Methods and systems for respiratory treatment management
The cloud-based platform for respiratory treatment management addresses intelligence and data management issues in respirators by generating intelligent display interfaces and automated parameter adjustments, improving treatment efficacy and user experience.
Patent Information
- Application Number
- PCT/CN2024/076456
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2024-02-06
- Publication Date
- 2025-08-14
AI Technical Summary
Existing respirators face issues with low intelligence, inaccurate parameter settings, and poor user experience due to manual control, leading to unsatisfactory treatment effects and difficulty in managing user data, which affects the quality and efficiency of respiratory treatment.
A cloud-based platform for respiratory treatment management that analyzes user and device data to generate intelligent display interfaces, recommended treatment plans, and automated parameter adjustments, enhancing user experience and treatment efficacy through intelligent data management and monitoring.
Improves the intelligence and accuracy of respiratory treatment by providing dynamic display interfaces, recommended settings, and real-time adjustments, ensuring effective treatment and efficient data management, thereby enhancing user experience and treatment outcomes.
Smart Images

Figure CN2024076456_14082025_PF_FP_ABST
Abstract
Description
METHODS AND SYSTEMS FOR RESPIRATORY TREATMENT MANAGEMENTTECHNICAL FIELD
[0001] The present disclosure relates to the technical field of respirators, and in particular, to a method and a system for respiratory treatment management.BACKGROUND
[0002] A respirator assists a spontaneous breathing and increases pulmonary ventilation of a patient through mechanical ventilation, which has a very obvious clinical effect on improving respiratory function and treating sleep diseases. At present, during the use of the respirator, parameters of the respirator usually need to be set manually, and errors, malfunctions, etc., of the respirator may be determined based on the usage of the patient. However, due to different professional levels or operating experiences of operators and different usage conditions of different patients, problems such as difficulty in use, inaccurate wearing, poor treatment effect, etc., are prone to occur, thereby causing a low degree of intelligence.
[0003] Therefore, it is desirable to provide a method and a system for respiratory treatment management that replace manual control through intelligent analysis and management, thereby improving use experience and ensuring a treatment effect.SUMMARY
[0004] One or more embodiments of the present disclosure provide a method for respiratory treatment management. The method may be implemented by a cloud platform. The method may comprise obtaining, based on a user terminal, at least one of user basic data, basic data of one or more respiratory devices, usage data of the one or more respiratory devices, or respiratory treatment data of a first subject. The user basic data may include basic data of the first subject or basic data of a second subject, the second subject may provide guidance or a suggestion regarding respiratory treatment to the first subject, and the respiratory treatment data may be obtained from a first monitoring device. The method may also comprise generating target information based on at least one of the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data. The method may further comprise sending the target information to the user terminal.
[0005] In some embodiments, wherein the generating target information based on at least one of the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data includes: generating a display interface based on the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data; wherein the display interface includes at least one of statistical information of the usage data of the one or more respiratory devices, statistical information of the respiratory treatment data, respiratory trend of the first subject, a recommended treatment plan of the first subject, and a supervisory relationship between the first subject and the second subject.
[0006] In some embodiments, wherein the display interface includes a plurality of sub-areas, the plurality of sub-areas correspond to different treatment parameters, and a display content of each of the plurality of sub-areas includes: a type of a treatment parameter and abnormal statistical data of the treatment parameter.
[0007] In some embodiments, the cloud platform may display different types of abnormal statistical data through different sub-areas in the display interface.
[0008] In some embodiments, the display content of each of the plurality of sub-areas may also include at least one of a proportion of data upload manner, a first subject board, a second subject board, a remote parameter regulation board, a report generation board, a comprehensive score of the first subject, an abnormal data graph and a data comparison diagram.
[0009] In some embodiments, the cloud platform may determine the comprehensive score of the first subject based on the respiratory treatment data and the respiratory device usage data of the first subject.
[0010] In some embodiments, different sub-areas may have different display parameters. In some embodiments, the cloud platform may determine importance levels of different abnormal statistical data based on the respiratory treatment data and the respiratory trend of the first subject and determine the display parameters of sub-areas corresponding to the different abnormal statistical data based on the importance levels. In some embodiments, the cloud platform may also determine, based on importance levels of different abnormal statistical data, the display parameters of the sub-areas corresponding to the different abnormal statistical data through a parameter determination model.
[0011] In some embodiments, the different sub-areas may have different refresh performance quotas. In some embodiments, the cloud platform may determine the refresh performance quotas allocated to the screen areas where the different sub-areas are located according to data generation frequencies and data update frequencies of different sub-areas.
[0012] In some embodiments, wherein the respiratory trend of the first subject is determined through operations including: predicting, based on the basic data of the first subject, the respiratory treatment data, and physiological data of the first subject, the respiratory trend of the first subject through a first prediction model; wherein the first prediction model is a trained machine learning model, the physiological data is acquired by a second monitoring device, and the second monitoring device is built into the user terminal or connected with a user terminal network.
[0013] In some embodiments, wherein the user terminal includes a second terminal or a first terminal, and the recommended treatment plan of the first subject is determined through operations including: determining a first candidate treatment plan; determining, based on the respiratory trend of the first subject and the first candidate treatment plan, a predicted treatment result corresponding to the first candidate treatment plan through a second prediction model, wherein the second prediction model is a trained machine learning model; determining a second candidate treatment plan based on the predicted treatment result and the first candidate treatment plan; sending the second candidate treatment plan and a predicted treatment result corresponding to the second candidate treatment plan to the second terminal; and in response to receiving a first recommended instruction fed back by the second terminal, generating the recommended treatment plan.
[0014] In some embodiments, the cloud platform may generate treatment reminder information to remind the first subject based on the recommended treatment plan, generate a treatment reminder parameter based on the respiratory trend of the first subject and / or the treatment cooperation degree of the first subject, and send the treatment reminder information to the first terminal based on the treatment reminder parameter. In some embodiments, the cloud platform may send the treatment reminder information to the first terminal according to the treatment reminder parameter. In some embodiments, the first terminal may determine a reminder intensity of the treatment reminder information according to a reminder time period of the treatment reminder information.
[0015] In some embodiments, the recommended treatment plan may further include an auxiliary treatment plan. In some embodiments, the cloud platform may generate a feature vector of the first subject based on basic data of the first subject, the respiratory trend of the first subject, and physiological data of the first subject and determine the auxiliary treatment plan by querying a first vector database based on the feature vector of the first subject.
[0016] In some embodiments, the cloud platform may determine one or more reference auxiliary vectors and the feature vector of the first subject through the similarity determination model based on the reference auxiliary treatment plan, relevant data of a first subject corresponding to the reference auxiliary treatment plan, and relevant data of a current first subject, determine a similarity between the feature vector of the first subject and each of the one or more reference auxiliary vectors, and determine a reference auxiliary treatment plan corresponding to a reference auxiliary vector with a highest similarity with the feature vector as the auxiliary treatment plan.
[0017] In some embodiments, the cloud platform may determine a relationship matching degree between a second subject and a first subject based on basic data of the first subject, a respiratory trend of the first subject, and basic data of the second subject; determine one or more candidate supervisory relationships based on the relationship matching degree; send the one or more candidate supervisory relationships to a second terminal, and generate one or more supervisory relationships based on a determination instruction fed back by the second termina.
[0018] In some embodiments, the cloud platform may generate at least one evaluation feature of the first subject based on historical feedback of the first subject, and determine a change probability of the supervisory relationship based on the basic data of the first subject, the at least one evaluation feature of the first subject, and the respiratory trend of the first subject. In response to a determination that the change probability is greater than a change probability threshold, the cloud platform may determine at least one alternative supervisory relationship based on the relationship matching degree. In some embodiments, the cloud platform may extract the positive feedback information and the negative feedback information from the first subject to the second subject through a feedback determination model based on the historical feedback of the first subject, and generate the at least one evaluation feature of the first subject based on the positive feedback information and the negative feedback information.
[0019] In some embodiments, the cloud platform may also construct an initial supervisory relationship map based on basic data of the first subject, a respiratory trend of the first subject, and basic data of the second subject, obtain a plurality of candidate supervisory relationship maps by adjusting the initial supervisory relationship map based on an adjustment rule, determine an estimated fatigue degree of each second node and a care perfection degree of each first node of the plurality of candidate supervisory relationship maps by processing the plurality of candidate supervisory relationship maps through a relationship assessment model, and generate a supervisory relationship based on the estimated fatigue degree of each second node and the care perfection degree of each first node of the plurality of candidate supervisory relationship maps.
[0020] In some embodiments, the cloud platform may generate a ward adjustment instruction with a frequency less than a preset adjustment frequency based on the supervisory relationship between the second subject and the first subject, and send the ward adjustment instruction to the user terminal to reminder the first subject to adjust a ward the first subject is admitted to.
[0021] In some embodiments, the cloud platform may determine whether the supervisory relationship needs to be adjusted based on a change in various information of the first subject over a period of time during which the current supervisory relationship is operated and a change in various information of the second subject. In response to a determination that the supervisory relationship needs to be adjusted, the cloud platform may be configured to generate a new supervisory relationship based on any of the manners mentioned above and send the new supervisory relationship to the first subject and / or the second subject. An actual supervisory relationship may be determined based on a feedback opinion of the first subject and / or the second subject, and the supervisory relationship may be adjusted based on the actual supervisory relationship.
[0022] In some embodiments, wherein the generating target information based on the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data includes: in response to receiving a report generation instruction from the user terminal, determining a report benchmark including a report content item and a report parameter; wherein the report benchmark is determined based on the user basic data or the respiratory treatment data; and generating a target report meeting the report benchmark.
[0023] In some embodiments, the cloud platform may also determine the content item based on basic data, a respiratory trend, and the respiratory treatment data of the first subject through a content item determination model.
[0024] In some embodiments, the report parameter may be divided into a first report parameter adapted to the first subject and a second report parameter adapted to the second subject. In some embodiments, the cloud platform may determine the first subject-adapted report parameter based on the basic data of the first subject.
[0025] In some embodiments, the cloud platform may determine a comprehensive recommendation of the second subject and a diagnostic analysis corresponding to the at least one content item through a report analysis model based on the basic data, the respiratory trend, and the respiratory treatment data of the first subject, and the at least one content item. In some embodiments, the cloud platform may generate the target report by lay outing the at least one content item, the diagnostic analysis corresponding to the at least one content item, and the comprehensive recommendation of the second subject according to the report parameter.
[0026] In some embodiments, wherein the generating target information based on the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data includes: generating a first result based on a relationship between the respiratory treatment data and a first index; wherein the first index is determined based on at least one of the basic data, posture data, and environmental data of the first subject; and generating reminder information based on the first result. In some embodiments, the cloud platform may determine the second subject-adapted report parameter based on basic data of the second subject.
[0027] In some embodiments, the cloud platform may determine the first reminder indicator based on at least one of basic data of the first subject, status data of the first subject, and environmental data of the first subject. In some embodiments, the cloud platform may determine the first reminder indicator through a reminder threshold model based on the basic data of the first subject, the status data of the first subject, the environmental data of the first subject, and the initial first reminder indicator.
[0028] In some embodiments, wherein the user terminal includes a second terminal and a first terminal, and the generating reminder information based on the first result includes: determining a type of a reason causing an abnormality of the respiratory treatment data; in response to a determination that the type of the reason causing the abnormality of the respiratory treatment data meets a condition, generating first reminder information based on the first result and sending the first reminder information to the first terminal; and in response to a determination that the type of the reason causing the abnormality of the respiratory treatment data does not meet the condition, generating second reminder information based on the first result and sending the second reminder information to the second terminal and the first terminal.
[0029] In some embodiments, the reason types of abnormalities in the respiratory treatment data may include a non-device fault and a device fault. In some embodiments, the cloud platform may determine the reason type of an abnormality in the respiratory treatment data based on a preset determination rule. In some embodiments, the preset determination rule may include that when an airway pressure stability is less than a stability threshold, and / or a difference between an oxygen supply rate and a rate set value is greater than a rate difference threshold, the reason type of an abnormality in the respiratory treatment data may be the device fault, or when the airway pressure stability is greater than or equal to the stability threshold, and / or the difference between the oxygen supply rate and the rate set value is less than or equal to the rate difference threshold, the reason type of an abnormality the respiratory treatment data may be the non-device fault.
[0030] In some embodiments, the cloud platform may predict a fault type of a device fault through a fault determination model based on the treatment data sequence of the respiratory device, the usage data of the respiratory device, a sequence of monitoring data of one or more auxiliary devices, and a sequence of physical measurement data of the first subject.
[0031] In some embodiments, the cloud platform may obtain image data and pressure data of the first subject wearing the respiratory device, and generate the reminder information based on the image data and / or the pressure data.
[0032] In some embodiments, the cloud platform may generate the reminder information based on the pressure data. In some embodiments, in response to a determination that at least one of the pressure values at one or more locations on the face of the first subject is higher than a pressure upper limit, the cloud platform may generate the reminder information indicating that the pressure exceeds the standard. In response to a determination that at least one of the pressure values at one or more locations on the face of the first subject is lower than a lower pressure limit, the cloud platform may generate the reminder information indicating that the mask does not fit the face of the first subject tightly.
[0033] In some embodiments, the cloud platform may generate a second determination result by determining a relationship between pressure data and a second reminder indicator, and generate reminder information based on the second determination result. In some embodiments, the second reminder indicator may be a pressure difference threshold corresponding to a pressure value difference between any two locations of a plurality of locations on the face of the first subject. In some embodiments, the cloud platform may determine the second reminder indicator based on at least one of facial data and posture data of the first subject.
[0034] In some embodiments, the cloud platform may determine a reminder intensity of the reminder information based on a current setting parameter and a duration of a current usage of the respiratory device currently used by a reminded first subject.
[0035] In some embodiments, the cloud platform may determine the abnormal wearing type of the first subject wearing the respiratory device based on the image data, determine a guidance content based on the abnormal wearing type, determine a guidance parameter based on basic data of the first subject, and generate guidance reminder information based on the guidance parameter and the guidance content through a guidance information determination model. In some embodiments, the cloud platform may determine the abnormal wearing type of the first subject wearing the respiratory device through a prediction model based on the image data.
[0036] In some embodiments, wherein the generating target information based on the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data includes: generating, based on the user basic data, the basic data of the one or more respiratory devices, and historical respiratory treatment data of the first subject, a recommended setting parameter for the one or more respiratory devices; and sending, through the user terminal, a control instruction to the one or more respiratory devices, wherein the control instruction includes the recommended setting parameter.
[0037] In some embodiments, wherein the generating, based on the user basic data, the basic data of the one or more respiratory devices, and historical treatment data of the first subject, a recommended setting parameter for the one or more respiratory devices includes: determining, based on the user basic data, the basic data of the one or more respiratory devices, and the historical respiratory treatment data, at least one candidate setting parameter through a third prediction model; wherein the third prediction model is a trained machine learning model; sending the at least one candidate setting parameter to the second terminal through a network; and generating the recommended setting parameter based on a second recommendation instruction fed back by the second terminal.
[0038] In some embodiments, the cloud platform may construct a map of respiratory devices, and determine the associated setting parameter by querying the map of the respiratory devices.
[0039] In some embodiments, wherein the one or more respiratory devices include a respiratory primary device and a respiratory auxiliary device, and the recommended setting parameter includes a first setting parameter of the respiratory primary device, a second setting parameter of the respiratory auxiliary device, and an associated setting parameter of the respiratory primary device and the respiratory auxiliary device.
[0040] In some embodiments, the cloud platform may provide an optimal respiratory primary device and / or a respiratory auxiliary device and a usage plan for each of different first subjects through a preset determination rule based on the general coordination parameter between the at least one respiratory primary device and the at least one respiratory auxiliary device, the applicable relationship between the at least one respiratory primary device and / or the at least one respiratory auxiliary device and the first subject, and the connectivity of different respiratory primary device nodes, respiratory auxiliary device nodes, and first nodes.
[0041] In some embodiments, in response to receiving a feedback signal of startup of the respiratory device by the first terminal, the cloud platform may obtain environmental data where the respiratory device are located, and status data and / or posture data of the first subject. The cloud platform may update the recommended setting parameter based on the environmental data, and the patient status data and / or the posture data of the first subject, and send an updated control instruction generated by the first terminal to the respiratory device through the first terminal. In some embodiments, the cloud platform may determine the status data and / or posture data of the first subject through a recognition model based on respiratory features detected by the at least one respiratory primary device and other features detected by the at least one respiratory auxiliary device.
[0042] In some embodiments, the cloud platform may also determine a parameter adjustment amount of the respiratory device using an adjustment determination model based on a current physiological parameter, a target physiological parameter, a current setting parameter, the environmental data, the status data and / or the posture data.
[0043] In some embodiments, the cloud platform may obtain a remote parameter adjustment instruction based on user input, and send the remote parameter adjustment instruction to the respiratory device through a remote control medium to remotely control the respiratory device to adjust a setting parameter. In some embodiments, the cloud platform may provide an adjustment interval for the setting parameter of the respiratory device and display the adjustment interval to the user through the remote parameter adjustment board. In some embodiments, the cloud platform may determine the adjustment interval of the setting parameter through an interval determination model based on the basic data of the first subject, the basic data of the respiratory device, the usage data of the respiratory device, and the target physiological monitoring data.
[0044] In some embodiments, wherein the generating target information based on the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data includes: determining a quality of the respiratory treatment data; and in response to a determination that the quality does not meet a quality condition, correcting the respiratory treatment data based on historical respiratory treatment data of the first subject to generate corrected respiratory treatment data.
[0045] In some embodiments, the cloud platform may determine the quality of the respiratory treatment data based on the respiratory treatment data and a preset evaluation condition. In some embodiments, the preset evaluation condition may include that the data signal intensity, the data integrity, and the sampling rate of the respiratory treatment data are respectively greater than corresponding thresholds, and an amplitude of the respiratory treatment data (e.g., a difference between a highest value and a lowest value in a data segment) is within a preset amplitude range, etc.
[0046] In some embodiments, the cloud platform may determine a degree of data similarity between the respiratory treatment data and standard respiratory treatment data, and determine a quality score of the respiratory treatment data based on the degree of data similarity.
[0047] In some embodiments, wherein the correcting the respiratory treatment data based on historical respiratory treatment data of the first subject includes: generating the corrected respiratory treatment data by processing a historical treatment sequence based on a correction model, wherein the correction model is a sequence model, and the historical treatment sequence is constructed based on the historical respiratory treatment data according to a preset time rule.
[0048] In some embodiments, wherein the generating target information based on the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data includes: determining, based on the historical respiratory treatment data and a respiratory trend of the first subject, acquisition parameters of different treatment parameters.
[0049] In some embodiments, wherein the determining, based on the historical respiratory treatment data and a respiratory trend of the first subject, acquisition parameters of different treatment parameters includes: predicting, based on the historical respiratory treatment data, a predicted acquisition data sequence of the different treatment parameters through a fourth prediction model, wherein the fourth prediction model is a sequence model; determining priorities of the different treatment parameters based on the predicted acquisition data sequence; and determining the acquisition parameters of the different treatment parameters based on the priorities of the different treatment parameters.
[0050] In some embodiments, the cloud platform may determine the priorities of the different treatment parameters based on a distance relationship between the predicted acquisition data sequence and a safety interval. In some embodiments, the cloud platform may determine the safety interval based on the respiratory trend of the first subject.
[0051] In some embodiments, the priorities of the treatment parameters may be positively correlated with the distance relationship between the predicted acquisition data sequence and the safe interval.
[0052] In some embodiments, the cloud platform may adjust the acquisition parameters of the different treatment parameters based on first bandwidth data between the respiratory device and the user terminal. The first bandwidth data may be data related to data transmission performance between the respiratory device and the user terminal. In some embodiments, the cloud platform may determine, based on the first bandwidth data of the acquisition time period, an upload congestion degree of the respiratory device uploading the acquisition data to the user terminal in the acquisition time period; and adjust, based on the upload congestion degree, the acquisition parameters of the different treatment parameters.
[0053] In some embodiments, the cloud platform may determine an adjustment parameter 1380 based on a life feature sequence and a treatment feature sequence of the first subject; and adjust the acquisition parameters of the different treatment parameters based on the adjustment parameter.
[0054] In some embodiments, the cloud platform may determine a respiratory impact situation of the life feature sequence and the treatment feature sequence on the first subject by temporally matching comparative change information of a historical acquisition data sequence of the first subject with the life feature sequence and the treatment feature sequence of the first subject; and determine the adjustment parameter based on the respiratory impact situation. In some embodiments, the cloud platform may temporally match the change time point in the comparative change information with the life feature sequence and the treatment feature sequence of the first subject and determine the magnitude of the change, the direction of the change, etc., of the change time point as the respiratory impact situation corresponding to the change time point.
[0055] In some embodiments, the cloud platform may generate a candidate adjustment parameter based on the respiratory impact situation; determine a missed acquisition rate of the candidate adjustment parameter through a data evaluation model; and determine the adjustment parameter by performing, based on the missed acquisition rate of the candidate adjustment parameter, at least one round of iterative updating on the candidate adjustment parameter.
[0056] In some embodiments, the cloud platform may adjust a transmission speed of the acquisition data based on second bandwidth data between a first terminal and the cloud platform. The second bandwidth data may be data related to data transmission performance between the first terminal and the cloud platform. In some embodiments, the cloud platform may determine a data synchronization transmission situation based on the second bandwidth data; in response to determining that the data synchronization transmission situation is that synchronization transmission is supported or allowable, support a current transmission speed; or in response to determining that the data synchronization transmission situation is that synchronization transmission is not supported or unallowable, adjust the transmission speeds of the acquisition data of the different treatment parameters.
[0057] In some embodiments, the cloud platform may adjust transmission speeds of the different treatment parameters based on a distance relationship between the acquisition data of the different treatment parameters and a safety interval corresponding to the acquisition data of the different treatment parameters.
[0058] In some embodiments, in response to receiving a preheating request, the cloud platform may generate a preheating instruction, and send the preheating instruction to a humidifier in the respiratory device to control the humidifier for preheating.
[0059] In some embodiments, the cloud platform may obtain a session initiation instruction based on a first terminal; predict one or more suspected questions based on respiratory treatment data and operation behavior data performed by a user on a display interface, and displayed the one or more suspected questions; in response to an operation instruction of the user for the one or more suspected questions, determine a target question; and determine a target answer and fed back to the user through the first terminal.
[0060] In some embodiments, the cloud platform may determine a question sequence based on a respiratory trend, the operation behavior data, historical dialogue data, respiratory treatment data, and usage data of a respiratory device through a fifth prediction model; and determine the one or more suspected questions based on the question sequence. In some embodiments, the cloud platform may determine, based on the greatest question confidence level in the question sequence, a count of questions that are determined as suspected questions in the question sequence generated by the fifth prediction model.
[0061] In some embodiments, the cloud platform may obtain, in response to determining that the target question does not satisfy a template condition, the target answer corresponding to the target question based on the target question, disease information of the first subject, and the respiratory treatment data through an answer generation model.
[0062] In some embodiments, the target answer may include key point data. In some embodiments, the cloud platform may obtain a keyword sequence based on a content input by the first subject; and generate, based on the keyword sequence, the portrait of the first subject, and historical key point data, key point data through a key point generation model.
[0063] In some embodiments, the cloud platform may generate an embellished statement by embellishing the target answer to be presented to the first subject. In some embodiments, the cloud platform may generate the embellished statement based on the content input by the first subject and the target question through a statement embellishment model.
[0064] In some embodiments, the cloud platform may deliver the embellished statement to the first subject in various forms. The form of delivery may be related to basic information about the first subject. In some embodiments, the cloud platform may determine the form of delivery to the first subject based on a text length of the embellished statement. In some embodiments, in response to the text length of the embellished statement being less than a preset length, the cloud platform may deliver the embellished statement to the first subject in the form of text; in response to the text length of the embellished statement being greater than or equal to the preset length, the cloud platform may deliver the embellished statement to the first subject in the form of text and voice playback.
[0065] One or more embodiments of the present disclosure provide a system for respiratory treatment management. The system may comprise a cloud platform, a user terminal, one or more respiratory devices, and a first monitoring device. The one or more respiratory devices and the first monitoring device may be connected with the user terminal, and the user terminal may be connected with the cloud platform. The one or more respiratory devices may be configured to acquire basic data of the one or more respiratory devices and usage data of the one or more respiratory devices, and transmit the basic data of the one or more respiratory devices and the usage data of the one or more respiratory devices to the user terminal through a network. The first monitoring device may be configured to acquire respiratory treatment data of a first subject and transmit the respiratory treatment data to the user terminal. The user terminal may be configured to acquire user basic data and transmit the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, the respiratory treatment data, and the user basic data to the cloud platform. The cloud platform may be configured to generate target information based on at least one of the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data and send the target information to the user terminal.
[0066] One or more embodiments of the present disclosure provide a non-transitory computer-readable storage medium storing computer instructions. After reading the computer instructions in the storage medium, a computer may execute the method for respiratory treatment management described above.BRIEF DESCRIPTION OF THE DRAWINGS
[0067] The present disclosure is further described in terms of exemplary embodiments. These exemplary embodiments are described in detail with reference to the drawings. These embodiments are non-limiting exemplary embodiments, in which like reference numerals represent similar structures throughout the several views of the drawings, and wherein:
[0068] FIG. 1A is a schematic structural diagram illustrating a system for respiratory treatment management according to some embodiments of the present disclosure;
[0069] FIG. 1B is a schematic diagram illustrating an application scenario of a system for respiratory treatment management according to some embodiments of the present disclosure;
[0070] FIG. 2 is a flowchart illustrating an exemplary method for respiratory treatment management according to some embodiments of the present disclosure;
[0071] FIG. 3 is a schematic diagram illustrating an exemplary process of determining a respiratory trend according to some embodiments of the present disclosure;
[0072] FIG. 4A is a schematic diagram illustrating exemplary statistical information according to some embodiments of the present disclosure;
[0073] FIG. 4B is a schematic diagram illustrating exemplary statistical information according to some embodiments of the present disclosure;
[0074] FIG. 4C is a schematic diagram illustrating exemplary statistical information according to some embodiments of the present disclosure;
[0075] FIG. 4D is a schematic diagram illustrating an exemplary display interface according to some embodiments of the present disclosure;
[0076] FIG. 4E is a schematic diagram illustrating an exemplary detailed data interface according to some embodiments of the present disclosure;
[0077] FIG. 4F is a schematic diagram illustrating an exemplary abnormal data map according to some embodiments of the present disclosure;
[0078] FIG. 5 is a flowchart illustrating an exemplary process of determining a recommended treatment plan according to some embodiments of the present disclosure;
[0079] FIG. 6 is a schematic diagram illustrating an exemplary process of determining a supervisory relationship according to some embodiments of the present disclosure;
[0080] FIG. 7 is a schematic diagram illustrating an exemplary process of determining a supervisory relationship according to some embodiments of the present disclosure;
[0081] FIG. 8 is a schematic diagram illustrating an exemplary process of generating a target report according to some embodiments of the present disclosure;
[0082] FIG. 9 is a flowchart illustrating an exemplary process of generating reminder information according to some embodiments of the present disclosure;
[0083] FIG. 10 is a flowchart illustrating an exemplary process of generating a recommended setting parameter according to some embodiments of the present disclosure;
[0084] FIG. 11A is a schematic diagram illustrating an exemplary application of a third prediction model according to some embodiments of the present disclosure;
[0085] FIG. 11B is a schematic diagram illustrating an exemplary application of a third prediction model according to some embodiments of the present disclosure;
[0086] FIG. 12 is a flowchart illustrating an exemplary process of generating corrected respiratory treatment data according to some embodiments of the present disclosure;
[0087] FIG. 13 is a schematic diagram illustrating an exemplary process of determining acquisition parameters according to some embodiments of the present disclosure; and
[0088] FIG. 14 is a flowchart illustrating an exemplary process of determining a target answer according to some embodiments of the present disclosure.DETAILED DESCRIPTION
[0089] In the following detailed description, numerous specific details are set forth by way of examples in order to provide a thorough understanding of the relevant disclosure. Obviously, drawings described below are only some examples or embodiments of the present disclosure. Those skilled in the art, without further creative efforts, may apply the present disclosure to other similar scenarios according to these drawings. It should be understood that the purposes of these illustrated embodiments are only provided to those skilled in the art to practice the application, and not intended to limit the scope of the present disclosure. Unless obviously obtained from the context or the context illustrates otherwise, the same numeral in the drawings refers to the same structure or operation.
[0090] It will be understood that the terms “system, ” “unit, ” and / or “module” used herein are one method to distinguish different components, elements, parts, sections, or assemblies of different levels in ascending order. However, the terms may be displaced by other expressions if they may achieve the same purpose.
[0091] The terminology used herein is for the purposes of describing particular examples and embodiments only and is not intended to be limiting. As used herein, the singular forms “a, ” “an, ” and “the” may be intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms “include” and / or “comprise, ” when used in this disclosure, specify the presence of integers, devices, behaviors, stated features, steps, elements, operations, and / or components, but do not exclude the presence or addition of one or more other integers, devices, behaviors, features, steps, elements, operations, components, and / or groups thereof.
[0092] The flowcharts used in the present disclosure illustrate operations that systems implement according to some embodiments of the present disclosure. It is to be expressly understood, the operations of the flowcharts may be implemented not in order. Conversely, the operations may be implemented in an inverted order, or simultaneously. Moreover, one or more other operations may be added to the flowcharts. One or more operations may be removed from the flowcharts.
[0093] At present, an application of a respirator has problems such as low intelligence, unsatisfactory treatment effect, and low accuracy in respirator abnormality assessment, which may be caused by various reasons. For example, the use of the respirator may involve various aspects of monitoring and control, the more important of which are the setting of treatment parameters, the determination of a treatment plan, and the dynamic display of a treatment report. Inaccuracies in the treatment parameters and the treatment plan may cause the respirator to be unable to assist or control the spontaneous breathing of the patient, thereby affecting the effect of treatment and reducing the user experience. The treatment report may not be displayed intelligently, which is not conducive for a doctor and a patient to understand a pathogenetic condition, etc. Moreover, different respirator types and patient conditions may have different requirements for the treatment parameters, the treatment plans, and the treatment reports. If different patients are treated with fixed treatment parameters, the quality and efficiency of respirator treatment cannot be guaranteed, and the pathogenetic condition under the treatment may be seriously affected.
[0094] In addition, during the use of the respirator, user data management may also be extremely important, since the user data management may provide a good data basis for the pertinency, intelligence, and the rapeutic effect of subsequent user treatment may be provided by. However, the user data management of a current respirator may include statistically summarizing the data and sending the summarized data to a doctor for review and analysis, and the doctor may inform the user of the relevant treatment. In this case, a pertinence and timeliness of data analysis may not be possible to be achieved when an amount of data is large and the user condition is complex. In addition, medical and nursing communication is not flexible and convenient, which may even affect the treatment effect of the user.
[0095] Accordingly, some embodiments of the present disclosure provide a method and a system for respiratory treatment management, which can generate display interfaces, target reports, reminder information, and recommended setting parameters for different users based on the user basic data, the basic data of one or more respiratory devices, the usage data of the one or more respiratory devices, and / or the respiratory treatment data, and display the types of treatment parameters and the abnormal statistical data of the treatment parameters for different sub-areas of the display interfaces, a recommended setting parameter for the one or more respiratory devices, so that the intelligence of the respiratory treatment process and the user experience can be improved, intelligent treatment can be realized, and the treatment effect of the respirator can be improved. Moreover, by setting up an intelligent user data management mechanism, carrying out intelligent monitoring management, downloading management, uploading management, and interactive management of the user data, hierarchical and classified management of the user data based on users and medical staff can be realized, and real-time dynamic adjustment can be performed at any time according to the users and the medical staff, thereby improving a level of data management and subsequent data application, allowing the medical staff to promptly and accurately understand the treatment status of the user and adjust the treatment plan in time, enabling the user to obtain treatment information through communication in time, and improving the treatment effect.
[0096] FIG. 1A is a schematic structural diagram illustrating a system for respiratory treatment management according to some embodiments of the present disclosure.
[0097] In some embodiments, a system 100 for respiratory treatment management (also referred to as respiratory treatment management system) may be applied to a hospital, a home place for home care, an ambulance, a rehabilitation center, a sports place, etc., that requires respiratory treatment, respiratory management, etc. For example, the system 100 for respiratory treatment management may be configured to monitor a condition of a first subject and provide a treatment plan for the first subject. As another example, the system 100 for respiratory treatment management may provide continuous respiratory treatment and monitoring for the first subject, and also provide necessary monitoring data to track a health status of the first subject. As still another example, the system 100 for respiratory treatment management may be arranged in an ambulance, so that a medical staff may monitor a respiratory status of the first subject in time more intuitively. As still another example, the system 100 for respiratory treatment management may be configured to monitor the respiratory status of the first subject to help the first subject perform respiratory rehabilitation training.
[0098] As shown in FIG. 1A, the system 100 for respiratory treatment management may include a cloud platform 110, one or more user terminals 120, a respiratory devices 130, a first monitoring device 140, and a second monitoring device 150.
[0099] The cloud platform 110 refers to a cloud computing platform configured to provide computing, network, and storage capabilities based on services provided by hardware resources and software resources.
[0100] In some embodiments, the cloud platform 110 may include one or more processing devices. Merely by way of example, the one or more processing devices may include a central processing unit (CPU) , an application specific integrated circuit (ASIC) , an application specific instruction processor (ASIP) , a graphics processing unit (GPU) , a physical processor (PPU) , a digital signal processor (DSP) , a field programmable gate array (FPGA) , a programmable logic circuit (PLD) , a controller, a microcontroller unit, a reduced instruction set computer (RISC) , a microprocessor, or the like, or any combination thereof.
[0101] In some embodiments, the cloud platform 110 may process data or information obtained from other devices or system components based on software implemented on the cloud platform 110, and execute methods for respiratory treatment management of some embodiments of the present disclosure based on the data, the information, and / or processing results to complete functions described in some embodiments of the present disclosure. In some embodiments, other components (e.g., the one or more user terminals 120, the respiratory device 130, the first monitoring device 140, and the second monitoring device 150) of the system 100 for respiratory treatment management may be connected with the cloud platform 110 through a network to communicate with the cloud platform 110. In some embodiments, the cloud platform 110 may obtain user basic data from the one or more user terminals 120, obtain basic data of a respiratory device and usage data of the respiratory device among the one or more respiratory devices 130, and obtain a statistical analysis result by performing statistical analysis on the obtained data. As another example, the cloud platform 110 may generate a display interface based on the user basic data, the basic data of the respiratory device, the usage data of the respiratory device, or respiratory treatment data. As another example, the cloud platform 110 may control display of the user basic data, the basic data of the respiratory device, the usage data of the respiratory device, or the respiratory treatment data based on different display areas of the display interface.
[0102] In some embodiments, the cloud platform 110 may obtain at least one of the user basic data, the basic data of the respiratory device, the usage data of the one or more respiratory devices, and / or respiratory treatment data based on the user terminal. The cloud platform 110 may generate target information based on at least one of the user basic data, the basic data of the respiratory device, the usage data of the respiratory device, or the respiratory treatment data. The cloud platform 110 may send the target information to the one or more user terminals 120.
[0103] In some embodiments, the basic data of the user, the basic data of the respiratory device 130, the usage data of the respiratory device 130, and the respiratory treatment data may be stored in the cloud platform 110. In some embodiments, when the basic data of the user, the basic data of the respiratory device 130, the usage data of the respiratory device 130, and the respiratory treatment data are updated to the cloud platform 110, the basic data of the respiratory device 130, the usage data of the respiratory device 130, and the respiratory treatment data may be correlated by a serial number of the respiratory device 130.
[0104] In some embodiments, monitoring data (e.g., the respiratory treatment data, physiological data, etc. ) obtained by the first monitoring device 140 and the second monitoring device 150 may be stored in the respiratory device 130 or uploaded to the cloud platform 110 through the respiratory device 130.
[0105] In some embodiments, a manner of obtaining the data from the respiratory device 130 by the one or more user terminals 120 or the cloud platform 110 may include but is not limited to wireless communication, uploading by scanning a QR code, etc.
[0106] In some embodiments, the cloud platform 110 may generate a display interface based on at least one of the user basic data, the basic data of the respiratory device, the usage data of the respiratory device, or the respiratory treatment data. The display interface may display at least one of statistical information of the usage data of the respiratory device, statistical information of the respiratory treatment data, a respiratory trend of a subject (i.e., the first subject) who uses the respiratory device, a recommended treatment plan of the first subject, and a supervisory relationship between the first subject and a second subject who is responsible for monitoring the first subject using the respiratory device. More descriptions regarding generating the display interface may be found in FIG. 2 and related descriptions thereof.
[0107] In some embodiments, the user may be embodied in various roles (e.g., the first subject, the second subject, a distributor, an administrator, etc. ) . The cloud platform 110 may display the data corresponding to the user on a display interface based on the role of the user and data display requirements of different user roles. For example, a respiratory disease situation and disease analysis may be displayed on the display interface for the first subject; the respiratory trend of the first subject and the recommended treatment plan of the first subject may be displayed on the display interface for the second subject; and statistical information of the usage data, statistical information of the respiratory treatment data, and the supervisory relationship between the first subject and the second subject may be displayed on the display interface for the administrator. More descriptions regarding the display interface of the first terminal may be found in FIG. 4A and FIG. 4B and related descriptions thereof. More descriptions regarding the display interface of the second terminal may be found in FIG. 4C and related descriptions thereof.
[0108] In some embodiments, the cloud platform 110 may generate a recommended setting parameter for the respiratory device 130 based on the user basic data, the basic data of the respiratory device 130, and historical respiratory treatment data of the first subject, and send a control instruction to the respiratory device 130 through the one or more user terminals 120. In some embodiments, the cloud platform 110 may generate a control instruction based on the recommended setting parameter of the respiratory device 130 and directly send the control instruction to the respiratory device 130. The control instruction may include the recommended setting parameter. More descriptions regarding generating the recommended setting parameter may be found in FIG. 10 and related descriptions thereof.
[0109] In some embodiments, in response to receiving a report generation instruction from the one or more user terminals 120, the cloud platform 110 may determine a report benchmark including a report content item and a report parameter. The report benchmark may be determined based on the user basic data or the respiratory treatment data. The cloud platform 110 may also generate a target report meeting the report benchmark. More descriptions regarding generating the target report may be found in FIG. 8 and related descriptions thereof.
[0110] In some embodiments, the cloud platform 110 may generate a first result based on a relationship between the respiratory treatment data and a first index. The first index may be determined based on at least one of basic data, posture data, and environmental data of the first subject. The cloud platform 110 may also generate reminder information based on the first result. More descriptions regarding generating the reminder information may be found in FIG. 9 and related descriptions thereof.
[0111] In some embodiments, the cloud platform 110 may determine a quality of the respiratory therapy data, and in response to a determination that the quality does not meet a quality condition, the cloud platform 110 may correct the respiratory treatment data based on historical respiratory treatment data of the first subject to generate corrected respiratory treatment data. More descriptions regarding generating the corrected respiratory treatment data may be found in FIG. 12 and related descriptions thereof.
[0112] In some embodiments, the cloud platform 110 may determine acquisition parameters of different treatment parameters based on the historical respiratory treatment data and the respiratory trend of the first subject. More descriptions regarding determining the acquisition parameters may be found in FIG. 12 and related descriptions thereof.
[0113] In some embodiments, the cloud platform 110 may obtain a session initiation instruction based on a user terminal, and predict and display one or more suspected questions based on the respiratory treatment data and operation behavior data performed by a user on the display interface. The cloud platform 110 may also determine a target question in response to a user operation instruction of the suspected questions, determine a target answer, and feed the target answer back to the user through the user terminal. More descriptions regarding determining the target answer may be found in FIG. 14 and related descriptions thereof.
[0114] The user terminal 120 refers to a terminal device or software used by the user. For example, the user terminal (s) 120 may include a PC terminal, a PAD terminal, a mobile terminal, software configured therein, etc. The user terminal (s) 120 may include a processing unit, a display unit, an input / output (I / O) unit, a sensing unit, a storage unit, etc. In some embodiments, the cloud platform 110 may provide the user with data related to use of the respiratory device and a solution to a respiratory problem through the one or more user terminals 120. In some embodiments, the user terminal (s) 120 may include various types of devices with information receiving and / or sending functions. For example, the user terminal (s) 120 may include a tablet computer, a laptop computer, a mobile phone, a desktop computer, or the like, or any combination thereof.
[0115] In some embodiments, each of the one or more user terminals 120 may be used by one or more users, which may include a user directly using a service provided by the user terminal, or another user. In some embodiments, the one or more users may include a medical staff, an operation and maintenance personnel, a subject (e.g., the first subject) with a respiratory disease, an elderly people needing respiratory monitoring, an athlete, etc. Merely by way of example, the respiratory disease may include an upper respiratory tract infection, a lower respiratory tract infection, pulmonary fibrosis, a chronic obstructive pulmonary disease, asthma, allergic rhinitis, or the like.
[0116] In some embodiments, the one or more user terminals 120 may include a first terminal 120-1 and a second terminal 120-2. The user of the first terminal 120-1 may be the first subject. The first subject may be a patient, a patient's family, etc. The user of the second terminal 120-2 may be the second subject. The second subject may be a doctor, a nurse, etc. The second subject may provide guidance and a suggestion on respiratory treatment to the first subject. For example, the user of the first terminal 120-1 may input the user basic data, view a target report, upload sleep data of the first subject to the cloud platform 110, etc., through the first terminal 120-1. The user of the second terminal 120-2 may download data related to the first subject, provide group management for the first subject, and / or input one or more respirator setting parameters, etc., through the second terminal 120-2.
[0117] In some embodiments, the one or more user terminals 120 may communicate with the cloud platform 110 through a network.
[0118] In some embodiments, the one or more user terminals 120 may obtain user basic data input by the user, obtain basic data of one or more respiratory devices and usage data of the one or more respiratory devices among the one or more respiratory devices 130, and send the user basic data, the basic data of the one or more respiratory devices, and the usage data of the one or more respiratory devices to the cloud platform 110.
[0119] In some embodiments, the one or more user terminals 120 may obtain a data migration instruction sent by the user. For example, the user may send the data migration instruction by operating an application program on the one or more user terminals 120, or by opening the application program and clicking a button, a page, or a link therein.
[0120] In some embodiments, the one or more user terminals 120 may obtain a type of data to be migrated in response to the data migration instruction sent by the user, determine at least one data keyword based on the type of data to be migrated, and perform data identification and labeling on a document to be migrated based on at least one data keyword.
[0121] The data migration instruction refers to an instruction for importing data information into a specific location for storage.
[0122] The type of data to be migrated refers to a data type of a document to be migrated. The document to be migrated refers to a data text to be imported into a specific location. For example, the document to be migrated may include basic information data of the user (e.g., age and gender) , other test data, medical history information, recommended setting parameters of the user, etc. In some embodiments, the document to be migrated may include monitoring data that needs to be migrated from the respiratory device 130, the first monitoring device 140, and / or the second monitoring device 150 to a specific location, such as the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, the respiratory treatment data of the first subject, the physiological data of the first subject, or other data types. The specific location may include a local storage space of the user terminal (s) 120, a cloud storage space of the cloud platform 110, or an external storage space (e.g., a hard disk, etc. ) .
[0123] The data keyword refers to a word used to describe or define a topic or a title of the document to be migrated. For example, the data keyword may include age, gender, medical history, medical examination information, etc.
[0124] In some embodiments, a corresponding relationship table between different data types to be migrated and different data keywords may be established. Data keywords corresponding to a current document to be migrated may be determined based on a current data type to be migrated and the corresponding relationship table between different data types to be migrated and different data keywords.
[0125] In some embodiments, the one or more user terminals 120 may determine a location of partial data corresponding to the data keyword in the document to be migrated based on the data keyword, extract the data at the location, and store the data based on a format corresponding to the data keyword. Different data keywords may correspond to different data storage formats, and the data storage formats may be preset. Different types and formats of data may be imported and stored by dividing the documents to be migrated based on the data keywords.
[0126] In some embodiments, the one or more user terminals 120 may obtain a data upload instruction sent by the user. For example, the user may send the data upload instruction by operating an application program on the user terminal, or by opening the application program and clicking on a button, a page, or a link therein.
[0127] In some embodiments, in response to the user sending the data upload instruction, the one or more user terminals 120 may obtain a type of data to be uploaded, determine a data sensitivity based on the type of data to be uploaded, obtain desensitized data by performing desensitization on the data to be uploaded based on the data sensitivity, and upload the desensitized data to the cloud platform 110.
[0128] The data upload instruction refers to an instruction for uploading data information to the cloud platform.
[0129] The type of data to be uploaded refers to a data type of the data to be uploaded. For example, the type of data to be uploaded may include identity information, case information, location information, etc., of the user.
[0130] The data sensitivity refers to a sensitivity level of the data to be uploaded. The more sensitive information contained in the data to be uploaded, the greater the data sensitivity may be. The sensitive information may include private information, important treatment data, etc., of the user. The data sensitivity may be expressed in various ways, such as a sensitivity value, a sensitivity grade, etc. In some embodiments, a corresponding relationship between an amount of sensitive information and the data sensitivity may be preset based on prior knowledge or historical data, and the data sensitivity of the data to be uploaded may be determined accordingly based on the amount of sensitive information contained in the data to be uploaded and the corresponding relationship between an amount of sensitive information and the data sensitivity.
[0131] The desensitized data refers to the data to be uploaded after desensitization. The desensitization may include encryption desensitization, disguise desensitization, anonymous desensitization, obfuscation desensitization, etc., for the sensitive information in the data to be uploaded. The encryption desensitization may include using a cryptographic algorithm to encrypt the sensitive information. The disguise desensitization may include replacing the sensitive information with specific symbols or simulated data. The anonymous desensitization may include desensitizing personal identification information, so that the information cannot be directly associated with an individual's identity. The obfuscation desensitization may include obfuscating certain fields in the data, so that the fields are no longer identifiable.
[0132] In some embodiments, the one or more user terminals 120 may perform the desensitization in various ways based on the data sensitivity of the type of data to be uploaded. For example, the one or more user terminals 120 may determine whether the data sensitivity of the data to be uploaded is greater than a sensitivity threshold based on the type of data to be uploaded, and in response to a determination that the data sensitivity of the data to be uploaded is greater than the sensitivity threshold, the one or more user terminals 120 may obtain the desensitized data by performing desensitization on the data to be uploaded using a desensitization rule. The sensitivity threshold may be a manual preset value, a system default value, etc.
[0133] In some embodiments, when the data (e.g., the treatment data of the respiratory device, the usage data of the respiratory device, etc. ) is uploaded to the cloud platform 110 using the respiratory device 130, the respiratory device 130 may perform a desensitization processing on the data based on a data sensitivity of the data.
[0134] In some embodiments, the desensitization rule may include simplification of sensitive information, shuffling of upload order, random selection of encoding modes, or the like, or any combination thereof.
[0135] Simplification of the sensitive information refers to encoding the sensitive information of the user and restricting a decoding authority of the data. For example, only users with permission can download data locally and use a local decoder to decode the data. Shuffling of upload order refers to packaging the data to be uploaded into segments based on a certain data length, and then uploading the data in a shuffled order. Random selection of encoding modes refers to that the data to be uploaded is segmented based on a certain data length, and for each segmented data, an encoding mode is randomly selected from various preset encoding modes for encoding the each segmented data.
[0136] In some embodiments, encoding the sensitive information of the user may include encrypting the sensitive information using an encryption algorithm, and using encrypted data as a result of encoding. The encryption algorithm may include a message-digest algorithm 5 (MD5) , a data encryption standard (DES) algorithm, an advanced encryption standard (AES) algorithm, etc.
[0137] In some embodiments, the encryption algorithm may include one or more encryption sub-algorithms. The encrypting the sensitive information using the encryption algorithm may include performing multi-level encryption on the sensitive information by using the encryption algorithm. The multi-level encryption refers to performing encryption of the sensitive information step-by-step by using one or more encryption sub-algorithms. The multi-level encryption is usually used in a scenario of distributed permission control. For example, when a piece of sensitive information needs to be accessed through level-by-level permission, the sensitive information may be subjected to multi-level encryption. Correspondingly, when the user accesses the sensitive information, the user needs to perform multi-level decoding to achieve distributed permission control.
[0138] In some embodiments, the restricting the decoding permission of data may include obtaining an encoding feature for encoding on the sensitive information of each user and allocating specific decoding key sets to different users based on the encoding features and permission features of different users. The encoding feature refer to an encoding mode used for encoding. The encoding features may include a type of the preset encryption algorithm and an encryption key. For example, when the encryption algorithm is the AES, the encoding feature may be the AES. As another example, when the encryption algorithm uses the AES and the DES as the two preset encryption sub-algorithms, the encoding feature may be expressed as (AES, DES) . The permission feature refers to a set of permissions that the user has. For example, if a user A has the permission to access data b1 and data b2 in the sensitive information, the permission feature of the user A may be expressed as (b1, b2) . In some embodiments, a method of allocating specific decoding keysets to different users based on the encoding features and the permission features of different users may include for each user and each piece of sensitive data in the sensitive information, determining the encoding feature of each piece of sensitive data and a permission for the sensitive data of the user, and in response to a determination that the user has an access permission to the sensitive data, the method may also include adding a decoding key corresponding to an encryption key of the sensitive data to a decryption keyset of the user based on the encoding feature and sending the decoding keyset to the user.
[0139] In some embodiments, the shuffling of upload order may include segmenting and packaging the data to be uploaded based on a data length, and uploading the data in a shuffled order. In some embodiments, the data length may be determined based on a total count of visits to clinic of the first subject. For example, the smaller the count of visits to clinic, the smaller the data length may be. The count of visits to clinic of the first subject may be determined based on a count of consultations with the second subject and / or a count of visits to a hospital of the first subject. For example, the count of visits to clinic of the first subject may be the count of consultations with the second subject of the first subject, the count of visits to the hospital of the first subject, or a sum of the count of consultations with the second subject of the first subject and the count of visits to the hospital of the first subject. The degree of desensitization performed on the treatment data of the first subject may be determined based on the count of visits to the clinic, so that data with better treatment effects may be better protected.
[0140] In some embodiments, the random selection of encoding mode may include segmenting the data to be uploaded based on a data length, and randomly selecting an encoding mode from a plurality of encoding modes for encoding each segmented data. The encoding mode may be an encryption algorithm and / or an encryption sub-algorithm in the encryption algorithm.
[0141] In some embodiments, the desensitized data may include self-destruct codes. The self-destruct codes may be configured to erase the data when data decoding fails. The self-destruct codes may be preset in each data segment to be uploaded.
[0142] In some embodiments, the cloud platform 110 may determine an upload channel and an upload bandwidth of the desensitized data based on data sensitivity of original data to be uploaded corresponding to the desensitized data to be uploaded. The upload channel refers to a network channel through which the desensitized data is uploaded. The upload channel may include an Internet channel, a wireless local area network (LAN) channel, a wired private network channel, etc. The upload bandwidth refers to a bandwidth allocated for an upload task of the desensitized data. For example, the higher the data sensitivity of the original data to be uploaded corresponding to the desensitized data to be uploaded is, a more secure upload channel (e.g., the wired private network channel) and a higher upload bandwidth may be allocated to the desensitized data.
[0143] By performing the desensitization and uploading the desensitized data to the cloud platform, sensitive private data of the user can be reliably protected. Setting the self-destruct modes can erase important data when decoding fails, thereby further protecting user privacy.
[0144] In some embodiments, the one or more user terminals 120 and related devices may include a short-range communication module. The short-range communication module may be a communication module that integrates Bluetooth communication, WI-FI communication, ZigBee communication, SD card reading, USB data transmission, and other functions. The one or more user terminals 120 and the related devices (e.g., the respiratory device 130, the first monitoring device 140, and / or the second monitoring module 150) may be connected through the short-range communication module (e.g., a Bluetooth communication module) . Merely by way of example, the short-range communication module may receive a broadcast signal from other devices (e.g., the one or more respiratory devices) in response to a turn-on instruction input by the user. The broadcast signal may reflect an identification information (ID) of other devices that have a signal connection with the short-range communication module. For example, the short-range communication module may return a list. The list may include the one or more respiratory devices that have a signal connection with the short-range communication module, and the identification information (e.g., a device serial number) of the one or more respiratory devices. In some embodiments, the list may be displayed through an application software installed on the user terminal (s) 120.
[0145] In some embodiments, the one or more user terminals 120 may determine a maximum count of connections of the user terminal 120 based on signal features of an environment where the user terminal 120 is located and signal features of an environment where the related devices are located. The signal features may include a signal intensity, a packet loss rate, a delay value, etc. The maximum count of connections refers to a maximum count of related devices that the user terminal (s) 120 can connect to simultaneously. For example, the greater the signal intensity is, the smaller the packet loss rate is, and the smaller the delay value of the environment where the one or more user terminals 120 and the related devices are located is, the larger the maximum count of connections may be.
[0146] In some embodiments, the one or more user terminals 120 may determine a bandwidth allocated to each connected related device based on one or more communication features with each connected related device. The connected related devices refer to related devices currently connected with the user terminal. The communication features may include a communication frequency and / or an average amount of communication data per minute. For example, for a connected related device with a higher communication frequency and a larger average amount of communication data per minute, a larger bandwidth may be allocated to the connected related device.
[0147] In some embodiments, the one or more user terminals 120 may display a device list searched by the short-range communication module (e.g., the device list may include devices that have the short-range communication module turned on and are located near the one or more user terminals 120) . The user may actively select a device for connection from the device list through the one or more user terminals 120.
[0148] In some embodiments, the one or more user terminals 120 may automatically establish a connection with a related device. In some embodiments, the one or more user terminals 120 may automatically establish a connection with the related device in response to a determination that the user does not actively select a device to be connected from the device list within a delay time. For example, if the user does not confirm the device to be connected within 5 seconds, automatic connection may be activated. The delay time may be a manual preset value, a system default value, etc., such as 5 seconds.
[0149] In some embodiments, the process for the one or more user terminals 120 to automatically establish the connection may include determining a recommended connection device based on the device list and historical connection data of the one or more user terminals 120, and automatically establishing a signal connection with the recommended connection device. The historical connection data may include a device type, identification information, a count of connections, and connection time of historical connected devices. The recommended connection device refers to a device recommended for connection.
[0150] In some embodiments, the one or more user terminals 120 may determine one or more devices each of which has a pairing relationship in the device list as one or more candidate connection devices based on the historical connection data, determine a recommendation value of each of the candidate connection devices through weighted summation based on the count of connections and the connection time between the candidate connection device and the user terminal 120, and determine the recommended connection device based on the recommendation value. The pairing relationship between a device and the user terminal refers to that the device and the user terminal are historically connected through the short-distance communication module. The count of connections refers to a total count of historical connections. The connection time refers to a time point of a last historical connection.
[0151] In some embodiments, a weight of the weighted summation may be positively related to the count of connections and negatively related to a time interval between the connection time and current time. For example, the more counts of connections of a certain candidate connection device with the user terminal is, the higher the weight corresponding to the count of connections is, and the longer the time interval between the connection time of the certain candidate connection device and the current time is, the smaller the weight corresponding to the connection time may be.
[0152] The recommendation value may be used to measure a priority of the recommended connection device to be automatically connected. In some embodiments, the one or more user terminals 120 may select the candidate connection device with a greatest recommendation value as the recommended connection device.
[0153] In some embodiments, the recommended connection device may be determined by the cloud platform 110. In some embodiments, the one or more user terminals 120 may send the device list and the historical connection data to the cloud platform 110. The cloud platform 110 may determine the recommended connection device based on the device list and historical connection data and then feed the recommended connection device back to the one or more user terminals 120.
[0154] In some embodiments, the one or more user terminals 120 may determine a computing terminal based on the count of candidate connection devices. In some embodiments, in response to a determination that the count of candidate connection devices is less than or equal to a computing threshold, the user terminal (s) 120 may be determined as the computing terminal, and in response to a determination that the count of candidate connection devices is greater than the computing threshold, the cloud platform 110 may be determined as the computing terminal. The computing threshold may be preset based on experience or requirements. In some embodiments, the computing threshold may be determined based on a current computing load of the user terminal (s) 120. The greater the current computing load of the user terminal (s) 120 is, the smaller the computing threshold may be.
[0155] The computing terminal refers to a computing device for determining the recommended connection device. The computing terminal may be the user terminal 120 and / or the cloud platform 110. In some embodiments, in response to a determination that the user terminal (s) 120 is determined as the computing terminal, the recommended connection device may be directly determined by the user terminal (s) 120, and in response to a determination that the cloud platform 1 10 is determined as the computing terminal, the recommended connection device may be determined by the cloud platform 110.
[0156] The respiratory device 130 may be a medical device used to provide respiratory support and treat various respiratory diseases and sleep disorders. In some embodiments, the respiratory device 130 may include, but is not limited to, ventilator devices such as a respirator and a high-flow oxygen therapy device, oxygen therapy devices such as an oxygen concentrator and an oxygen bottle, monitoring devices such as a sleep apnea monitor and a sleep apnea screener, etc.
[0157] In some embodiments, the respiratory device 130 may include a respiratory primary device and a respiratory auxiliary device. The respiratory primary device may include a respirator used for treatment. The respiratory auxiliary device refers to a device used for cooperating with the respiratory primary device for treatment. For example, the respiratory auxiliary device may include an oxygen device, a sputum suction device, etc.
[0158] In some embodiments, the respiratory device 130 may communicate with the cloud platform 110 through the one or more user terminals 120. In some embodiments, the respiratory device 130 may be configured to directly communicate with the cloud platform 110.
[0159] In some embodiments, the short-distance communication module may be integrated into the respiratory device 130, and the respiratory device 130 may achieve short-distance communication with the one or more user terminals 120 through the short-distance communication module. In some embodiments, the respiratory device 130 may communicate with the one or more user terminals 120 through the network. Merely by way of example, the respiratory device 130 may communicate with the one or more user terminals 120 (e.g., a PC terminal, a PAD terminal, a mobile terminal, etc., used by the user) via WI-FI communication or Bluetooth communication, etc., to obtain data input by the one or more user terminals 120. Furthermore, the one or more user terminals 120 may transmit the data to the cloud platform 110 via WI-FI communication or Bluetooth communication, etc. The respiratory device 130 may store the data in an SD card, and the user may download the data stored in the SD card to the PC terminal, the PAD terminal, and / or the mobile terminal device. As another example, the cloud platform 110 may transmit parameter setting information of the respiratory device 130 to the one or more user terminals 120 via WI-FI communication or Bluetooth communication, etc. In addition, the one or more user terminals 120 may send the parameter setting information to the respiratory device 130 through the short-range communication module.
[0160] In some embodiments, the respiratory device 130 may transmit the basic data of the respiratory device and the usage data of the respiratory device to the one or more user terminals 120 through the short-range communication module and / or network. More descriptions regarding the basic data of the one or more respiratory devices and the usage data of the one or more respiratory devices may be found in FIG. 2 and related descriptions thereof.
[0161] The first monitoring device 140 may be configured to collect data of the first subject during respiratory treatment and send the data to the one or more user terminals 120. In some embodiments, the first monitoring device may include a ventilation meter, a pressure sensor, or the like. Correspondingly, the data during the respiratory treatment may include at least one of a tidal volume, an airway pressure, a respiratory frequency, etc.
[0162] In some embodiments, the first monitoring device 140 may be independent from the respiratory device 130 or may be a part of the respiratory device 130.
[0163] The second monitoring device 150 may be an associated monitoring device operating in conjunction with the one or more respiratory devices. In some embodiments, the second monitoring device 150 may be configured to collect physiological data of the first subject. For example, the second monitoring device 150 may obtain the physiological data of the user during sleep, such as electroencephalogram (EEG) , electrocardiograph (ECG) , air flow, snoring, etc.
[0164] In some embodiments, the second monitoring device 150 may be a wearable device, such as a smart bracelet, smart shoes and socks, smart glasses, a smart helmet, a smart watch, a smart clothe, a smart backpack, a smart accessory, or the like, or any combination thereof.
[0165] In some embodiments, the second monitoring device 150 may be independent from the one or more user terminals 120 or may be integrated into or built into the one or more user terminals 120.
[0166] In some embodiments, the one or more user terminals 120 may establish a signal connection with the second monitoring device 150. In some embodiments, the respiratory device 130, the first monitoring device 140, and / or the second monitoring device 150 may upgrade software and / or firmware integrated therein based on upgrade instructions input by the user. The upgrade instructions refer to control instructions that control related software and / or firmware to be upgraded. In some embodiments, the respiratory device 130, the first monitoring device 140, and / or the second monitoring device 150 may perform recovery processing on the upgraded software and / or firmware based on a version recovery instruction input by the user. The upgrade instructions and / or version recovery instructions may be input and obtained by the user from the one or more user terminals 120.
[0167] In some embodiments, the system 100 for respiratory treatment management may also include other structures.
[0168] For example, the system 100 for respiratory treatment management may also include a storage device configured to store data or information generated by other devices. In some embodiments, the storage device may store data collected by the respiratory device 130, such as the basic data of the one or more respiratory devices and the usage data of the one or more respiratory devices, etc. In some embodiments, the storage device may store data and / or information processed by the cloud platform, such as the recommended setting parameters, etc.
[0169] As another example, the system 100 for respiratory treatment management may also include a network and / or other components that connect the system to external resources.
[0170] More descriptions regarding the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, the display interface, the recommended setting parameters, the target report, etc., may be found in FIGs. 2-14 and related descriptions thereof.
[0171] Exemplary illustrations of application scenarios of the system 100 for respiratory treatment management are provided below with reference to FIG. 1 B. FIG. 1 B is a schematic diagram illustrating an application scenario of the system 100 for respiratory treatment management according to some embodiments of the present disclosure.
[0172] In some embodiments, as shown in FIG. 1 B, the system 100 for respiratory treatment management may perform data collection from related devices (e.g., the respiratory device 130, the first monitoring device 140, and / or the second monitoring device 150) used by a first subject 170-1 through a first terminal 120-1. The first terminal 120-1 may realize download management, upload management, interaction management, hierarchical classification, and other applications of the data by communicating with various cloud platforms (e.g., at least one of a cloud platform 110-1, a cloud platform 110-2, a cloud platform 110-3, and a cloud platform 110-4) . Second subjects, such as a doctor, an agent, etc., may access the data information through the cloud platform 110. In some embodiments, the system 100 for respiratory treatment management may learn, analyze, and summarize the data related to respiratory treatment through an Artificial Intelligence (AI) module 180 to triggering an integrated hardware and software operation and achieve synchronized terminal response.
[0173] In some embodiments, various components of the system 100 for respiratory treatment management may perform data interaction by communicating with each other in various ways. In some embodiments, the system 100 for respiratory treatment management may include a server (data center) 160. The server 160 may analyze, learn, summarize, and provide feedback on the data related to respiratory treatment management.
[0174] In some embodiments, the server 160 may communicate with the cloud platform 110 through a network. The server 160 may perform data interaction through at least one of the cloud platform 110-1, the cloud platform 110-2, the cloud platform 110-3, and the cloud platform 110-4.
[0175] For example, the cloud platform 110-1 and the cloud platform 110-3 may be used by a foreign user 170-4 (e.g., the second subject, such as an agent, a doctor) to perform data interaction with the server 160. Merely by way of example, the user 170-4 may obtain one or more configuration parameters of the respiratory device 130, etc., from the server 160 based on the cloud platform 110-1 and the cloud platform 110-3 through the network, or upload the data related to respiratory treatment management to the server 160. The data interaction between the cloud platform 110-1 and the cloud platform 110-3 may be carried out through an Application Programming Interface (API) .
[0176] The cloud platform 110-2 and the cloud platform 110-4 may be used by a domestic user (e.g., a domestic agent 170-3) to perform data interaction with the server 160. Merely by way of example, the domestic agent 170-3 may directly perform data interaction with the server 160 through the cloud platform 110-2 based on the network, or may first perform data interaction with the cloud platform 110-4 based on the network, and then perform data interaction with the cloud platform 110-2 through the cloud platform 110-4, and ultimately realize data interaction with the server 160 through the cloud platform 110-2.
[0177] In some embodiments, the second subject 170-2 may perform data interaction with the cloud platform 110-2 or the cloud platform 110-4 through the second terminal 120-2 to ultimately realize data interaction with the server. For example, the second subject 170-2 (e.g., a doctor) may log into the software of a hospital through the second terminal 120-2, generate a data acquisition instruction or a data upload instruction by performing an operation such as data downloading, data uploading, etc., in the software, and upload the data acquisition instruction and / or the data upload instruction to the server 160 through the cloud platform 110-2 or the cloud platform 110-4 to acquire or upload the data.
[0178] In some embodiments, the server 160 may be provided with technical support and technical maintenance by a platform operator 170-5. In some embodiments, the platform operator 170-5 may also perform data operation of the cloud platform 110.
[0179] In some embodiments, the first subject 170-1 (e.g., a patient) may perform data interaction with the cloud platform 110-1 and / or the cloud platform 110-2 through a first terminal 120-1, which in turn enables data interaction with the server 160. The first terminal 120-1 may include a Web terminal, a tablet, a mobile phone, or the like. For example, the first subject 170-1 may upload collected data related to respiratory treatment to the cloud platform 110-1 and / or the cloud platform 110-2 through the first terminal 120-1 for further uploading the data to the server 160. Similarly, the first subject 170-1 may obtain the data (e.g., a recommended setting parameter of the one or more respiratory devices, a target report, etc. ) from the server 160 based on the first terminal 120-1 through the cloud platform 110-1 and / or the cloud platform 110-2.
[0180] In some embodiments, the first subject 170-1 may perform data interaction with the respiratory device 130 based on the first terminal 120-1 through at least one of WI-FI communication, cellular communication, SD storage, and the short-range communication module. For example, the first terminal 120-1 may send the data related to respiratory treatment management, or send the recommended setting parameter to the respiratory device 130. As another example, the respiratory device 130 may send the usage data of the one or more respiratory devices, etc., to the first terminal 120-1. In some embodiments, the first subject 170-1 may perform data interaction with the first monitoring device 140 and / or the second monitoring device 150 based on the first terminal 120-1 through at least one of WI-FI communication, TF card / micro card analysis, an API of a monitoring software, and the short-range communication module. For example, the first monitoring device 140 and / or the second monitoring device 150 may send the physiological data of the first subject, the respiratory treatment data of the first subject, etc., to the first terminal 120-1.
[0181] It should be noted that the system 100 for respiratory treatment management is provided merely for illustrative purposes and is not intended to limit the scope of the present disclosure. For those having ordinary skills in the art, various variations and modifications can be made based on the description in the present disclosure. For example, the system 100 for respiratory treatment management may also include a database, an information source, etc. As another example, the system 100 for respiratory treatment management may be implemented on other devices to achieve similar or different functionality. However, such variations and modifications may not depart from the scope of the present disclosure.
[0182] FIG. 2 is a flowchart illustrating an exemplary method for respiratory treatment management according to some embodiments of the present disclosure. In some embodiments, a process 200 may be performed by a cloud platform. The operations of process 200 presented below is intended to be illustrative. In some embodiments, process 200 may be accomplished with one or more additional operations not described and / or one or more operations not discussed. In addition, the order in which the operations of process 200 as illustrated in FIG. 2 and described below is not intended to be limiting. As shown in FIG. 2, process 200 may include the following operations.
[0183] In 210, at least one of user basic data, basic data of a respiratory device, usage data of the respiratory device, or respiratory treatment data of a first subject may be obtained based on a user terminal.
[0184] The user basic data refers to data information related to a user (e.g., a first subject and / or a second subject) .
[0185] In some embodiments, the user basic data may include basic data of the first subject and / or basic data of a second subject. The first subject may include a subject who receives a service (e.g., a treatment) provided by the respiratory device. The first subject may also be referred to as a service requester. For example, the first subject may include a human who has a respiratory disturbance, a sportsman, etc. The second subject may include a subject who provides a service (e.g., a treatment) to the first subject based on the respiratory device. The second subject may also be referred to as a service provider. For example, the second subject may include a doctor, a nursing worker, a coach, a family of the first subject, etc.
[0186] The basic data of the first subject refers to data information related to a basic situation of the first subject. For example, the basic data of the first subject may include basic information (e.g., a gender, an age, a weight, an occupation, an education, or other information contained in the first subject's account) of the first subject, a past medical history (e.g., a respiratory disease type, a disease extent, or a treatment) of the first subject, etc., or any combination thereof. The basic data of the first subject may be obtained by the first subject inputting the basic data in a first terminal. For descriptions of the first terminal, please refer to related contents in FIG. 1A.
[0187] The basic data of the second subject refers to data related to a basic situation of the second subject. For example, the basic data of the second subject may include a gender, an age, an education, a specialty in disease, a time of medical practice, a count of subjects treated in history, a professional title, a department of the hospital of the second subject, etc., or any combination thereof. The basic data of the second subject may be obtained by the second subject or the hospital inputting the basic data in a second terminal. For descriptions of the second terminal, please refer to related contents in FIG. 1A.
[0188] The basic data of the one or more respiratory devices refers to data related to a basic situation of the respiratory device. For example, the basic data of the respiratory device may include a type, a model number, a serial number, a software version, a pairing situation, etc., of the respiratory device. The pairing situation refers to information that reflects pairing of the respiratory device with the first subject, e.g., a pairing relationship between a product serial number of the respiratory device (Device SN) , a usage status of the respiratory device (Device Status, e.g., in use or not in use) , a device model (e.g., G3A20 or G3C20) , or a device type (e.g., Auto CPAP or BPAP) and an identification number of the first subject (first subject ID) , etc.
[0189] The usage data of the respiratory device refers to data related to the respiratory device when respiratory treatment is performed on the first subject. For example, the usage data of the one or more respiratory devices may include a setting parameter of the respiratory device and usage data of the first subject. The setting parameter of the respiratory device may include a tidal volume, a respiratory rate, an inspiratory-expiratory ratio, a pressure trigger range, a flow rate trigger range, an inspiratory time, an oxygen concentration, etc. The usage data of the first subject may include a type, a model number, a general parameter, a usage frequency, a usage duration, a usage period, a count of usage, a usage duration each time, etc., of the respiratory device used by the first subject The general parameter of the respiratory devices may include a treatment pressure, auto-on, auto-off, delayed shutdown, reslex\ISes, ESens, a respiratory frequency, an expiration-inspiration ratio, a minimum inspiratory time, a maximum inspiratory time, a high-pressure alarm, a low-pressure alarm, etc. In some embodiments, the setting parameter of the respiratory device may be related to a condition and a treatment goal of the first subject. For example, a healthcare professional may determine or adjust setting parameters according to different first subjects and different treatment goals. The usage data of the one or more respiratory devices may be determined by healthcare professional inputting. In some embodiments, the setting parameter of the respiratory devices may be also determined by the cloud platform. More descriptions may be found in FIG. 10, and the relevant descriptions thereof.
[0190] The respiratory treatment data refers to data related to usage of the respiratory devices of the first subject. For example, the respiratory treatment data may include an inspiratory tidal volume, an expiratory tidal volume, a respiratory rate, a peak airway pressure, a minute ventilation, air leakage, a plateau pressure, an inspiratory-expiratory ratio, airway resistance, lung compliance, etc., that may be monitored when the first subject actually uses the respiratory device.
[0191] In some embodiments, the respiratory treatment data may be autonomously collected, recorded in a local repository, and / or migrated to be stored in the user terminal by the respiratory device or a first monitoring device.
[0192] In some embodiments, the cloud platform may read the user basic data, the basic data of the respiratory device, the usage data of the one or more respiratory devices, and / or the respiratory treatment data of the first subject from the user terminal, a storage device of the respiratory devices and / or the first monitoring device, and / or via an interface. The interface may include a program interface, a data interface, a transmission interface, etc., or any combination thereof. More descriptions regarding the user terminal may be found in FIG. 1A and the relevant descriptions thereof. In some embodiments, the data may be transmitted through the data interface using a preset communication protocol. In some embodiments, the preset communication protocol may include a Health Level 7 (HL7) protocol, a Digital Imaging and Communications in Medicine (DICOM) protocol, an American Society for Testing and Materials (ASTM) protocol, an ISO / IEEE 11073 standard series protocol, a Gateway Discovery Protocol (GDP) , a Modbus protocol, a dedicated communication protocol, or any combination thereof.
[0193] In some embodiments, when the dedicated communication protocol is used for data transmission, the cloud platform and the respiratory device may actively send request messages to each other, and each of the request messages may have a reply message corresponding to the each of the request messages. A command code of the request message that is actively sent may be of a first type (e.g., an even number) , and a command code of the reply message may be of a second type (e.g., an odd number) . Message identities (IDs) of different request messages may be strictly increased (e.g., increasing from 0) , and a message ID of the reply message may be the same as the message ID of the request message. In some embodiments, the message ID may include a sending timestamp and a sending sequence. The sending timestamp refers to time information of the request message sent from a data sender, and the sending sequence refers to an order of request messages during current communication connection. In some embodiments, when performing data transmission communication, the cloud platform and the respiratory device may check a preset encoding algorithm (e.g., a Cyclic Redundancy Check (CRC) algorithm) of the request message and / or the reply message, in response to a failure of the check, a sender of the request message and / or the reply message may resend the request message and / or the reply message, and in response to an absence of data transmission of the cloud platform and / or the respiratory device within a preset time period, the communication connection may be disconnected. In some embodiments, the preset time period may be determined based on a difference between a time when the request message is actually received last time and the sending timestamp in the message ID. For example, the difference between the time when the request message is actually received last time and the sending timestamp in the message ID, or a product of the difference and a time coefficient may be determined as the preset time period. The time coefficient may be preset by the system or manually (e.g., 1.5 or 2) . When there is no data transmission within the preset time period, the cloud platform or the respiratory equipment may no longer need data transmission, and the communication connection may be turned off to save communication resources.
[0194] The device (e.g., the respiratory device, the first monitoring device, and / or second monitoring device) and the cloud platform may both serve as the sender of the request message. The sender of the request message refers to a data terminal that sends the request message. The request message may be related to a type of the data terminal. For example, if the data terminal is the respiratory device, the request message may include a model number of the respiratory device, an identity (ID) of the respiratory device, a data type and a data amount of transmission data, etc.
[0195] In some embodiments, when the device is the sender of the request message, in response to the cloud platform firstly receiving the request message sent by the device, the cloud platform may determine a communication protocol type of the sender of the request message based on the request message; determine a data format of the request message based on the communication protocol type; and decode the request message based on the data format of the request message and determine a data format of the reply message. In some embodiments, in response to the cloud platform firstly receiving the request message sent by the device, the cloud platform may determine the sender of the request message through an Internet Protocol (IP) address that sends the request message, a device identifier, etc., and query a preset communication protocol type in the sender of the request message. In some embodiments, after determining the communication protocol type, the cloud platform may determine the data format of the request message based on a data format specified in the communication protocol type to decode the request message and obtain the data type and the data volume of the transmitted data. The cloud platform may also determine, based on the data format specified in the communication protocol type, the data format of the reply message corresponding to the data format specified in the communication protocol type.
[0196] In some embodiments, when the cloud platform serves as the sender of the request message, the cloud platform may obtain a communication protocol type of a receiver of the request message, determine the request message based on a data format and content of the request message corresponding to the communication protocol type, and send the request message through the communication protocol type. For example, the cloud platform may determine the communication protocol type of the receiver of the request message based on relevant information (e.g., a device type or a device ID) of the receiver of the request message. The cloud platform may obtain the request message by encoding based on the communication protocol type and the content of the request message and send the request message to the receiver of the request message.
[0197] In some embodiments of the present disclosure, the communication protocol type of the sender and / or the receiver of the request message may be automatically identified when the data is transmitted, which is conducive to being compatible with communication protocols of different enterprises and different devices. The treatment data and device attribute information of the respiratory device may be sent to the cloud platform through the dedicated communication protocol, so that the treatment data may be uploaded to the cloud platform in real time, and the user (e.g., a doctor or a manufacturer) may understand the treatment data of the first subject timely through the cloud platform, thereby adjusting the treatment plan according to the treatment result to solve the problems of limited communication and cumbersome operation and maintenance of the respirator in the prior art.
[0198] In some embodiments, the cloud platform may also perform a check operation before data transmission. In some embodiments, the check operation may include checking legitimacy of modified setting data and / or checking one or more user identities associated with the transmitted data. Taking the respiratory device as an example, in some embodiments, a Rivest-Shamir-Adleman (RSA) algorithm may be made to perform a first identity check. The respiratory device may be built with an RSA public key at the factory, and a paired private key may be saved in the cloud platform. The respiratory device may encrypt a string of numbers using the public key and send the string of encrypted numbers to the cloud platform. The cloud platform may usually decrypt the string of encrypted numbers using the private key, perform an operation of a certain type of algorithm on the data, encrypt a result using the private key, and return the encrypted result to the respiratory device. The respiratory device may decrypt, after receiving the data, the encrypted result using the public key and obtain the result. If the data meets expectations, the first identity check may be considered to be passed. If the data does not meet the expectations, the first identity check may be considered not to be passed. If the first identity check is failed, communication connection between the respiratory device and the cloud platform may be disconnected. If the first identity check is passed, the respiratory device and the cloud platform may be confirmed through an encryption key. Although the RSA algorithm is highly secure and difficult to be cracked, an operation speed thereof may be too slow. Therefore, after confirming the identity of the cloud platform, the respiratory device may generate a Triple Data Encryption Standard (3DES) key, encrypt the 3DES key using the RSA public key, and send the encrypted key to the cloud platform. The cloud platform may decrypt the encrypted key using the RSA private key and confirm that a format of the 3DES key is correct. Communication data between the respiratory device and the cloud platform may be encrypted using a 3DES encryption algorithm, and the cloud platform may feedback a key confirmation message to the respirator. If the cloud platform does not feedback the key confirmation message to the respirator, the confirmation may be considered to be failed.
[0199] In some embodiments, after the first identity check and encryption key confirmation, the cloud platform may need to confirm an identity of the respiratory device, that is, the respiratory device and the cloud platform may perform a second identity check. In the embodiment, the respiratory device may send the attribute information to the cloud platform. The attribute information may include an SN and an identity identification code of the respiratory device. The SN and the identification code may be uploaded to a cloud platform database before the respiratory device leaves the factory. If the second identity check passes, the respirator may upload real-time information including the treatment data. If the second identity check is failed, the communication connection between the respiratory device and the cloud platform may be disconnected.
[0200] In some embodiments, the second identity check may include that: when receiving the attribute information of the respiratory device, the cloud platform may compare the attribute information with information in the cloud platform database; If the attribute information is the same as the information in the cloud platform database, the second identity check may be passed and the communication may continue; and If the attribute information is different from the information in the cloud platform database, the cloud platform may actively disconnect the communication connection. In the embodiment, to prevent brute-force attacks, the cloud platform may record information such as the IP address and the uploaded SN of the respiratory device. If the respiratory device with a certain IP address frequently fails to check, the cloud platform may block the IP address for a time period.
[0201] In some embodiments, the cloud platform may determine, according to a check success rate of the check operation in a historical time period and a data sensitivity of the transmission data during current data transmission, a check intensity of the current data transmission, and determine a count of checks, a check algorithm, and a communication protocol type after the check is successful based on the check intensity of the current data transmission. In some embodiments, the cloud platform may determine a ratio of a count of successful checks to a total count of check operations in the historical time period as the check success rate.
[0202] The check intensity refers to a strict degree of the check operation. The higher the check intensity is, the stricter the check operation may be. In some embodiments, the cloud platform may determine the check intensity of the current data transmission by looking up a table based on the check success rate and the data sensitivity of the transmission data during the current data transmission. The table may include a correspondence between different check success rates, different data sensitivities, and different check intensities. For example, the higher the success rate is, and the higher the data sensitivity is, the greater the check intensity may be. In some embodiments, the correspondence between the different check success rates, the different data sensitivities, and the different check intensities may be predetermined based on historical data or prior knowledge. The data sensitivity of the transmitted data may be determined based on the data type of the transmitted data. More descriptions regarding the data sensitivity may be found in FIG. 1A and the descriptions thereof.
[0203] In some embodiments, different check intensities may correspond to different counts of checks, different check algorithms, and different communication protocol types. For example, the greater the check intensity is, and the larger the count of checks is, the more complex (time-consuming and resource-consuming) but more secure the check algorithm may be selected, and the higher the security factor of a selected communication protocol type after successful check may be. A correspondence between the check intensities and the counts of checks, the check algorithms, and the communication protocol types may be predetermined based on historical data or prior knowledge.
[0204] In 220, target information may be generated based on at least one of the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data, and the target information may be sent to the user terminal.
[0205] The target information refers to information that the user needs to obtain. For example, the target information may include information displayed by the user terminal, the setting parameter of the respiratory device, recent treatment data of the first subject, etc.
[0206] In some embodiments, the cloud platform may generate the target information based on at least one of the user basic data, the basic data of the one or more respiratory device, the usage data of the respiratory device, or the respiratory treatment data in various ways.
[0207] For example, the cloud platform may generate the target information based on at least one of the user basic data, the basic data of the one or more respiratory devices, the usage data of the respiratory device, or the respiratory treatment data through a preset algorithm, a machine learning model, etc.
[0208] In some embodiments, the target information may include a display interface.
[0209] In some embodiments, the cloud platform may generate the display interface based on at least one of the user basic data, the basic data of the respiratory device, the usage data of the respiratory device, or the respiratory treatment data.
[0210] The display interface refers to a content interface that is displayed to the user in a display screen of the user terminal. For example, the display interface may be a display interface of a user terminal (e.g., a computer or a mobile phone) or a display interface of a display screen on the respiratory device.
[0211] In some embodiments, the display interface may include at least one of statistical information of the usage data of the respiratory device, statistical information of the respiratory treatment data, a respiratory trend of the first subject, a recommended treatment plan of the first subject, or a supervisory relationship between the first subject and the second subject.
[0212] The statistical information of the usage data of the respiratory device refers to a statistical result of the usage data of the one or more respiratory devices during a period of time. For example, the statistical information of the usage data of the respiratory device may include statistical data of the setting parameter of the respiratory device when the first subject is treated during a certain period of time, statistical data of a length of time the first subject uses the respiratory device, etc.
[0213] The statistical information of the respiratory treatment data refers to a statistical result of the respiratory treatment data during a period of time. For example, the statistical information of the respiratory treatment data may include statistical data such as an inspiratory tidal volume, an expiratory tidal volume, a respiratory rate, or a pressure, which are monitored during a certain period of time when the first subject actually uses the respiratory device.
[0214] In some embodiments, the cloud platform may obtain the usage data of the respiratory device and the respiratory treatment data during a period of time and obtain the statistical information of the usage data of the respiratory device and statistical information of the respiratory treatment data through statistical analysis.
[0215] In some embodiments, the statistical information of the usage data of the respiratory device and the statistical information of the respiratory treatment data may be represented in text. As shown in FIG. 4A, the display interface may include the statistical information of the usage data of the respiratory device such as a count of days with a respiratory treatment within a period, a ratio of the count of days with a respiratory treatment to a total count of days within the period, a total usage time, a count of days with a daily usage time greater than or equal to 4h, and a ratio of the count of days with a daily usage time greater than or equal to 4h to the total count of days, a count of days with a daily usage time less than 4h, a ratio of the count of days with a daily usage time less than 4h to the total count of days, an average daily usage time (which is determined based on the count of days with a respiratory treatment within the period) , an average daily usage time (which is determined based on the total count of days) , a count of days without respiratory treatment, a ratio of the count of days without respiratory treatment to the total count of days during the period, etc. As shown in FIG. 4A, The display interface may include the statistical information of respiratory treatment data such as a mean value, an intermediate value, a 95th percentile, or a maximum value of Inspired Positive Airway Pressure (IPAP) with a unit of cmH2O, and the statistical information of respiratory treatment data such as a mean value, an intermediate value, a 95th percentile, or a maximum value of Expiratory Positive Airway Pressure (EPAP) with a unit of cmH2O. IPAP and EPAP are parameters in the Bi-Level Positive Airway Pressure (BiPAP) ventilation mode, and a difference between IPAP and EPAP is (Pressure Support, PS) . In some embodiments, the statistical information of the usage data of the respiratory devices may be represented in a chart. As shown in FIG. 4B, the respiratory treatment data such as time information of the respiratory treatment, the pressure, or an IPAP of the first subject during a period of time may be represented in a bar chart.
[0216] In some embodiments, different statistical information of the usage data of different respiratory devices and / or respiratory treatment data may be displayed in different types of terminals. For example, the statistical data of the setting parameters of the respiratory devices when the first subject is treated during a certain period of time, the statistical data of a length of time the first subject using the respiratory devices, etc., may be displayed in the first terminal. The statistical information of the usage data of the respiratory device of a first subject whose data is abnormal among a plurality of first subjects may be displayed in the second terminal. As another example, the statistical data such as an inspiratory tidal volume, an expiratory tidal volume, a respiratory rate, or a pressure monitored during a certain period of time during which the first subject actually uses the respiratory devices may be displayed in the first subject terminal. The statistical information of the respiratory treatment data of the first subject whose data is abnormal among the plurality of first subjects may be displayed in the second terminal. Whether the data is abnormal may include whether the data exceeds an industry standard. As shown in FIG. 4C, a count of persons with a High AHI, a count of persons with High Leak, a count of persons with Low usages (usage time <4h per day continues 10 days after 90 days) , etc., may be displayed in the display interface of the second terminal.
[0217] The respiratory trend of the first subject may reflect probability information about a developing direction of a respiratory disease that the first subject suffers from. In some embodiments, the respiratory trend may include a disease progression trend of the first subject. The disease progression trend refers to a probability of a next stage of the respiratory disease that the first subject currently suffers from. For example, if the respiratory disease that the first subject currently suffers from is an upper respiratory tract infection, then the next stage thereof may include a lower respiratory tract infection, pneumonia, bronchitis, sinusitis, otitis media, or the like, or any combination thereof. Each specific stage in the next stage may have a specific probability value.
[0218] In some embodiments, the cloud platform may construct a disease comparison table based on historical basic data of the first subject, historical respiratory treatment data, historical physiological data, and actual diseases corresponding to the historical basic data of the first subject, the historical respiratory treatment data, and the historical physiological data. The cloud platform may determine a respiratory trend of the first subject by querying the disease comparison table based on current basic data of the first subject, current respiratory treatment data, and current physiological data. The physiological data of the first subject refers to relevant data reflecting a physiological condition of the body of the first subject. For example, the physiological data may include a body temperature, a heart rate, a respiratory rate, a respiratory oxygen saturation, etc. The physiological data of the first subject may be collected by a second monitoring device and transmitted to the user terminal for storage. More descriptions regarding the second monitoring device may be found in FIGs. 1A-1 B and the relevant descriptions thereof.
[0219] In some embodiments, the cloud platform may determine the respiratory trend of the first subject based on the basic data, the respiratory treatment data, and the physiological data of the first subject through a first prediction model. More descriptions regarding determining the respiratory trend may be found in FIG. 3 and the relevant descriptions thereof.
[0220] The respiratory trend of the first subject may also be determined based on an input of the second subject through the second terminal. The second subject may make a manual assessment of the respiratory trend of the first subject in conjunction with the usage data of the respiratory device and the respiratory treatment data of the first subject.
[0221] The recommended treatment plan of the first subject refers to a treatment program that recommends setting the respiratory device with a specific parameter and / or using the respiratory device in a specific manner (e.g., using at a specific usage frequency or for a specific usage duration) to achieve the treatment of the first subject. For example, the recommended treatment plan of the first subject may include a type of the respiratory device, a setting parameter of the respiratory device, a usage frequency, a usage duration, etc., that are recommended for the first subject.
[0222] In some embodiments, the recommended treatment plan may be determined by a professional, such as a doctor, and uploaded to the cloud platform through the second terminal.
[0223] In some embodiments, the cloud platform may send the recommended treatment plan to the first subject and / or second subject and determine, based on feedback of the first subject and / or second subject, a treatment plan actually used for the first subject. For example, according to the feedback, when the first subject and the second subject approve a certain recommended treatment plan at the same time, the recommended treatment plan may be determined to be the treatment plan actually used for the first subject.
[0224] In some embodiments, the cloud platform may generate a machine instruction based on the treatment plan actually used for the first subject and send the machine instruction to the device (e.g., the respiratory device) used for the treatment of the first subject. The machine instruction may include a control parameter (e.g., a time to start running or a parameter with which to run) of the respiratory device.
[0225] In some embodiments, the cloud platform may determine, based on the monitoring data (e.g., respiratory treatment data) of the first subject in the treatment process, whether the treatment plan needs to be adjusted (e.g., there is an abnormality in the respiratory treatment data) ; in response to the treatment plan needing to be adjusted, the cloud platform may regenerate a new recommended treatment plan and send the new recommended treatment plan to the first subject and / or the second subject, determine, based on the feedback of the first subject and / or second subject, an updated actual treatment plan actually used for the first subject and generate an updated machine instruction, and send the updated machine instruction to the device used for the treatment of the first subject.
[0226] In some embodiments, the cloud platform may determine one or more first candidate treatment plans and a predicted treatment result corresponding to each of the one or more first candidate treatment plans and generate the recommended treatment plan based on the first candidate treatment plans and the predicted treatment results corresponding to the first candidate treatment plans. More descriptions regarding determining the recommended treatment plan may be found in FIG. 5 and the relevant descriptions thereof.
[0227] The supervisory relationship between the first subject and the second subject refers to a correspondence between the second subject and the first subject. For example, the supervisory relationship between the first subject and the second subject may include a correspondence between a patient and an attending doctor, a correspondence between a patient and an augmented doctor, a correspondence between a patient and a nurse, or the like, or any combination thereof.
[0228] In some embodiments, the supervisory relationship between the first subject and the second subject may be determined by a professional, such as a doctor, and uploaded to the cloud platform through the second terminal.
[0229] In some embodiments, the cloud platform may send the regulatory relationship between the first subject and the second subject to the first subject and / or second subject, determine an actual regulatory relationship based on the feedback of the first subject and / or second subject, and adjust according to the actual regulatory relationship. For example, if the first subject and the second subject approve a certain regulatory relationship at the same time according to the feedback, the regulatory relationship may be determined as the actual regulatory relationship.
[0230] In some embodiments, the cloud platform may determine a matching degree between the second subject and the first subject based on the basic data of the first subject, the respiratory trend, and the basic data of the second subject, thereby determining the supervisory relationship. More descriptions regarding determining the supervisory relationship between the first subject and the second subject may be found in FIG. 6 and the relevant descriptions thereof.
[0231] In some embodiments, the display interface may also include a cooperation degree trend of the first subject. The cooperation degree trend refers to a change in a treatment cooperation degree of the first subject during subsequent respiratory treatments. For example, the cooperation degree trend may include a plurality of predicted treatment cooperation degree at a plurality of future time points. The treatment cooperation degree refers to a cooperation degree of the first subject in the respiratory treatment according to the treatment plan. The higher the treatment cooperation degree, the more actively the first subject cooperates with the treatment plan for respiratory treatment. In some embodiments, the cloud platform may predict the cooperation degree trend of the first subject through a cooperation degree prediction model based on the basic data of the first subject, the usage data of the respiratory device, and the physiological data of the first subject. The cooperation degree prediction model is a model used to determine the cooperation degree trend. In some embodiments, the cooperation degree prediction model may be a machine learning model, such as a Neural Network (NN) model, etc. In some embodiments, an input of the cooperation degree prediction model may include the basic data of the first subject, the usage data of the respiratory device, and the physiological data of the first subject. More descriptions regarding generating the basic data of the first subject, the usage data of the respiratory device, and the physiological data of the first subject may be found in FIG. 2 and relevant descriptions thereof. In some embodiments, the cooperation degree prediction model may be trained based on a large number of training samples with labels. The training manner of the cooperation degree prediction model is similar to a training manner of an effect prediction model. For more explanation, please refer to the description below in FIG. 2. In some embodiments, the training samples used to train the cooperation degree prediction model may include sample basic data and sample physiological data of a sample first subject, and sample usage data of the respiratory device, and the training samples may be obtained based on historical data. The label corresponding to each of the training samples may be an actual treatment cooperation degree within a time period after a time point corresponding to the training sample. In some embodiments, the cloud platform may determine the actual treatment cooperation degree corresponding to the training sample through weighted calculation based on a daily usage duration and a count of consecutive days of usage of the sample usage data of the respiratory device in the training sample as a label. In some embodiments of the present disclosure, by determining the cooperation degree trend, it is helpful to analyze the cooperation degree of the first subject in respiratory treatment according to the treatment plan. In some embodiments, the cloud platform may determine a device optimization plan in response to a treatment cooperation degree at a future time point meeting a first preset condition. The device optimization plan refers to a plan that optimizes treatment reminder parameters, treatment reminder information, etc. For example, when the treatment cooperation degree of the first subject at the future time point meets the first preset condition, a frequency of reminders may be increased. As another example, when the treatment cooperation degree of the first subject at the future time point meets the first preset condition, the treatment reminder information displayed on the patient terminal may be highlighted. For more explanation about the treatment reminder parameters and treatment reminder information, refer to FIG. 5 and related description thereof. In some embodiments, the first preset condition may be that the treatment cooperation degree is lower than a preset lower limit value of cooperation degree. The preset lower limit value of cooperation degree may be preset manually or by the system. In some embodiments, the cloud platform may update the recommended treatment plan and send the recommended treatment plan to the first subject in response to the treatment cooperation degree of the first subject at the future time point meeting a second preset condition. For example, the cloud platform may increase a duration and frequency of daily treatment in response to the treatment cooperation degree of the first subject at the future time point meeting a second preset condition to make the treatment more intense. In some embodiments, the second preset condition may be that the treatment cooperation degree is higher than a cooperation degree threshold. The cooperation degree threshold may be preset manually or by the system. The cooperation degree threshold may be greater than or equal to the preset lower limit value of cooperation degree.
[0232] In some embodiments, the display interface may include a plurality of sub-areas, the plurality of sub-areas may correspond to different treatment parameters, and a display content of each of the plurality of sub-areas may include a type of a treatment parameter and abnormal statistical data of the treatment parameter.
[0233] Each of the plurality of the sub-areas refers to a portion of an area in the display interface. In some embodiments, the plurality of sub-areas may be a plurality of portions of the area of the display interface in different sizes.
[0234] The treatment parameter refers to a parameter related to a treatment of the first subject. The plurality of sub-areas may correspond to the different treatment parameters. As shown in FIG. 4D, the plurality of sub-areas are included in the display interface 400. A treatment parameter corresponding to sub-area 401 is “asupervisory relationship involving between first subject A and second subject B. ” A treatment parameter corresponding to sub-area 402 is “the respiratory trend of the first subject A is a chronic obstructive pulmonary disease. ” A treatment parameter corresponding to sub-area 403 is “the heart rate of the first subject A is 80 beats / min. ” A treatment parameter corresponding to sub-area 404 is “actual data and predicted data of an Apnea-Hypopnea Index (AHI) of first subject A. ” A treatment parameter corresponding to sub-area 405 is “the recommended treatment plan of first subject A. " A treatment parameter corresponding to sub-area 406 is “usage data (e.g., an actual usage duration, a required usage duration, etc. ) of the respiratory device. ” A treatment parameter corresponding to sub-area 407 is “abnormal statistical data of first subject A. ” A treatment parameter corresponding to sub-area 408 is “an abnormal data graph of first subject A. ”
[0235] The abnormal statistical data refers to statistical data that reflect an abnormality of the treatment parameter. For example, the abnormal statistical data may include a type of data where the abnormality exists, an abnormality degree, etc.
[0236] In some embodiments, the cloud platform may determine the abnormal statistical data based on the treatment parameter and abnormality determination conditions corresponding to different types of treatment parameters.
[0237] The abnormality determination conditions corresponding to different types of treatment parameters may be different. For example, the abnormality determination conditions may be determined based on industry-standard data or priori knowledge.
[0238] In some embodiments, a detailed data page of a treatment parameter corresponding to a sub-area may be displayed by a user clicking on the sub-area. For example, when the user clicks on a sub-area displaying "respiratory treatment data" , a detailed data page corresponding to the respiratory treatment data may be generated and displayed. The detailed data page may include the inspiratory tidal volume, the expiratory tidal volume, the respiratory rate, the peak airway pressure, the minute ventilation volume, the air leakage, the plateau pressure, the inspiratory-expiratory ratio, the airway resistance, the lung compliance, etc., that may be monitored when the first subject actually uses the one or more respiratory devices. As shown in FIG. 4E, when the user clicks on a sub-area displaying "respiratory treatment data" , a date or a length of time the first subject actually uses the respiratory device, and actual monitored data information such as Mean EPAP, Mean IPAP, EPAP95, IPAP95, AHI, Leak, Vt, or Res rate, MV may be displayed in the detailed data page corresponding to the sub-area.
[0239] In some embodiments, the cloud platform may display different types of abnormal statistical data through different sub-areas in the display interface, e.g., as shown in FIG. 4C, the count of persons with a High AHI, the count of persons with High Leak, and the count of persons with Low usages (usage time < 4h per day continues 10 days after 90 days) , etc., displayed in the display interface of the second terminal. In some embodiments, a detailed data page corresponding to the abnormal statistical data may be displayed by the user clicking the abnormal statistical data displayed in the different sub-areas in the display interface. For example, when the user clicks on a sub-area displaying “the count of persons with a High AHI” in the display screen, a first subject list corresponding to the High AHI may be generated and displayed. The first subject list may display information about all first subjects with the abnormal statistical data (i.e., High AHI) . For example, the information may include a number of the first subject, a name of the first subject, a gender of the first subject, a city of the first subject, a supervising doctor of the first subject, a last update time, a compliance proportion, etc. The compliance proportion refers to a proportion of a count of first subjects who use the respiratory device correctly to a count of first subjects with a certain piece of abnormal statistical data.
[0240] In some embodiments, the display content of each of the plurality of sub-areas may also include a proportion of data upload manner. The data upload manners may include Quick Response (QR) code scanning upload, wireless upload, etc. For example, the proportion of data upload manner may include 7.7%for QR code scanning transmission, 92.3%for wireless transmission, etc.
[0241] In some embodiments, the display content of each of the plurality of sub-areas may also include a first subject board. The first subject board refers to a board interface that shows a situation of different first subjects receiving the respiratory treatment. In some embodiments, the first subject board may include statistical information of a plurality of first subjects receiving the respiratory treatment. For example, the statistical information may include a name of the first subject, an ID of the first subject, a type of device used by the first subject, a duration and a date when the first subject receives the respiratory treatment, etc. In some embodiments, a detailed data page displaying more information (e.g., respiratory treatment data when the respiratory treatment is performed each time) about the first subject may be displayed by the user clicking on an area where each of the different first subjects is located in the first subject board.
[0242] In some embodiments, the display content of the plurality of sub-areas may also include a second subject board. The second subject board refers a board interface that displays information about the second subject and the first subject supervised by the second subject. In some embodiments, the second subject board may include a user name of the second subject, a name of the second subject, a gender of the second subject, a city of the second subject, an address of the second subject, a last login time of the second subject, a user type of the second subject, an e-mail address of the second subject, a count of first subjects supervised by the second subject, first subject information, etc.
[0243] In some embodiments, the display content of each of the plurality of sub-areas may also include a remote parameter regulation board and a report generation board.
[0244] The remote parameter regulation board refers to a board interface configured to remotely regulate a parameter of the respiratory device. The parameter may be viewed and the device parameter may be regulated through the remote parameter regulation board. The parameter viewing may include viewing the device parameter, a mask type, a tubing type, a historical device replacement record, etc.
[0245] The report generation board refers to a board interface configured for online generation, viewing, and downloading of a report. The report may include a statistical report (including a general situation of the first subject, a general situation of the device, statistical information such as a usage situation, a pressure, a respiratory index, air leakage, a blood oxygen, a pulse rate, etc. ) , a summary report (including the general situation of the first subject, the general situation of the device, a summary waveform, a usage period, AHI&AI, air leakage, a pressure, etc. ) , a detailed report (including the general situation of the first subject, the general situation of the device, a detailed waveform, a pressure, an airflow, a respiratory event, air leakage, a blood oxygen, a pulse rate, etc. ) , and all available reports (including all the foregoing reports, etc. ) .
[0246] In some embodiments, the display content of the plurality of sub-areas may also include a comprehensive score of the first subject. The comprehensive score may allow the user to quickly know a respiratory device usage condition of the first subject on a most recent day (from 12: 00 noon on a previous day to 12: 00 noon on a current day) .
[0247] In some embodiments, the cloud platform may determine the comprehensive score of the first subject based on the respiratory treatment data and the respiratory device usage data of the first subject. For example, the cloud platform may determine an indicator score of a key indicator in the respiratory treatment data and the respiratory device usage data and determine the comprehensive score by weighting indicator scores of a plurality of key indicators. The key indicators in the respiratory treatment data may include the oxygen saturation, the air leakage, the AHI, etc. The key indicators in the respiratory device usage data may include the usage duration etc. The key indicators and weights of the different key indicators may be preset by the system or manually. In some embodiments, the weights assigned to the plurality of key indicators may be different in a case where the oxygen saturation is included and where the oxygen saturation is not included in the key indicators. For example, in the case where the oxygen saturation is included in the key indicators, the weight corresponding to the indicator score of the oxygen saturation may be 0.1, the weight of the usage duration may be 0.6, the weight of the air leakage may be 0.2, and the weight of the AHI may be 0.1. In the case where the oxygen saturation is not included in the key indicators, the weight of the usage duration may be 0.6, the weight of the air leakage may be 0.2, and the weight of the AHI may be 0.2.
[0248] In some embodiments, the cloud platform may determine indicator scores of the different key indicators in the plurality of key indicators based on preset calculation rules. The calculation rules of the indicator scores of the different key indicators may be different. For example, the calculation rule of the indicator score of oxygen saturation may include: determining a calculated value of oxygen saturation based on a detection value of oxygen saturation of the first subject and a proportion of the usage duration (e.g., a proportion of an actual usage duration to a day) and determining, based on a correspondence between the calculated value of oxygen saturation and the indicator score, the indicator score corresponding to the oxygen saturation. The correspondence between the calculated value of oxygen saturation and the indicator score may be preset by the system or manually. The calculation rules of indicator scores of other key indicators may be similar to that of the indicator score of oxygen saturation, which may not be repeated herein. The content of the calculation rules is merely illustrative and is not intended to constitute a restriction on the implementation.
[0249] In some embodiments, different sub-areas may have different display parameters. In some embodiments, the display parameter may include at least one of a position, an appearance parameter, a reminder level, etc., of the sub-area. The position of the sub-area refers to a position (e.g., a center position, a secondary center position, a secondary edge position, or an edge position) where the sub-area is located in the display interface. The appearance parameter refers to information related to an appearance of the sub-area. For example, the appearance parameter may include dimensional information (e.g., the length, the width, or the acreage) , a display form of the content displayed on the sub-area (e.g., the content may be displayed in a numerical form, in an image form, or in a combined numerical-image form) , etc. The reminder level refers to a degree of differentiated display of the sub-area. For example, the higher the reminder level of a certain sub-area is, the higher the degree of differentiated display of the sub-area may be. Different reminder levels may be reflected through different interface design languages. For example, a relatively high reminder level may be identified through an operation such as box coloring or overstriking. The different interface design languages corresponding to the different reminder levels may be system presets, system defaults, or manual presets.
[0250] The different display parameters may be set for the different sub-areas, so that it is convenient for the user to quickly grasp the key information and reduce the professional requirements for the user to view the information.
[0251] In some embodiments, the cloud platform may determine importance levels of different abnormal statistical data based on the respiratory treatment data and the respiratory trend of the first subject and determine the display parameters of sub-areas corresponding to the different abnormal statistical data based on the importance levels.
[0252] The importance level of the abnormal statistical data refers to information that reflects an importance degree of the abnormal statistical data. For example, the higher the importance level of the abnormal statistical data is, the higher the importance degree of the abnormal statistical data may be.
[0253] In some embodiments, the cloud platform may determine the importance level of the abnormal statistical data based on the respiratory treatment data and the respiratory trend of the first subject through a preset rule. The preset rule refers to a rule preset in advance to determine the importance level of the abnormal statistical data. For example, the preset rule may include determining whether there is a type of data that needs to be focused on by the first subject in the abnormal statistical data and, in response to determining there is the type of data that needs to be focused on by the first subject in the abnormal statistical data, setting the type of data that needs to be focused on with the highest importance level. It is understood that for first subjects with different types of diseases, types of data needs to be focused on may be different. For example, for a first subject with a sleep disorder, the AHI may need to be focused on, and for a first subject with a lung disease, the tidal volume, the minute ventilation, etc., may need to be focused on. The type of data that needs to be focused on may be preset based on experience or needs. As another example, the preset rule may be that the higher the abnormality degree of the abnormal statistical data is, the higher the importance level corresponding to the abnormal statistical data may be.
[0254] In some embodiments, the cloud platform may preset a correspondence between different importance levels and different display sizes, and determine, based on the correspondence between different importance levels and different display sizes, the display sizes corresponding to the different abnormal statistical data. The display size refers to size information of a sub-area in which a certain content is displayed. For example, the higher the importance level of the abnormal statistical data is, the larger the size of the sub-area for display may be.
[0255] In some embodiments, the cloud platform may design a plurality of interface display layouts in advance. The interface display layout may include a distribution of positions of sub-areas of various sizes. Through a selected interface display layout and the previously determined display size, the different abnormal statistical data may be displayed on the sub-areas of the sizes and positions corresponding to the different abnormal statistical data. The distribution of positions of sub-areas of different sizes in the display interface may be determined by presetting the interface display layouts instead of randomly arranging the positions of sub-areas or arranging the positions of sub-areas in a diffuse manner according to a central position after the display sizes are determined, which can effectively avoid the error of correspondence between the positions and importance levels of the sub-areas.
[0256] In some embodiments, the cloud platform may preset a correspondence between the display form of the content and the size information of the sub-area. After the display size of the abnormal statistical data is determined, the display form of the abnormal statistical data may be determined according to the correspondence between the size information and the display form of the content.
[0257] In some embodiments, the cloud platform may determine the reminder level of the sub-area corresponding to the abnormal statistical data based on the abnormality degree and the importance level of the abnormal statistical data. For example, the cloud platform may determine a reminder level of a sub-area corresponding to current abnormal statistical data based on an abnormality degree and an importance level of the current abnormal statistical data through a level comparison table. The level comparison table may be determined based on historical data or priori knowledge. In some embodiments, the cloud platform may also determine the reminder level of the sub-area corresponding to the abnormal statistical data based on a dependence degree of the first subject on the respiratory device and the abnormality degree and the importance level of the abnormal statistical data of the first subject. For example, for two users using a respiratory device, one may be a first subject with a lung disease who has a high dependence degree on the respiratory device, and another may be an ordinary first subject who has a moderate dependence degree on the respiratory device. The required usage durations of the two users are the same, and the required usage durations may be 8 hours. However, for the former with a lung disease who has a high dependence degree on the respiratory device, the required usage duration may need to be strictly met due to the disease condition of the former, and the latter who has a moderate dependence degree on the respiratory device may merely use the respiratory device as a sleep aid. Therefore, for the former, the importance level of the usage duration may be relatively high. For the latter, the importance level of the usage duration may be relatively low. In this case, even though for the two users, the required usage durations (e.g., 6 hours) are not met, abnormal statistical data of the required usage durations may correspond to different importance levels and different reminder levels. The reminder level of the former may be higher than the reminder level of the latter.
[0258] In some embodiments, the cloud platform may also determine, based on importance levels of different abnormal statistical data, the display parameters of the sub-areas corresponding to the different abnormal statistical data through a parameter determination model. The parameter determination model refers to a model used to determine the display parameter of a sub-area. In some embodiments, the parameter determination model may be a machine learning model. For example, the parameter determination model may include a convolutional neural network (CNN) model, a neural network (NN) model, or other custom model structures, or the like, or any combination thereof.
[0259] In some embodiments of the present disclosure, the importance levels of the different abnormal statistical data may be determined based on the respiratory treatment data and the respiratory trend of the first subject, and the display parameters of the sub-areas corresponding to the different abnormal statistical data may be determined, which facilitates the prioritization of displaying the important data to the user, thereby avoiding the negative impact on the treatment result caused by ignoring the important data.
[0260] In some embodiments, the different sub-areas may have different refresh performance quotas. The refresh performance quota refers to an extent to which a refresh performance is allocated to a sub-area. The higher the refresh performance quota is, the higher the refresh rate allocated to the sub-area may be. The refresh rate refers to a software refresh rate of a software page where the sub-area is located, or a physical refresh rate of a screen area where the sub-area is located.
[0261] In some embodiments, the cloud platform may determine the refresh performance quotas allocated to the screen areas where the different sub-areas are located according to data generation frequencies and data update frequencies of different sub-areas. For example, the higher the data generation frequency and the data update frequency of a sub-area are, the higher the refresh performance quota allocated to the sub-area may be.
[0262] The data generation frequency refers to a frequency with which raw data of a treatment parameter corresponding to the sub-area is generated. The data update frequency refers to a frequency with which the treatment parameter corresponding to the sub-area is updated. The update refers to an update of data that is actually stored inside the storage device, not an update that appears on the page. For example, the data generation frequency of sub-area 403 may be equal to a frequency (an average interval at which the heartbeat is collected, i.e., a reciprocal of a heart rate value) at which a heartbeat is collected by a respiratory auxiliary device. The data update frequency of sub-area 403 may be equal to a frequency with which a heart rate is updated (e.g., every 1 minute or every 10 seconds) . As another example, the data generation frequency of sub-area 406 may be equal to a frequency with which the respiratory device make statistic on the usage data, and the data update frequency may be equal to a frequency with which the usage data is updated.
[0263] It is understood that when the user does not click on the detailed data page corresponding to a certain sub-area, the user is in a page where a plurality of sub-areas and the treatment parameters thereof are displayed together, and the data generation frequencies and the data update frequencies of different sub-areas may be usually different. If all sub-areas are refreshed at a same refresh rate, the refresh performance may be wasted (e.g., display content may not change when some sub-areas with a relatively low data update frequency are refreshed repeatedly and continuously) or the refresh may not be timely (e.g., if internal actual data of some sub-areas with a high data update frequency has been updated, but it is not yet time for a next refresh, there may be a lag in the data displayed on the page) . The data generation frequencies and the data update frequencies of different sub-areas may be considered, which may effectively balance the refresh performance and refresh timeliness, thereby improving the user experience.
[0264] In some embodiments, the display content of each of the plurality of sub-areas may further include an abnormal data graph.
[0265] The abnormal data graph refers to a graph used to reflect an association relationship between the abnormal statistical data, a reason causing an abnormality, and a treatment measure. The abnormal data graph may be a data structure consisting of nodes and edges. The edges may connect the nodes, and the nodes and the edges may have attributes. In some embodiments, the cloud platform may construct the abnormal data graph based on the abnormal statistical data, the reason causing the abnormality, and the treatment measure.
[0266] The reason causing the abnormality refers to a reason causing an abnormality of the treatment parameter. For example, a reason causing an abnormality may include a malfunction of the respiratory device, an improper diet of the first subject, an insufficient usage duration of the first subject, etc. The treatment measure refers to a measure to deal with the reason causing the abnormality of the abnormal statistical data. For example, the treatment measure may include repairing the respiratory devices, adjusting the diet of the first subject, increasing the usage duration of the first subject, etc. In some embodiments, the cloud platform may determine the reason causing the abnormality and the treatment measure based on the abnormal statistical data.
[0267] The abnormal data graph may include a plurality of types of nodes and a plurality types of edges.
[0268] In some embodiments, the abnormal data graph may include a node of abnormal statistical data (also referred to as an abnormal statistical data node) , a node of a reason causing an abnormality (also referred to as a reason node) , and a node of a treatment measure (also referred to as a treatment measure node) . In some embodiments, a detailed data page corresponding to a node may be jumped to by the user clicking on the node. For example, by the user clicking on the abnormal statistical data node, the detailed data page corresponding to the abnormal statistical data node may be jumped to, and the user may view the data type, the abnormality degree, the importance level, etc., of the abnormal statistical data.
[0269] In some embodiments, a node attribute of each type of nodes may include a node size and a node color. The node size may have a maximum size threshold (i.e., an upper threshold of the node size) , and the maximum size threshold may be preset based on experience or needs.
[0270] In some embodiments, the node size of the abnormal statistical data node may be determined based on the abnormality degree. The higher the abnormality degree is, the larger the node size of the abnormal statistical data node may be. A specific correspondence between the abnormality degree and the node size may be preset based on experience or needs.
[0271] In some embodiments, the node size of the reason causing the abnormality node may be determined based on a confidence level of the reason causing the abnormality. The higher the confidence level is, the larger the node size of the reason causing the abnormality node may be. Merely by way of example, the cloud platform may set a node size corresponding to a confidence level of 100%as a unit node size. When the confidence level of the reason causing the abnormality is 50%, the node size of the node corresponding to the reason causing the abnormality node may be half of the node size of the unit node size. Similarly, node sizes corresponding to other confidence levels may be determined. The unit node size may be preset by the system or manually.
[0272] In some embodiments, different abnormal statistical data may correspond to the same reason causing an abnormality. The node size of the reason causing the abnormality node may be determined based on a sum of confidence levels of the same reasons causing the abnormality corresponding to the different abnormal statistical data. The sum of the confidence levels may be greater than 100%, and accordingly, the node size of the reason causing the abnormality node may be greater than the unit node size.
[0273] In some embodiments, the cloud platform may determine a probability of occurrence of a reason causing the abnormality corresponding to the same or similar basic data of the first subject and the respiratory treatment data of the first subject as the confidence level of the reason causing the abnormality.
[0274] In some embodiments, the node size of the treatment measure node may be determined based on a feasibility degree of the treatment measure. The higher the feasibility degree is, the larger the node size of the treatment measure node may be. The node size of the treatment measure node may be determined based on the feasibility degree of the treatment measure in a similar way of determining the node size of the reason causing the abnormality node based on the confidence level of the reason causing the abnormality. More descriptions may be found in the relevant descriptions above.
[0275] The feasibility degree refers to a parameter reflecting a degree of the treatment measure being feasible. In some embodiments, the feasibility degree may be determined based on an operation difficulty of the user and a recommendation degree.
[0276] The operation difficulty refers to a parameter that measures how easy or difficult the operation is for the user to operate. For example, for a treatment measure (e.g., adjusting a setting parameter of a respiratory device, etc. ) that is inconvenient for the user to operate, the operation difficulty of the treatment measure may be set to be relatively high. The recommendation degree refers to a parameter that measures a degree of the cloud platform recommending the treatment measure. In some embodiments, the cloud platform may determine a probability of occurrence of a treatment measure corresponding to the same or similar basic data of the first subject, the respiratory treatment data of the first subject, and the same abnormality as the recommendation degree of the treatment measure.
[0277] In some embodiments, the cloud platform may determine the feasibility degree of the treatment measure through weighted operation based on the operation difficulty and the recommendation degree. Weights of the operation difficulty and the recommendation degree may be preset by the system or manually.
[0278] In some embodiments, the abnormal statistical data node, the reason causing the abnormality node, and the treatment measure node may be set to different color schemes and / or different textures (or shapes) .
[0279] The cloud platform may determine the colors and / or textures (or shapes) of different types of nodes in various ways.
[0280] In some embodiments, the cloud platform may determine a user classification based on age and vision of the first subject in the basic data of the first subject through a classification model and select a color combination and / or texture combination (or shape combination) corresponding to the user classification based on the user classification. Exemplary classification models may include a K-nearest neighbor model, a decision tree model, etc.
[0281] For example, the cloud platform may classify users into a group with a visual impairment (e.g., color weakness or color blindness) and a group with a normal vision through the classification model. For the group with the visual impairment, the cloud platform may select a preset texture combination and randomly assign one texture in the texture combination to each of the different types of nodes. For the group with the normal vision, the cloud platform may select a preset color combination and randomly assign one color in the color combination to each of the different types of nodes. The preset texture combination and the preset color combination may be determined based on an input of the user or based on a user preference. The user preference may be determined based on the user inputting or a color or texture corresponding to a historical highest usage frequency of the user.
[0282] In some embodiments, the cloud platform may determine an associated first subject of the current first subject and determine a target color combination based on historical viewing records of the abnormal data graph of the associated first subject under different color combinations.
[0283] The target color combination refers to a node color combination of the abnormal data graph displayed to the current first subject. In some embodiments, the associated first subject may be other first subjects having the same or similar basic data as the current first subject. In some embodiments, the historical viewing records may include a count of clicks on each node by the associated first subject, a length of time the detailed data page of the node is viewed, a count of times the treatment measure is adopted, etc.
[0284] In some embodiments, for each associated first subject, the cloud platform may perform a normalization processing on the count of clicks on each node in the abnormal data graph, the length of time the detailed data page of the node is viewed, and the count of times the treatment measure is adopted based on the associated first subject under different node color combinations; determine combination scores of the different node color combinations by weighting the count of clicks on each node, the length of time the detailed data page of the node is viewed, and the count of times the treatment measure is adopted after the normalization processing; and determine a node color combination with a highest combination score as the target color combination among a plurality of node color combinations of a plurality of associated first subjects. The normalization processing may normalize the count of clicks on each node, the length of time the detailed data page of the node is viewed, and the count of times the treatment measure is adopted to a preset value range (e.g., [0, 1] ) . In the embodiments of the present disclosure, there may be no special limitation on the normalization processing, and the normalization processing may be performed using the operation known to those skilled in the art. For example, exemplary normalization treatments may include linear normalization, Z-Score normalization, fractional calibration normalization, etc., or any combination thereof. The weights corresponding to the count of clicks on each node, the length of time the detailed data page of the node is viewed, and the count of times the treatment measure is adopted may be preset by the system or manually.
[0285] In some embodiments, the cloud platform may determine the count of times the treatment measure is adopted based on a count of operations of the user in the detailed data page corresponding to the treatment measure node.
[0286] In some embodiments, the cloud platform may determine the count of times the treatment measure is adopted based on a change in the parameter of the respiratory device. For example, when the treatment measure is pressurization, the cloud platform may determine the count of times the treatment measure is adopted based on a count of times pressure data changes.
[0287] In some embodiments, color shades of the node color of a same type of nodes may be different. In some embodiments, the cloud platform may determine the color shades of the abnormal statistical data node, the color shades of the reason causing the abnormality node, and the color shades of the treatment measure node based on the abnormality degree of the abnormal statistical data, the confidence level of the reason causing the abnormality, and the feasibility degree of the treatment measure, respectively. For example, the higher the abnormality degree of the abnormal statistical data is, the higher the confidence level of the reason causing the abnormality is, and the higher the feasibility degree of the treatment measure is, the darker the color of the abnormal statistical data node may be. A specific correspondence between a color shade range and the abnormality degree of the abnormal statistical data, the confidence level of the reason causing the abnormality, and the feasibility degree of the treatment measure may be preset based on experience or needs.
[0288] In some embodiments, an edge may exist between the abnormal statistical data node and the one or more reason nodes corresponding to the abnormal statistical data node, and an edge may exist between the reason node and one or more treatment measure nodes corresponding to the reason node. In some embodiments, an edge attribute may include a color of the edge, a thickness of the edge, a length of the edge, etc.
[0289] In order to facilitate the user to intuitively and quickly select an appropriate abnormality cause and treatment measure according to the abnormal data graph, the cloud platform may distinguish a most recommended reason causing the abnormality and a most recommended treatment measure through different edge attributes. For example, the cloud platform may determine a reason node with a highest confidence level among the plurality of reason nodes connected to the abnormal statistical data node and a treatment measure node with a highest feasibility degree among the plurality of treatment measure nodes connected to the reason node, and bold or color an edge between the abnormal statistical data node and the reason node with the highest confidence level and an edge between the reason node and the treatment measure node with the highest feasibility degree to differentially display the most recommended reason causing the abnormality and treatment measure of the cloud platform. As another example, the cloud platform may determine the length of the edge based on the confidence level of the reason causing the abnormality and the feasibility degree of the treatment measure. The greater the confidence level of the reason causing the abnormality is, and the greater the feasibility degree of the treatment measure is, the shorter the edge corresponding to the reason causing the abnormality node and the treatment measure node, and the closer the relationship between the nodes corresponding to the edge on the abnormal data graph may be. In some embodiments, the largest threshold and the smallest threshold of the edge may be set based on experience or needs, which can avoid overlapping of nodes caused by the edges being too short, a too large graph caused by the edges being too long, etc.
[0290] In some embodiments, the cloud platform may construct the abnormal data graph by connecting nodes of different sizes and edges of different lengths through a force-directed algorithm or any other optional means.
[0291] Merely by way of example, FIG. 4F shows an exemplary schematic diagram of an abnormal data graph 408. Different types of nodes may correspond to different textures (or colors) , and different nodes may have different node sizes. An edge 4009 exists between an abnormal statistical data node 4001 and a node 4004, an edge 4010 exists between an abnormal statistical data node 4001 and a reason node 4005, and the edge 4010 is thicker than the edge 4009, which may indicate that the abnormal statistical data 1 may be caused by the reason causing the abnormality 1 and the reason causing the abnormality 2, and the confidence level of the reason causing the abnormality 2 is higher than the confidence level of the reason causing the abnormality 1. The abnormal statistical data nodes 4001, 4002, and 4003 are connected to the reason node 4005, which may indicate that the abnormal statistical data 1, 2 and 3 may be caused by the reason causing the abnormality 2. The reason causing the abnormality node 4005 is connected to the treatment measure nodes 4007 and 4008, and the edge 4011 is thicker than the edge 4012, which may indicate that: the reason causing the abnormality 2 may be treated by the treatment measure 1 and the treatment measure n, and the feasibility degree of the treatment measure 1 is higher than the feasibility degree of the treatment measure n. The abnormal statistical data node 4003 is connected to the reason causing the abnormality nodes 4005 and 4006, which may indicate that the abnormal statistical data 3 may be caused by the reason causing the abnormality 2 and the reason causing the abnormality n. The rest of the diagram is similarly understood, which is not repeated herein.
[0292] In some embodiments, the display content of each of the plurality of sub-areas may further include a data comparison diagram.
[0293] The data comparison diagram refers to a diagram that includes a comparison relationship between two or more types of data. In some embodiments, a data comparison diagram may include a comparison relationship between one or more of actual data, standard data, predicted data, etc.
[0294] The actual data refers to indication data actually detected by the user, for example, respiratory treatment data actually detected by the user. The cloud platform may directly obtain the actual data based on the respiratory devices, the first monitoring device, etc.
[0295] The standard data refers to indication data in a normal range, for example, a normal range of the AHI may be 0-5. The cloud platform may obtain the standard data based on user input information, web crawling, etc.
[0296] The predicted data refers to predicted indication data, for example, a predicted AHI, an oxygen saturation, etc., of the user in a future period of time. It is understood that the predicted data is merely for predictable indication data (e.g., the AHI or blood oxygen concentration) and not for unpredictable indication data (e.g., the usage duration of the respiratory device) of the user.
[0297] In some embodiments, the cloud platform may obtain the predicted data based on an effect prediction model.
[0298] The effect prediction model refers to a model used to determine the predicted data. In some embodiments, the effect prediction model may be a machine learning model. For example, the effect prediction model may include a CNN model, an NN model, other custom model structures, or the like, or any combination thereof.
[0299] In some embodiments, an input of the effect prediction model may include the basic data and the respiratory treatment data of the first subject, and an output of the effect prediction model may include the predicted data.
[0300] In some embodiments, the effect prediction model may be obtained based on a plurality of training samples with labels. For example, the training of the effect prediction model may be performed based on a gradient descent manner. Merely by way of example, the plurality of training samples with labels may be input into an initial effect prediction model, a loss function may be constructed through the labels and results of the initial effect prediction model, parameters of the initial effect prediction model may be iteratively updated based on the loss function, the training of the effect prediction model may be completed when a preset iteration condition is satisfied, and the effect prediction model may be obtained. The preset iteration condition may be that the loss function converges, a count of iterations reaches a threshold, etc.
[0301] In some embodiments, a training sample may include sample basic data and sample respiratory treatment data of a sample first subject, and the first label may include actual data corresponding to the training sample. The training sample and the first label may be obtained based on historical data.
[0302] In some embodiments, the data comparison diagram may reflect a comparison relationship between the actual data and the standard data and / or the predicted data.
[0303] In some embodiments, the data comparison diagram may be in the form of a chart such as a table, a line chart, a bar chart, or a pie chart, etc. Understandably, different first subjects may have different levels of understanding of different forms of the chart. In some embodiments, the cloud platform may determine the form of the chart of the data comparison diagram based on the basic data of the first subject.
[0304] In some embodiments, the cloud platform may determine the form of the chart of the data comparison diagram based on basic data of a large number of first subjects and readability scores of the first subjects to different forms of the chart.
[0305] The readability score refers to a degree of the first subject understanding a particular form of the chart. The larger the readability score of a form of a chart is, the better the first subject understands the form of the chart may be.
[0306] In some embodiments, the cloud platform may determine the readability scores of different forms of the chart in various ways. For example, the cloud platform may determine the readability scores by processing user behavior data based on a scoring model. The user behavioral data refers to relevant data reflecting a behavior of the user, e.g., durations of the user views the different forms of the chart, counts of clicks on the different forms of the chart, etc. The scoring model may be a machine learning model, for example, the scoring model may be a NN model.
[0307] In some embodiments, different data in the data comparison diagram may be differentially displayed using different colors, different font sizes, different line thicknesses, etc. For example, the actual data and the standard data may be differentially displayed using different colors.
[0308] In some embodiments, the cloud platform may construct the data comparison diagram based on at least two of the actual data, the standard data, and the predicted data in various feasible ways (e.g., Echarts, Excel, etc. ) .
[0309] In some embodiments of the present disclosure, the data comparison diagram may be constructed based on the actual data, the standard data, and the predicted data, and the data comparison diagram may be displayed through the sub-area, which can make the first subject more accurately perceive the treatment condition and can be convenient for the first subject to conduct self-examination. For example, when the first subject finds that the actual data is significantly different from the predicted data, the first subject may take the initiative to take the measure or seek help from the second subject. On the other hand, the second terminal may also see a difference between the actual data and the predicted data during the treatment period, which is convenient for the second subject not only to accurately grasp the situation of the first subject, but also to test and adjust the treatment plan in time.
[0310] In some embodiments, the cloud platform may generate the display interface based on the user basic data, the basic data of the respiratory devices, the usage data of the respiratory devices, and / or the respiratory treatment data through a preset algorithm, a machine learning model, etc.
[0311] In some embodiments of the present disclosure, the display interface may be generated based on the user basic data, the basic data of the respiratory devices, the usage data of the respiratory devices, and / or the respiratory treatment data, so that various physiological indications of the user and the treatment data based on the respiratory devices may be displayed to the user intuitively and graphically, thereby facilitating the smooth progress of the treatment process and the timely adjustment of the treatment plan.
[0312] In some embodiments, the target information may include a target report. In some embodiments, the cloud platform may, in response to receiving a report generation instruction from the user terminal, determine a report benchmark and generate the target report meeting the report benchmark. More descriptions regarding generating the target report may be found in FIG. 8 and relevant descriptions thereof.
[0313] In some embodiments, the target information may include reminder information. In some embodiments, the cloud platform may generate a first result based on a relationship between the respiratory treatment data and a first index and generate the reminder information based on the first result. More descriptions regarding generating the reminder information may be found in FIG. 9 and relevant descriptions thereof.
[0314] In some embodiments, the target information may include a recommended setting parameter. In some embodiments, the cloud platform may generate the recommended setting parameter of the respiratory devices based on the user basic data, the basic data of the respiratory devices, and historical respiratory treatment data of the first subject. More descriptions regarding generating the recommended setting parameter may be found in FIG. 10 and relevant descriptions thereof.
[0315] In some embodiments, the target information may include corrected respiratory treatment data. In some embodiments, the cloud platform may determine a quality of the respiratory treatment data, and in response to a determination that the quality does not meet a quality condition, correct the respiratory treatment data based on the historical respiratory treatment data of the first subject to generate the corrected respiratory treatment data. More descriptions regarding generating the corrected respiratory treatment data may be found in FIG. 12 and relevant descriptions thereof.
[0316] In some embodiments, the target information may include a target answer of a target question. In some embodiments, the cloud platform may obtain a session initiation instruction based on the first terminal, predict and display one or more suspected questions based on the respiratory treatment data and operation behavior data of the user on the display interface, in response to an operation instruction of the user for the suspected questions determine a target question, and determine the target answer and feedback the target answer to the user through the first terminal. More descriptions regarding generating the target answer may be found in FIG. 14 and relevant descriptions thereof.
[0317] In some embodiments, the cloud platform may send the target information to the user terminal through wired transmission, wireless transmission, etc.
[0318] In some embodiments, the cloud platform may determine a transmission bandwidth allocated to a target information sending task according to an amount of data contained in target information. The target information sending task refers to a task of sending the target information to the user terminal through the wired transmission, the wireless transmission, etc. For example, the larger the amount of data contained in the target information is, the larger the transmission bandwidth allocated to the target information sending task may be.
[0319] In some embodiments of the present disclosure, the user basic data, the basic data of the respiratory devices, the usage data of the respiratory devices, and / or the respiratory treatment data of the first subject may be obtained based on the user terminal, and the target information may be generated and sent to the user terminal, so that various data related to the user and the association relationship between the data may be comprehensively considered. Intelligent analysis may be performed, and the target information such as the display interface, the target report, the reminder information, the recommended setting parameter, the corrected respiratory treatment data, the target answer, etc., can be efficiently and accurately obtained. Health of the first subject can be comprehensively managed in an all-round and multi-dimensional manner, which facilitates promoting a treatment process of the first subject and enhancing the treatment experience of the user.
[0320] In some embodiments, in response to receiving a preheating request (which may be issued by the user terminal) , the cloud platform may generate a preheating instruction, and send the preheating instruction to a humidifier in the respiratory device to control the humidifier for preheating. The preheating instruction may include a parameter such as a target preheating temperature (i.e., a temperature reached after preheating) , a duration to reach the target preheating temperature for the first time, or a maximum preheating duration. The parameter in the preheating instruction may be a default value or may be set by the user (e.g., the first subject or second subject) before or during each usage.
[0321] In some embodiments, the cloud platform may determine a default value of the duration to reach the target preheating temperature for the first time in the preheating instruction based on an average of actual preheating durations when the first subject historically uses the preheating function. For example, the average of the actual preheating durations when the first subject historically uses the preheating function may be determined as the default value of the duration to reach the target preheating temperature for the first time. It is understandable that, for example, if an initial value of the maximum preheating duration is 20 minutes, an initial value of the duration to reach the target preheating temperature for the first time is 5 minutes, and the first subject does not start using the respiratory device until the respiratory device is preheated for 10 minutes in the actual usage process of the preheating function, the default value of the duration to reach the target preheating temperature for the first time in the preheating instruction may be set to be 10 minutes to reduce resource waste.
[0322] In some embodiments, the cloud platform may determine the default value of the target preheating temperature based on the treatment plan of the first subject. For example, a heating temperature of the humidifier in the respiratory device used by the first subject in the usage parameter of the respiratory device in the treatment plan of the first subject may be determined as the target preheating temperature.
[0323] In some embodiments, the user terminal may provide the first subject with an interactive button to turn on the preheating function. In response to a state of a humidifier gear being "off, " an icon and an entry of the interaction button may be non-optional. In response to the state of the humidifier gear being "on, " the icon and the entry of the interaction button may be normal and optional.
[0324] In some embodiments, the cloud platform may determine whether a preheating stop condition is satisfied. In response to a determination that the preheating stop condition is satisfied, the cloud platform may generate a preheating stop instruction, and send the preheating stop instruction to the humidifier in the respiratory device to control the humidifier to stop preheating.
[0325] The preheating stop condition refers to a condition used to determine whether to stop the preheating. In some embodiments, the preheating stop condition may include that a temperature of the humidifier reaches a temperature threshold, the user clicks an interactive button of "stop preheating, " and the duration of preheating reaching a duration threshold, or the like, or any combination thereof. The temperature threshold and the duration threshold may be preset by the system or manually.
[0326] In some embodiments of the present disclosure, the preheating of the humidifier may effectively improve the experience of the first subject when using the respiratory device.
[0327] FIG. 3 is a schematic diagram illustrating an exemplary process of determining a respiratory trend according to some embodiments of the present disclosure.
[0328] As shown in FIG. 3, in some embodiments, a cloud platform may predict, based on basic data 310 of a first subject, respiratory treatment data 320, and physiological data 330 of the first subject, the respiratory trend 350 of the first subject through a first prediction model 340.
[0329] The first prediction model refers to a model used to determine the respiratory trend. In some embodiments, the first prediction model may be a machine learning model, e.g., an NN model.
[0330] In some embodiments, an input of the first prediction model may include the basic data of the first subject, the respiratory treatment data, and the physiological data of the first subject. More descriptions regarding the basic data of the first subject, the respiratory treatment data, the physiological data, and the respiratory trend of the first subject may be found in FIG. 2 and relevant descriptions thereof.
[0331] In some embodiments, the first prediction model may include a plurality of branches. An input of each branch may include the basic data of the first subject, the respiratory treatment data, and the physiological data of the first subject, and an output of each branch may include a plurality of underlying diseases and development possibilities of the plurality of underlying diseases.
[0332] The underlying disease refers to a disease that the first subject may have. In some embodiments, the underlying disease may include a complication of a disease that the first subject currently suffers from, a treatment side effect and a treatment sequela after treatment of the disease that the first subject currently suffers from, an undiagnosed new disease, etc.
[0333] The development possibility of an underlying disease refers to a possibility that the first subject may suffer from the underlying disease next, and the possibility may be usually expressed as a probability, i.e., the development possibility of an underlying disease refers to a probability that the first subject may suffer from the underlying disease next.
[0334] In some embodiments, the branches of the first prediction model may have same inputs and different outputs. In some embodiments, the output of each branch of the first prediction model may include the plurality of underlying diseases and the development possibilities of the plurality of underlying diseases. For example, the output of each branch may be represented by a vector consisting of the plurality of underlying diseases and the development possibilities of the underlying diseases. For example, the output of a particular branch of the first prediction model may be [(underlying disease 1, development possibility 1) , (underlying disease 2, development possibility 2) , ..., (underlying disease n, development possibility n) ] .
[0335] In some embodiments, the cloud platform may determine a final underlying disease and a development possibility of the final underlying disease as the respiratory trend of the first subject by weighting according to the plurality of underlying diseases and the development possibilities of the plurality of underlying diseases output by the each branch of the first prediction model. For example, if there are 5 branches in the first prediction model, 5 underlying diseases A and development possibilities of the 5 underlying diseases A may be output accordingly, and the development possibilities of the underlying diseases A in the 5 branches may be e1~e5, respectively. The cloud platform may determine a mean value f as (e1+... +e5) / 5 and determine the final development possibility g of the underlying diseases A as h1*e1+... + h5*e5, where h1-h5 denote preset coefficients, and the cloud platform may determine the preset coefficients based on the output result of each branch and the mean value f. For example, the cloud platform may determine h1 as k / (f-e1) , ..., and h5 as k / (f-e5) , k denotes a preset threshold, and k is proportional to a difference between the output result of the each branch and the mean value f. The larger the difference between the output result of the each branch and the mean value f is, the larger the k may be.
[0336] In some embodiments of the present disclosure, the respiratory trend of the first subject may be predicted through the first prediction model, so that a plurality of factors affecting the respiratory trend and an association relationship between the plurality of factors may be considered comprehensively, and the respiratory trend may be determined efficiently and accurately. The first prediction model may include the plurality of branches, the plurality of underlying diseases and the development possibilities of the plurality of underlying diseases may be output using the plurality of branches, and the outputs of the branches may be fused to obtain the final output. Compared to an ordinary machine learning model where only one result is output, in the multi-branch prediction model, the respiratory trend of the first subject may be determined by outputting the plurality of results simultaneously through the plurality of branches and obtaining the final output result of the model by weighting the output results of branches and determining a mean value, which can prevent overfitting of a single branch and make the model output more accurate. When the outputs of branches are weighted and summed, weights may be set to be related to the differences between the output results of branches and the mean value, which may make the output of a branch that deviates from the mean value have a smaller weight, thereby ensuring that the overall prediction is accurate. The weights may be set to be related to the preset threshold, which may make a magnitude of the weights affected by the "difference between the output result of each branch and the mean value" related to the preset threshold, so that the larger the preset threshold is, the smaller the magnitude of the weights affected by the "difference between the output result of each branch and the mean value" may be. Since it is considered that the uncertainty of the disease trend of the first subject is inherently greater, even an underlying disease and a development possibility that deviates far from the mean value may need to be considered as normal values.
[0337] In some embodiments, when training the first prediction model, the cloud platform may train each branch using different sampling training datasets.
[0338] In some embodiments, the cloud platform may divide a plurality of sampling training datasets from a total training dataset for training different branches in the first prediction model. In some embodiments, the cloud platform may determine the sampling training dataset in the manner described in four operations below.
[0339] In S11, a first disease distribution of the training dataset may be determined, and a preset similarity threshold may be determined based on the first disease distribution.
[0340] The training dataset refers to a set that includes at least one piece of training data. The training data may include a training sample and a second label corresponding to the training sample. The training sample may include sample basic data of a sample first subject, sample respiratory treatment data, and sample physiological data of the sample first subject, and the label may include an actual disease of the sample first subject. The training data may be obtained based on historical data.
[0341] The disease distribution refers to a parameter that reflects the distribution of the conditions of different sample first subjects in the training sample. The first disease distribution refers to a disease distribution corresponding to the training dataset.
[0342] In some embodiments, the cloud platform may convert each training sample in the training dataset into a vector form (e.g., splicing the basic data of the sample first subject, the sample respiratory treatment data, and the sample physiological data, and constructing a corresponding feature vector) , cluster based on a feature vector corresponding to each training sample, and obtain the first disease distribution based on a clustering result. The clustering result may include a clustering center of each cluster and a count of feature vectors in each cluster. An exemplary first disease distribution may be:
[0343] It should be noted that in the embodiments of the present disclosure, there is no special limitation on the clustering manner, and clustering may be performed using an operation (e.g., a density-based spatial clustering of applications with noise (Dbscan) algorithm) known to those skilled in the art.
[0344] The preset similarity threshold refers to a preset value characterizing a uniformity degree of the disease distribution. The more uniform the disease distribution is, the smaller the preset similarity threshold may be.
[0345] In some embodiments, the cloud platform may determine a uniformity degree of the first disease distribution based on the count of feature vectors in each cluster in the first disease distribution, and determine, based on the uniformity degree of the first disease distribution, the preset similarity threshold. In some embodiments, the cloud platform may determine a variance of the count of feature vectors in each cluster in the first disease distribution. The smaller the variance is, the more uniform the count of feature vectors in each cluster in the first disease distribution and the smaller the preset similarity threshold may be. In some embodiments, the cloud platform may preset a correspondence between the variance of the count of feature vectors in each cluster in the first disease distribution and the preset similarity threshold, for example, a reciprocal of the variance may be determined as the preset similarity threshold.
[0346] In S12, an intermediate dataset may be formed by randomly extracting training data within a preset count range from the training dataset
[0347] The intermediate dataset may be used as a sampling training dataset. The preset count range may be a system default, an empirical value, a manually preset value, or the like, or any combination thereof, which may be set according to actual needs and not limited in the present disclosure.
[0348] In S13, a second disease distribution of the intermediate dataset may be determined. Whether the intermediate dataset may be used as a sampling training dataset may be determined based on the second disease distribution and the first disease distribution. In response to determining a difference between the second disease distribution and the first disease distribution satisfies a preset difference condition, the intermediate dataset may be determined as the sampling training dataset. In response to determining that the difference between the second disease distribution and the first disease distribution does not satisfy the preset difference condition, the intermediate dataset may be discarded.
[0349] The second disease distribution refers to a disease distribution corresponding to the intermediate dataset. The second disease distribution may be determined in a similar way of calculating the first disease distribution. More descriptions may be found in the relevant descriptions above.
[0350] The preset difference condition refers to a preset condition used to screen the sampling training dataset. In some embodiments, the preset difference condition may be that the difference between the second disease distribution and the first disease distribution is smaller than or equal to the preset similarity threshold.
[0351] In some embodiments, the cloud platform may determine a mean value, a median value, and a variance of the count of feature vectors in each cluster and a count of clusters in the first disease distribution and the second disease distribution, respectively, determine a first difference between the mean value of the count of feature vectors in each cluster in the first disease distribution and the mean value of the count of feature vectors in each cluster in the second disease distribution, a second difference between the median value of the count of feature vectors in each cluster in the first disease distribution and the median value of the count of feature vectors in each cluster in the second disease distribution, a third difference between the variance of the count of feature vectors in each cluster in the first disease distribution and the variance of the count of feature vectors in each cluster in the second disease distribution, and a fourth difference between the count of clusters in the first disease distribution and the count of clusters in the second disease distribution, respectively, and determining the difference between the second disease distribution and the first disease distribution by weighting the first difference, the second difference, the third difference, and the fourth difference. For example, for the first disease distribution A and the second disease distribution B, the mean value a1, median value a2, variance a3, and count of clusters a4 of the first disease distribution A may be determined, the mean value b1, median value b2, variance b3, and count of clusters b4 of the second disease distribution B may be determined. The first difference between a1 and b1, the second difference between a2 and b2, the third difference between a3 and b3, and the fourth difference between a4 and b4 may be determined, respectively, and the difference between the first disease distribution A and the second disease distribution B may be determined by weighting the four differences. The weights of the first difference, the second difference, the third difference, and the fourth difference may be a system default value, an empirical value, a manually preset value, or the like, or any combination thereof, which may be set according to actual needs and not limited in the present disclosure.
[0352] In S14, S12 and S13 may be repeated until a specific count of sampling training datasets are determined as the training data for each branch, respectively. The specific count of sampling training datasets may be a system default, an empirical value, a manually preset value, or the like, or any combination thereof, which may be set according to actual needs and be limited in the present disclosure.
[0353] In some embodiments, the cloud platform may train the each branch in the first prediction model using different sampling training datasets. An exemplary training process may include inputting a plurality of training samples with labels in one of the sampling training datasets into an initial branch, constructing a loss function based on the labels and results of the initial branch, iteratively updating parameters of the initial branch based on the loss function, and completing the branch training when the loss function of the initial branch satisfies a preset iteration condition. The preset iteration condition may be that the loss function converges, a count of iterations reaches a threshold, etc. The initial branch may be a model branch whose training is not completed.
[0354] In some embodiments of the present disclosure, instead of using an original training dataset directly as the training dataset of each branch, the plurality of sampling training datasets may be determined from the original training dataset, and each branch may be trained based on different sampling datasets, which may reduce the complexity of the data and prevent the first prediction model from failing to learn the "general rules" in the dataset due to insufficient learning ability caused by the relatively high data complexity of the original training dataset, resulting in underfitting. Each branch may be trained through the sampling training dataset, which may make trained parameters of different branches different and each branch have a certain degree of specialization. When whether the intermediate dataset is used as the sampling training dataset is determined, the intermediate dataset where the difference between the second disease distribution and the first disease distribution is greater than the preset similarity threshold may be discarded, which can effectively prevent the difference between the data distribution of the sampling training dataset and the data distribution of the original dataset from being too large.
[0355] FIG. 5 is a flowchart illustrating an exemplary process for determining a recommended treatment plan according to some embodiments of the present disclosure. In some embodiments, the process 500 may be performed by a cloud platform. As shown in FIG. 5, the process 500 may include the following operations.
[0356] In 510, one or more first candidate treatment plans may be determined.
[0357] The treatment plan refers to a program for treating a first subject by performing a specific parameter setting on a respiratory device and / or using the respiratory device in a specific manner (e.g., used at a specific usage frequency or for a specific usage duration) . For example, the treatment plan may include a parameter (e.g., a setting parameter of the respiratory device) related to the device for treating the first subject using the respiratory device, a usage phase, a usage duration (e.g., a usage duration of the respiratory device corresponding to each treatment phase) , a usage point in time (e.g., a starting time of a treatment phase) , etc.
[0358] A first candidate treatment plan refers to a preliminarily determined treatment plan. A first candidate treatment plan may be determined by a second subject based on personal experience and / or historical data, then input into a second terminal, and uploaded to the cloud platform through the second terminal.
[0359] In 520, a predicted treatment result corresponding to each of the one or more first candidate treatment plans may be determined through a second prediction model based on a respiratory trend of the first subject and the first candidate treatment plan.
[0360] The predicted treatment result corresponding to a first candidate treatment plan refers to a probability that each physical indicator of the first subject reaches a different value after treating the first subject according to the first candidate treatment plan. In some embodiments, the predicted treatment result may be expressed according to a predicted score (e.g., a value in a range of 0-1) of the each physical indicator of the first subject and / or a predicted score of recovery of the first subject.
[0361] The second prediction model refers to a model used to determine the predicted treatment result corresponding to a first candidate treatment plan. The second prediction model may be a machine learning model. For example, the second prediction model may include a neural network (e.g., a deep neural network (DNN) ) model or other feasible structures.
[0362] In some embodiments, an input of the second prediction model may include the respiratory trend of the first subject and a first candidate treatment plan, and an output of the second prediction model may be the predicted treatment result corresponding to the first candidate treatment plan. More descriptions regarding the respiratory trend of the first subject may be found in FIGs. 2 and 3 and the relevant descriptions thereof.
[0363] In some embodiments, the input of the second prediction model may also include a treatment cooperation degree of the first subject and / or one or more life features of the first subject.
[0364] The treatment cooperation degree of the first subject refers to a degree of the first subject cooperating with the treatment, e.g., the degree to which the first subject cooperates with the treatment plan. More descriptions regarding the treatment cooperation degree may be found in FIG. 2 and the relevant descriptions thereof.
[0365] In some embodiments, the cloud platform may determine the treatment cooperation degree of the first subject based on a similarity between usage data of the respiratory device of the first subject and reference usage data. The reference usage data refers to a usage condition that is expected to be achieved by the first subject when the first subject receives respiratory treatment implemented by g the respiratory device. For example, when the first subject receives respiratory treatment implemented by the respiratory device, the first subject may be expected to use the respiratory device for treatment according to a reference usage frequency, a reference usage duration, a reference usage period, etc. The reference usage data may be related to a condition of the first subject and a treatment goal. For example, a healthcare professional may determine or adjust, according to different first subjects and different treatment goals, reference usage data corresponding to the different first subjects and the different treatment goals. The reference usage data may be determined based on an input from the healthcare professional.
[0366] In some embodiments, the cloud platform may construct a usage vector based on the usage data of the respiratory device of the first subject, construct a reference vector based on the reference usage data, determine a vector similarity between the usage vector and the reference vector, and determine the vector similarity as the treatment cooperation degree of the first subject. The vector similarity may be represented by a cosine similarity, a Euclidean distance, etc.
[0367] A life feature of the first subject refers to a feature that is related to the life status of the first subject. In some embodiments, the life feature of the first subject may include a work-rest feature of the first subject and a dietary feature of the first subject.
[0368] The work-rest feature of the first subject refers to a feature that characterizes a work-rest condition of the first subject. In some embodiments, the work-rest feature of the first subject may include a daily time to fall asleep, a frequency of getting up in the night, an average interval between getting up in the night, a wake-up time, etc., of the first subject in a period of time. The period of time refers to a period before a current time. The average interval between getting up in the night refers to an average of intervals each of which is between two adjacent getting ups of the first subject in the night.
[0369] The dietary feature of the first subject refers to a feature that characterizes a diet of the first subject. In some embodiments, the dietary feature of the first subject may include a distribution of daily dietary time, a type of food eaten, an amount of food eaten, etc., of the first subject. The dietary time distribution refers to a set including dietary times of the first subject.
[0370] In some embodiments, the life feature of the first subject may be determined based on an input of a user (e.g., a healthcare professional, the first subject, or a family member of the first subject) via a user terminal (e.g., a first terminal or a second terminal) .
[0371] In some embodiments of the present disclosure, in conjunction with the treatment cooperation degree and / or the life feature of the first subject, the predicted treatment result corresponding to the first candidate treatment plan may be determined from various perspectives, which makes the evaluation of the predicted treatment result more accurate and reliable.
[0372] In some embodiments, the second prediction model may include a first embedding layer, a second embedding layer, and a result prediction layer. The first embedding layer and the second embedding layer may encode input data and adjust data dimensionality of the input data.
[0373] In some embodiments, an input of the first embedding layer may include the treatment cooperation degree of the first subject and the first candidate treatment plan, and an output of the first embedding layer may be a treatment feature embedding vector. The treatment feature embedding vector refers to a feature vector related to the treatment cooperation degree and a treatment plan.
[0374] In some embodiments, an input of the second embedding layer may include the life feature (including the dietary feature of the first subject and the work-rest feature of the first subject) , and an output of the second embedding layer may be a life feature embedding vector. The life feature embedding vector refers to a feature vector related to a life situation of the first subject.
[0375] The result prediction layer refers to a model that determines the predicted treatment result. In some embodiments, the result prediction layer may be a machine learning model, e.g., a convolutional neural network (CNN) model.
[0376] In some embodiments, an input of the result prediction layer may include the respiratory trend of the first subject, the life feature embedding vector, and the treatment feature embedding vector, and an output of the result prediction layer may be the predicted treatment result corresponding to the first candidate treatment plan.
[0377] In some embodiments, the second prediction model may be determined by joint training of the first embedding layer, the second embedding layer, and the result prediction layer through a large number of treatment data samples with labels. In some embodiments, each of the treatment data samples may include a sample respiratory trend of a sample first subject and a sample treatment plan. In some embodiments, the each of the treatment data samples may also include a sample treatment cooperation degree of the sample first subject and / or a sample life feature of the sample first subject (including a sample work-rest feature of the sample first subject and a sample dietary feature of the sample first subject) . In some embodiments, the treatment data samples may be determined based on historical data. For example, a historical treatment plan of the sample first subject in the historical data may be determined as the sample treatment plan, and a historical respiratory trend may be determined as the sample respiratory trend, etc. In some embodiments, each of the labels may include an actual treatment result of the sample first subject. In some embodiments, the cloud platform may determine similarities between various physical indicators of the sample first subject and standard indicators after the sample first subject is treated using the historical treatment plan and designate a score (e.g., a value in a range of 0-1 obtained through normalization) obtained by normalizing values of the similarities as the actual treatment result (i.e., the label) .
[0378] A process of exemplary joint training is illustrated as follows. The sample treatment cooperation degree of the sample first subject and the sample treatment plan in the treatment data sample may be used as an input of an initial first embedding layer. The sample life feature of the sample first subject (including the sample work-rest feature of the sample first subject and the sample dietary feature of the sample first subject) may be used as an input of an initial second embedding layer. An output of the initial result prediction layer may be determined by inputting a treatment feature embedding vector output by the initial first embedding layer, a life feature embedding vector output by the initial second embedding layer, and the sample respiratory trend of the sample first subject into the initial result prediction layer. A loss function may be constructed based on the predicted treatment results output by the initial result prediction layer and the labels. Model parameters of the initial first embedding layer, the initial second embedding layer, and the initial result prediction layer may be iteratively updated based on the loss function until a termination condition is met. A trained first embedding layer, a trained second embedding layer, and a trained result prediction layer may be obtained, thereby obtaining the second prediction model. The termination condition may be that the loss function is smaller than a threshold or converges, or a count of training periods reaches a threshold.
[0379] In some embodiments of the present disclosure, the treatment cooperation degree of the first subject, the first candidate treatment plan, and the life feature of the first subject may be transformed into the feature vectors of unified dimensions through the first embedding layer and the second embedding layer, which is conducive to better extraction of features, so that the predicted treatment result corresponding to the first candidate treatment plan can be determined more accurately through the result prediction layer.
[0380] In 530, one or more second candidate treatment plans may be determined based on the predicted treatment result and the one or more first candidate treatment plans.
[0381] In some embodiments, the cloud platform may determine the one or more second candidate treatment plans in various ways. For example, the cloud platform may select one or more first candidate treatment plans whose predicted treatment results exceed a preset result threshold as the one or more second candidate treatment plans. The preset result threshold may be preset manually based on experience or set by default by the system.
[0382] In 540, the one or more second candidate treatment plans and a predicted treatment result corresponding to each of the one or more second candidate treatment plans may be sent to the second terminal.
[0383] In some embodiments, the cloud platform may send, after determining the one or more second candidate treatment plans, the one or more second candidate treatment plans and the predicted treatment result corresponding to each of the one or more second candidate treatment plans to the second terminal.
[0384] In 550, in response to receiving a first recommendation instruction fed back by the second terminal, a recommended treatment plan may be generated.
[0385] The first recommended instruction refers to an instruction given by the second subject to determine the recommended treatment plan. For example, the first recommendation instruction may include selecting a second candidate treatment plan with a best predicted treatment result as the recommended treatment plan. As another example, the first recommendation instruction may include determining any one of a plurality of second candidate treatment plans as the recommended treatment plan.
[0386] In some embodiments of the present disclosure, the predicted treatment result of the candidate treatment plan of the first subject may be determined based on the respiratory trend of the first subject using the second prediction model, and the recommended treatment plan of the first subject may be determined according to the predicted treatment result and the feedback of the second subject, which combines artificial intelligence (AI) processing with human experience to make the prediction of the treatment result of the candidate treatment plan more intuitive and accurate and to make the determination of the recommended treatment plan have more practical therapeutic potential to achieve a better treatment result.
[0387] In some embodiments, the cloud platform may generate treatment reminder information to remind the first subject based on the recommended treatment plan, generate a treatment reminder parameter based on the respiratory trend of the first subject and / or the treatment cooperation degree of the first subject, and send the treatment reminder information to the first terminal based on the treatment reminder parameter.
[0388] The treatment reminder information refers to information used to remind the first subject to undergo treatment. The treatment reminder information may be used to remind the first subject to undergo a corresponding treatment at a treatment time point. The treatment time point refers to a time point at which the respiratory device is needed to be used. For example, the treatment reminder information may be used to remind the first subject to wear a respirator at 9 p. m., etc.
[0389] The cloud platform may generate the treatment reminder information to remind the first subject based on the recommended treatment plan in various ways. For example, the cloud platform may generate the treatment reminder information reminding the first subject to carry out the treatment plan corresponding to the treatment time point within a preset time range before the treatment time point according to the treatment time point in the recommended treatment plan. The preset time range may be a system default value, an empirical value, an artificial preset value, or the like, or any combination thereof, which may be set according to actual needs and may not be limited in the present disclosure.
[0390] The treatment reminder parameter refers to a parameter related to sending the treatment reminder information to the first subject, e.g., a time period in which a reminder is performed, a frequency of the reminder, etc.
[0391] In some embodiments, the cloud platform may generate the treatment reminder parameter in various ways based on the respiratory trend of the first subject and / or the treatment cooperation degree of the first subject. For example, when the respiratory trend of the first subject trends toward a severe trend and / or the treatment cooperation degree of the first subject is relatively low, the frequency of reminder may be increased. More descriptions regarding the respiratory trend of the first subject may be found in FIGs. 2 and 3 and the relevant descriptions thereof.
[0392] In some embodiments, the cloud platform may send the treatment reminder information to the first terminal according to the treatment reminder parameter.
[0393] In some embodiments, the first terminal may determine a reminder intensity of the treatment reminder information according to a reminder time period of the treatment reminder information. The reminder time period refers to a time period in which the treatment reminder information is sent. The reminder intensity refers to an intensity which the treatment reminder information stimulates a sense organ of the first subject. For example, the reminder intensity may include a light intensity of a warning light emitted by the treatment reminder information through a display screen or a physical light-emitting device, a volume intensity of sound emitted outward by the treatment reminder information through a sound device, etc. For example, when the reminder time period of the treatment reminder information is located at a lunch break time such as 13: 00~14: 00, the reminder intensity may be relatively large.
[0394] In some embodiments, the cloud platform may determine the reminder intensity of the treatment reminder information according to the reminder time period of the treatment reminder information by querying a preset table. The preset table may store a correspondence between the reminder time period of the treatment reminder information and the reminder intensity of the treatment reminder information. The preset table may be obtained based on historical feedback of the first subject.
[0395] In some embodiments of the present disclosure, the treatment reminder information may be generated to remind the first subject to carry out the treatment based on the recommended treatment plan, which prevents the first subject from forgetting or omitting. The treatment reminder parameter that is more in line with an actual situation of the first subject may be generated based on the respiratory trend of the first subject and / or the treatment cooperation degree of the first subject, and the treatment reminder information may be sent to the first terminal according to the treatment reminder parameter, which is conducive to ensuring the normal implementation of the treatment plan.
[0396] In some embodiments, the recommended treatment plan may further include an auxiliary treatment plan.
[0397] The auxiliary treatment plan refers to a plan that treats the first subject other than using the respiratory device. For example, the auxiliary treatment plan may include medication arrangement (including a medication type, a medication cycle, or a daily dosage) , examination arrangement (including a type of medical testing device or a testing time) , a plan related to dietary regulation, and a plan related to auxiliary rehabilitation exercises, etc., of the first subject.
[0398] In some embodiments, the auxiliary treatment plan may be related to at least one of work-rest, activity, or diet of the first subject.
[0399] In some embodiments, the cloud platform may generate a feature vector of the first subject (also referred to as a subject feature vector) based on basic data of the first subject, the respiratory trend of the first subject, and physiological data of the first subject and determine the auxiliary treatment plan by querying a first vector database based on the feature vector of the first subject.
[0400] In some embodiments, the cloud platform may determine a first reference vector that meets a preset matching condition as a first association vector by querying the first vector database based on the feature vector of the first subject (i.e., a first vector to be matched) and determine a reference auxiliary treatment plan corresponding to the first association vector as the auxiliary treatment plan corresponding to the feature vector of the first subject. The first vector database may be constructed based on historical data, or the first vector database may be artificially modified and supplemented. The first vector database may include first reference vectors constructed based on historical basic data, historical respiratory trends, and historical physiological data of a large number of first subjects, and reference auxiliary treatment plans each of which corresponds to one of the first reference vectors. The reference auxiliary treatment plans corresponding to the first reference vectors may be obtained from historical data. In some embodiments, the preset matching condition may include a vector distance between two vectors (e.g., the first reference vector and the feature vector of the first subject) being smaller than a distance threshold, the vector distance being minimized, etc. The distance threshold may be a system default value, an empirical value, an artificially preset value, or the like, or any combination thereof, which may be set according to actual needs and may not be limited in the present disclosure. An exemplary vector distance may include a cosine distance, a Euclidean distance, etc.
[0401] In some embodiments of the present disclosure, the auxiliary treatment plan may be determined from the plurality of reference auxiliary treatment plans in the first vector database, so that the determination of the auxiliary treatment plan is more reasonable and accurate, which is conducive to improving the treatment result.
[0402] In some embodiments, the cloud platform may generate the feature vector of the first subject in various ways based on the basic data of the first subject, the respiratory trend of the first subject, and the physiological data of the first subject. For example, the cloud platform may obtain the feature vector of the first subject by inputting the basic data of the first subject, the respiratory trend of the first subject, and the physiological data of the first subject into a feature extraction model. The feature extraction model may include a trained machine learning model, e.g., a convolutional neural network model, etc.
[0403] In some embodiments, the cloud platform may determine the feature vector of the first subject based on a similarity determination model.
[0404] In some embodiments, the cloud platform may determine one or more reference auxiliary vectors and the feature vector of the first subject through the similarity determination model based on the reference auxiliary treatment plan, relevant data of a first subject corresponding to the reference auxiliary treatment plan, and relevant data of a current first subject, determine a similarity between the feature vector of the first subject and each of the one or more reference auxiliary vectors, and determine a reference auxiliary treatment plan corresponding to a reference auxiliary vector with a highest similarity with the feature vector as the auxiliary treatment plan. The reference auxiliary treatment plan may be an auxiliary treatment plan taken when a historical first subject undergoes auxiliary treatment. The reference auxiliary treatment plan may be determined based on historical data. The relevant data of the first subject may include the basic data of the first subject, the respiratory trend, and physiological data.
[0405] The similarity determination model refers to a model used to determine the similarity between the feature vector of the first subject the and the reference auxiliary vector. In some embodiments, the similarity determination model may include a deep structured semantic model (DSSM) .
[0406] In some embodiments, the similarity determination model may include an input layer, a feature representation layer, and a matching layer.
[0407] In some embodiments, the input layer may include a reference vector branch and a feature vector branch. In some embodiments, each branch in the input layer may perform encoding processing and feature stitching processing on input data.
[0408] The encoding processing may include performing One-Hot encoding and / or Embedding encoding on the input data. The One-Hot encoding may include processing a dense feature of the input data. For example, the dense feature may include consecutive indicator data of the relevant data of the reference subject or the current first subject. As another example, the dense feature may include consecutive indicator data of the reference auxiliary treatment plan of the sample first subject. The Embedding encoding may include processing a sparse feature of the input data. For example, the sparse feature may include partial indicator data of the relevant data of the reference subject or the current first subject. The sparse features corresponding to the basic data of the first subject, the respiratory trend of the first subject, and the physiological data of the first subject may be labeled and determined by the cloud platform in advance.
[0409] The feature stitching processing may include stitching encoding results. For example, a first encoding result may be obtained by the reference vector branch stitching an encoding result of the One-Hot encoding and an encoding result of the Embedding encoding through the feature stitching processing. A second encoding result may be obtained by the feature vector branch stitching the encoding result of the One-Hot encoding and the encoding result of the Embedding encoding through the feature stitching processing.
[0410] In some embodiments, in the input layer, an input of the reference vector branch may include the reference auxiliary treatment plan, the basic data of the first subject of the reference subject corresponding to the reference auxiliary treatment plan, the respiratory trend of the first subject, the physiological data of the first subject, and an output of the reference vector branch may be the first encoding result. An input of the feature vector branch may include the basic data of the first subject of the current first subject, the respiratory trend of the first subject, the physiological data of the first subject, and an output of the feature vector branch may be the second encoding result.
[0411] The feature representation layer refers to a processing layer used to transform an encoding result into a form of a feature vector for representation. In some embodiments, the feature representation layer may include a fully connected layer and an embedding layer. In some embodiments, the feature representation layer may include the reference vector branch and the feature vector branch. In some embodiments, in the feature representation layer, an input of the reference vector branch may include the first encoding result and an output of the reference vector branch may be the reference auxiliary vector. An input of the feature vector branch may include the second encoding result, and an output of the feature vector branch may be the feature vector of the first subject.
[0412] The matching layer refers to a processing layer used to determine a similarity between vectors. In some embodiments, the matching layer may determine the similarity between the vectors using a cos function, etc. In some embodiments, an input of the matching layer may be the reference auxiliary vector and the feature vector of the first subject, and an output of the matching layer may be the similarity between the feature vector of the first subject and the reference auxiliary vector.
[0413] In some embodiments, the similarity determination model may be obtained through training based on a training dataset in various ways. In some embodiments, the training dataset may include a positive sample and a negative sample. The positive sample may correspond to a training label of 1 and the negative sample may correspond to a training label of 0. A ratio of the count of positive samples to the count of negative samples may be a preset ratio. In some embodiments, the positive sample may include relevant data of the sample first subject and a reference auxiliary treatment plan corresponding to the relevant data of the sample first subject, and relevant data (including the basic data, the respiratory trend, and the physiological data of the first subject) of a reference subject whose similarity with the sample first subject is greater than or equal to a similarity threshold and a reference auxiliary treatment plan corresponding to the relevant data of the reference subject whose similarity with the sample first subject is greater than or equal to the similarity threshold. The negative sample may include relevant data of the sample first subject and a reference auxiliary treatment plan corresponding to relevant data of the sample first subject, and relevant data (including the basic data, the respiratory trend, and the physiological data of the first subject) of a reference subject whose similarity with the sample first subject is smaller than the similarity threshold and a reference auxiliary treatment plan corresponding to the relevant data of the reference subject whose similarity with the sample first subject is smaller than the similarity threshold. In some embodiments, the cloud platform may construct the training dataset based on historical data through random sampling according to the preset ratio. The preset ratio may be preset by the cloud platform, etc. A training of the similarity determination model may be similar to the training of the result prediction model. More descriptions may be found in FIG. 2 and the relevant descriptions thereof.
[0414] In some embodiments, the cloud platform may combine the input layer and the feature representation layer of the trained similarity determination model into the feature extraction model, so that the feature vector of the first subject may be generated by inputting the basic data of the first subject, the respiratory trend of the first subject, and the physiological data of the first subject into the feature extraction model.
[0415] In some embodiments of the present disclosure, based on the branches of the deep structured semantic model, the generation of the feature vector of the first subject is more accurate and faster, which facilitates the subsequent determination of the auxiliary treatment plan. Meanwhile, the similarity between the feature vector of the current first subject and the reference auxiliary vector may be determined through the similarity determination model to determine the auxiliary treatment plan, so that patterns may be found from a large amount of data using a self-learning ability of machine learning, the association relationship between the data can be obtained, and the accuracy and efficiency of determining the similarity between the vectors can be improved.
[0416] FIG. 6 is a flowchart illustrating an exemplary process of determining a supervisory relationship according to some embodiments of the present disclosure. In some embodiments, a process 600 may be performed by a cloud platform. As shown in FIG. 6, the process 600 may include the following operations.
[0417] In 610, a relationship matching degree between a second subject and a first subject may be determined based on basic data of the first subject, a respiratory trend of the first subject, and basic data of the second subject.
[0418] The relationship matching degree refers to a parameter that characterizes an adaptation degree between the second subject and the first subject. The higher the relationship matching degree is, the higher the adaptation degree between the second subject and the first subject may be, and the better the treatment effect may be.
[0419] In some embodiments, the cloud platform may determine the relationship matching degree between the second subject and the first subject by processing the basic data of the first subject, the respiratory trend of the first subject, and the basic data of the second subject based on a matching model.
[0420] The matching model may be a machine learning model. A plurality of types of matching models may be provided. For example, the matching model may include a neural network (NN) model, a deep neural network (DNN) model, or the like, or any combination thereof.
[0421] In some embodiments, an input of the matching model may include the basic data of the first subject, the respiratory trend of the first subject, and the basic data of the second subject, and an output of the matching model may include the relationship matching degree between the second subject and the first subject.
[0422] In some embodiments, the matching model may be trained in various feasible ways through a large number of training samples (also referred to as matching training samples) with labels (also referred to as matching degree labels) . The training of the matching model may be similar to the training of an effect prediction model. More descriptions may be found in FIG. 2 and related descriptions thereof.
[0423] In some embodiments, the training samples may include basic data of sample first subjects, sample respiratory trends, and basic data of sample second subjects. A treatment for the sample first subjects may be performed by the sample second subjects. The training samples may be obtained based on historical data. The labels may include actual relationship matching degrees between the sample first subjects and the sample second subjects. The labels may be labeled manually.
[0424] In some embodiments of the present disclosure, the relationship matching degree may be determined by processing the basic data of the first subject, the respiratory trend of the first subject, and the basic data of the second subject based on the matching model. A correlation between the basic data of the first subject, the respiratory trend of the first subject, and the basic data of the second subject, and the matching degree may be obtained by determining rules from a large amount of data using a self-learning ability of the machine learning model, thereby improving the accuracy and efficiency of determining the relationship matching degree.
[0425] In 620, one or more candidate supervisory relationships may be determined based on the relationship matching degree.
[0426] A candidate supervisory relationship refers to an initially determined supervisory relationship. More descriptions regarding the supervisory relationship may be found in FIG. 2 and related descriptions thereof.
[0427] In some embodiments, the cloud platform may establish a candidate supervisory relationship between a second subject and a first subject whose relationship matching degree exceeds a matching degree threshold. The matching degree threshold may be a system default value, an empirical value, a manually preset value, or the like, or any combination thereof. The matching degree threshold may also be set according to actual needs, which is not limited in the present disclosure.
[0428] In 630, the one or more candidate supervisory relationships may be sent to a second terminal, and one or more supervisory relationships may be generated based on a determination instruction fed back by the second terminal.
[0429] The determination instruction refers to an instruction of the second subject for confirming at least one of the one or more candidate supervisory relationships. Merely by way of example, in response to a determination that only one candidate supervisory relationship is provided, the second subject may feed the corresponding determination instruction back to the cloud platform by confirming and establishing the supervisory relationship through operations in the second terminal; in response to a determination that a plurality of candidate supervisory relationships are provided, the second subject may feed the corresponding determination instruction back to the cloud platform by selecting one or more candidate supervisory relationships from the plurality of candidate supervisory relationships to establish the one or more supervisory relationships through operations in the second terminal based on a load of the second subject.
[0430] In some embodiments of the present disclosure, the supervisory relationship may be established between a second subject and a first subject whose relationship matching degree as high as possible, so that the second subject and the first subject may cooperate with each other for subsequent treatment processes. The second subject may perform confirmation through the second terminal, so that the second subject may perform autonomous selection based on a current load status of the second subject to avoid a situation of poor experience of the first subject or resource vacancy of the second subject caused by too many or too few supervisory relationships automatically assigned to the second subject by the system.
[0431] In some embodiments, the cloud platform may generate at least one evaluation feature of the first subject based on historical feedback of the first subject, and determine a change probability of the supervisory relationship based on the basic data of the first subject, the at least one evaluation feature of the first subject, and the respiratory trend of the first subject. In response to a determination that the change probability is greater than a change probability threshold, the cloud platform may determine at least one alternative supervisory relationship based on the relationship matching degree.
[0432] The historical feedback of the first subject refers to related feedback information from the first subject to the second subject (e.g., a doctor or nurse) in a past time period, such as positive feedback from the first subject to the second subject (e.g., the first subject considers that the second subject has good medical skills, good attitude, etc. ) , negative feedback from the first subject to the second subject (e.g., the first subject considers that the second subject has poor medical skills, poor attitude, etc. ) etc.
[0433] An evaluation feature refers to a related parameter that reflects an evaluation situation of the first subject to the second subject. For example, the evaluation feature may include a feedback score. The higher the feedback score is, the better the evaluation of the first subject to the second subject may be.
[0434] In some embodiments, the cloud platform may extract the positive feedback information and the negative feedback information from the first subject to the second subject through a feedback determination model based on the historical feedback of the first subject, and generate the at least one evaluation feature of the first subject based on the positive feedback information and the negative feedback information. In some embodiments, the feedback determination model may be a language model (LM) , etc. The feedback determination model may be obtained based on existing models or training.
[0435] In some embodiments, the cloud platform may determine a scoring table including various feedback information and the feedback scores corresponding to the various feedback information based on historical data, look up the scoring table to determine the feedback scores corresponding to the positive feedback information and the negative feedback information, and obtain the at least one evaluation feature of the first subject based on the feedback scores and influence coefficients corresponding to the various feedback information included in the historical feedback of the first subject. For example, the cloud platform may determine a product of the feedback score and the influence coefficient corresponding to each feedback information as a sub-score, and determine a sum of the sub-scores of all feedback information as the feedback score of the first subject.
[0436] The change probability refers to a probability of change in the supervisory relationship. The change in the supervisory relationship may include changing the second subject, adding a second subject, reducing a second subject, etc.
[0437] In some embodiments, the cloud platform may perform normalization processing based on the basic data of the first subject, the at least one evaluation feature of the first subject, and the respiratory trend of the first subject, and determine the change probability of the supervisory relationship by performing a weighting operation on the basic data of the first subject, the evaluation feature (s) of the first subject, and the respiratory trend of the first subject subjected to the normalization processing. The normalization processing may normalize the basic data of the first subject, the at least one evaluation feature of the first subject, and the respiratory trend of the first subject to a preset value range (e.g., a range of [0, 1] ) . In some embodiments, the cloud platform may obtain the change probability of the supervisory relationship by performing normalization processing on a count of historical changes and then perform the weighting operation on the basic data of the first subject, the at least one evaluation feature of the first subject, and the respiratory trend of the first subject after the normalization processing. A weight may be a default value, a preset value, or a value determined based on the historical data.
[0438] More descriptions regarding the normalization processing may be found in FIG. 2 and related descriptions thereof. The weights corresponding to the basic data of the first subject, the at least one evaluation feature of the first subject, and the respiratory trend of the first subject, respectively, may be preset by the system or preset manually.
[0439] An alternative candidate supervisory relationship refers to a supervisory relationship used to replace an original supervisory relationship. In some embodiments, the alternative supervisory relationship (s) may include a replacement supervisory relationship and an additional supervisory relationship. The replacement supervisory relationship refers to a supervisory relationship used to replace a current supervisory relationship. The additional supervisory relationship refers to another supervisory relationship newly added to the current supervisory relationship.
[0440] In some embodiments, in response to a determination that the change probability is greater than the change probability threshold, the cloud platform may determine the alternative supervisory relationship based on the relationship matching degree. The change probability threshold may include a first change probability threshold and a second change probability threshold, and the first change probability threshold may be greater than the second change probability threshold. For example, in response to a determination that the change probability is greater than the first change probability threshold, the cloud platform may select another second subject (e.g., another doctor) with a highest relationship matching degree with the first subject to establish a replacement supervisory relationship based on a relationship matching degree between the first subject and the another second subject (i.e., other second subjects other than the second subject in the current supervisory relationship) ; in response to a determination that the change probability is between the first change probability threshold and the second change probability threshold, the cloud platform may select another second subject (e.g., another second subject) with a highest relationship matching degree with the first subject to establish an additional supervisory relationship based on a relationship matching between the first subject and other second subjects; and in response to a determination that the change probability is less than the second change probability threshold, the cloud platform may keep the current supervisory relationship to be unchanged. The change probability threshold may be preset based on experience or preset based on actual needs.
[0441] In some embodiments of the present disclosure, the at least one evaluation feature may be determined based on the historical feedback of the first subject, and the change probability of the supervisory relationship may be determined based on the basic data of the first subject, the at least one evaluation featur of the first subject, and the respiratory trend of the first subject to determine the alternative evaluation relationship (s) , so that another alternative supervisory relationship may be timely and accurately established for the first subject to avoid delaying treatment progress when the treatment effect of the current second subject is poor or the experience of the first subject is poor.
[0442] As illustrated in FIG. 7, in some embodiments, the cloud platform may also construct an initial supervisory relationship map 720 based on basic data 310 of the first subject, a respiratory trend 320 of the first subject, and basic data 710 of the second subject, obtain a plurality of candidate supervisory relationship maps 730 by adjusting the initial supervisory relationship map 720 based on an adjustment rule, determine an estimated fatigue degree 750 of each second node and a care perfection degree 760 of each first node of the plurality of candidate supervisory relationship maps 730 by processing the plurality of candidate supervisory relationship maps 730 through a relationship assessment model 740, and generate a supervisory relationship 770 based on the estimated fatigue degree 750 of each second node and the care perfection degree 760 of each first node of the plurality of candidate supervisory relationship maps 730.
[0443] A supervisory relationship map refers to a map used to reflect the supervisory relationships between the second subject (e.g., a doctor, a nurse) and the first subject (e.g., a patient) . The supervisory relationship map may be a data structure composed of nodes and edges. An edge may connect nodes. A nodes and An edge may have attributes. The initial supervisory relationship map refers to an initially determined supervisory relationship map.
[0444] In some embodiments, the supervisory relationship map may include nodes of multiple node types. The node types of the supervisory relationship map may include a second node (e.g., a doctor node, a nurse node) , and a first node (e.g., a patient node) . The node attributes of the second node may include the basic data of the second subject (e.g., a doctor, a nurse) . For example, the node attributes of the nurse node may include information related to the nurse (i.e., data information related to the nurse t, such as the gender of the nurse, the disease in which the nurse specializes, a count of patients cared for in a historical time period, a recovery status of the patients, the time period the nurse has been working, the department (e.g., a department of a hospital) the nurse belongs to, etc. ) . The node attributes of the first node may include the basic data of the first subject, physiological data of the first subject, and the respiratory trend of the first subject. More descriptions regarding the basic data of the first subject, the physiological data of the first subject, and the respiratory trend of the first subject may be found in FIGs. 2-3 and related descriptions thereof.
[0445] In some embodiments, the node features of the first node may include a treatment schedule set of the first subject.
[0446] A treatment schedule set refers to a set including at least one group of treatment schedules of the first subject. In some embodiments, a treatment schedule may include one treatment phase of a recommended treatment plan of the first subject and a time period corresponding to the treatment phase.
[0447] In some embodiments, the cloud platform may determine a plurality of candidate treatment schedules based on the recommended treatment plan of the first subject, and determine the treatment schedule set based on the plurality of candidate treatment schedules. More descriptions regarding the recommended treatment plan may be found in FIG. 2 and FIG. 5 and related descriptions thereof.
[0448] A candidate treatment schedule refers to an initially determined treatment schedule. In some embodiments, the cloud platform may determine the plurality of candidate treatment schedules based on the treatment phases, treatment durations, treatment time points, etc., in the recommended treatment plan. For example, the cloud platform may determine a start time and an end time of a treatment phase based on the treatment time point and the treatment duration of the treatment phase, and determine the plurality of candidate treatment schedules based on the start times and the end times of the plurality of treatment phases. Merely by way of example, the treatment phases in the recommended treatment plan may include A, B, C, D, E, and F, start times corresponding to the treatment phases A, B, C, D, E, and F may be denoted as ts1-ts6, respectively, and end times corresponding to the treatment phases A, B, C, D, E, and F may be denoted as te1-te6, respectively. The constructed candidate treatment schedules may include (ts1, te1) , (ts2, te2) , (ts3, te3) , (ts4, te4) , (ts5, te5) , (ts6, te6) , etc.
[0449] In some embodiments, the cloud platform may obtain the treatment schedule set by combining some or all candidate treatment schedules. In some embodiments, the recommended treatment plan may include replaceable treatment phases and optional treatment phases. Correspondingly, when combining the treatment schedules, the cloud platform may replace the replaceable treatment phases and / or delete or retain the optional treatment phases to obtain a plurality of candidate treatment schedule sets. For example, in the above example, the treatment phase C and the treatment phase D may be replaceable each other, and the treatment phase F may be optional, then the constructed candidate treatment schedule sets may include a candidate treatment schedule set 1 denoted as [ (ts1, te1) , (ts2 , te2 ) , (ts3, te3) , (ts5, te5) , (ts6, te6) ] , a candidate treatment schedule set 2 denoted as [ (ts1, te1) , (ts2, te2) , (ts4, te4) , (ts5, te5) , (ts6, te6) ] , a candidate treatment schedule set 3 denoted as [ (ts1, te1) , (ts2, te2) , (ts3, te3) , (ts5, te5) ] , etc. In some embodiments, the cloud platform may send the candidate treatment schedule sets to a first terminal and / or a second terminal, and determine a final treatment schedule set based on feedback from the first subject and / or the second subject.
[0450] In some embodiments of the present disclosure, by setting the node attributes of the first node including the treatment schedule set of the first subject, the treatment progress of the first subject may also be considered in subsequent work such as supervisory relationship assessment, or the like, thereby making the data for subsequent analysis more accurate, and also facilitating to match the second subject who has sufficient time for the treatment of the first subject based on a treatment progress.
[0451] In some embodiments, an edge of the supervisory relationship map connecting two nodes may represent historical supervisory relationships between the two nodes. A historical supervisory relationship refers to a supervisory relationship between the first subject and the second subject during a historical treatment process. Correspondingly, when the first subject and the second subject have the historical supervisory relationship, an edge between nodes may be used to connect the corresponding nodes of the first subject and the second subject.
[0452] In some embodiments, an edge attribute may include a supervisory feature. The supervision feature refers to feature information of the second subject supervising the first subject. For example, the supervision feature may include supervision time, a supervision frequency, a supervision intensity, etc. The supervision time refers to time information related to supervision, such as an average duration required for a single supervision, etc. The supervision frequency refers to frequency information related to supervision, such as a ratio of a count of times for supervisions in a certain historical time period to the length of the historical time period, etc. The supervision intensity refers to parameter information representing an amount of supervision content. The greater the supervision intensity is, the more the amount of supervision content performed by the second subject for the first subject may be. The amount of supervision content may include each indicator data for of respiratory treatment data of the first subject. The supervision feature may be obtained based on historical data, user input information, or other ways.
[0453] In some embodiments, the cloud platform may obtain a plurality of candidate supervisory relationship maps by adjusting the initial supervisory relationship map based on an adjustment rule. In some embodiments, the adjustment rule may be randomly increasing or reducing edges between the second node and the first node (e.g., a patient node and a doctor node, a patient node and a nurse node) . The adjustment rule may also be implemented in other manners, which may be set based on experience or actual needs.
[0454] In some embodiments, the relationship assessment model may be a graph neural network (GNN) model.
[0455] An input of the relationship assessment model may be the plurality of candidate supervisory relationship maps, and an output of the relationship assessment model may be an estimated fatigue degree of each second subject node and a care perfection degree of each first node in the plurality of candidate supervisory relationship maps. The relationship assessment model may determine the estimated fatigue degree of each second subject node and the care perfection degree of each first node in the plurality of candidate supervisory relationship maps by processing the plurality of candidate supervisory relationship maps for multiple times, respectively.
[0456] The relationship assessment model may also be other graph model, such as a graph convolutional neural network (GCNN) model, or may be obtained by adding other processing layers to the GNN model, or modifying processing manners in the GNN model, etc.
[0457] An estimated fatigue degree refers to a parameter that characterizes a fatigue level of the second subject estimated based on the actual workload of the second subject.
[0458] The care perfection degree refers to a parameter that characterizes a perfection level of care the first subject receives.
[0459] The relationship assessment model may be trained based on training data. The training data may include relationship assessment samples and relationship assessment labels. For example, a relationship assessment sample may include a the historical supervisory relationship map determined based on the historical data. The nodes, the node attributes, the edges, and the edge attributes of the historical supervisory relationship map may be similar to those of the supervisory relationship map. The relationship assessment sample may include a historical estimated fatigue degree of a historical second node (e.g., a historical doctor node, a historical nurse node) and a historical care perfection degree of a historical first node (e.g., a historical patient node) in the historical supervisory relationship map. The relationship assessment samples may be obtained based on the historical data, and the relationship assessment labels may be determined based on manual labeling. The historical estimated fatigue of each historical second node in the relationship assessment label may be determined based on average working hours per day (e.g., the average working hours per day may be determined based on clock-in time, etc. ) of each second subject, regular physical examination results, or other indicators of the historical second node under a certain historical supervisory relationship through weighting after the normalization processing is performed on the indicators of the historical second nodes mentioned above. Weights may be set based on experience. The historical care perfection degree of each historical first node may be determined based on a count of on-time cares (e.g., a count of times that the second subject provides a service such as on-time medication, on-time infusion, on-time accompany, etc., to the first subject) of each first subject, a count of demand responses (e.g., a count of times the second subject responds to a call bell of the first subject) to the first subject, or other indicators of the historical first node under a certain historical supervisory relationship through weighting after the normalization processing is performed on the indicators of historical first node mentioned above. Weights may be set based on experience. More descriptions regarding the normalization processing may be found in the related descriptions of FIGs. 2-3.
[0460] In some embodiments, the cloud platform may determine a relationship score of the first subject and the second subject (e.g., a doctor-patient relationship score) based on the estimated fatigue of each second node and the care perfection degree of each first node in the plurality of candidate supervisory relationship maps. The cloud platform may select the candidate supervisory relationship map with the highest relationship score as a target supervisory relationship map. The cloud platform may generate a supervisory relationship based on edges in the target supervisory relationship map.
[0461] The relationship score refers to a parameter that represents a quality of a relationship between the second subject and the first subject. The higher the relationship score is, the better the relationship between the second subject and the first subject may be. In some embodiments, the cloud platform may determine a ratio of a sum of the care perfection degrees of all the first nodes to a sum of the estimated fatigue degrees of all the second nodes as the relationship score of the first subject and the second subject.
[0462] In some embodiments, the cloud platform may establish the supervisory relationship between the second subject and the first subject based on the second subject and the first subject with the edges in the target supervisory relationship map.
[0463] In some embodiments of the present disclosure, the supervisory relationship map is constructed based on the basic data of the first subject, the respiratory trend of the first subject, and the basic data of the second subject, the estimated fatigue degree of the second node and the care perfection degree of the first node are determined by processing the supervisory relationship map through the relationship assessment model, and the supervisory relationship is generated based on the estimated fatigue degree and the care perfection degree, so that a reasonable supervisory relationship can be determined by comprehensively considering the workload of the second subject, and the care experience of the first subject based on actual situations, thereby ensuring a good and healthy relationship between the first subject and the second subject.
[0464] In some embodiments, the cloud platform may generate a ward adjustment instruction with a frequency less than a preset adjustment frequency based on the supervisory relationship between the second subject and the first subject, and send the ward adjustment instruction to the user terminal to reminder the first subject to adjust a ward the first subject is admitted to. The adjustment frequency may be a preset value. It should be understood that a purpose of setting the adjustment frequency is to avoid a frequent adjustment to the ward, which wastes human resources and reduces experience of the first subject. In some embodiments, the ward adjustment instruction may include that admission wards of first subjects supervised by a same second subject may be adjusted to a same ward or wards physically adjacent to each other. It should be understood that by adjusting the admission wards of the first subjects supervised by the same second subject to the same ward or the wards physically adjacent to each other, a supervision efficiency of the second subject may be improved.
[0465] In some embodiments, the cloud platform may determine whether the supervisory relationship needs to be adjusted based on a change in various information (e.g., the respiratory trend, the respiratory treatment data, the actual care perfection degree during a period of time, etc. ) of the first subject over a period of time during which the current supervisory relationship is operated and a change in various information (e.g., the actual fatigue degree of the second subject during a period of time) of the second subject. In response to a determination that the supervisory relationship needs to be adjusted, the cloud platform may be configured to generate a new supervisory relationship based on any of the manners mentioned above and send the new supervisory relationship to the first subject and / or the second subject. An actual supervisory relationship may be determined based on a feedback opinion of the first subject and / or the second subject, and the supervisory relationship may be adjusted based on the actual supervisory relationship. For example, if a frequency of abnormal occurrence in the respiratory treatment data of a first subject is higher than a preset value, and / or the actual care perfection degree of a first subject is lower than a preset value over a period of time (e.g., a week) during which the current supervisory relationship is operated, the cloud platform may determine that the supervisory relationship needs to be adjusted. As another example, if the actual fatigue degree of a second subject is higher than a preset value over a period of time (e.g., a week) during which the current supervisory relationship is operated, the cloud platform may determine that the supervisory relationship needs to be adjusted.
[0466] FIG. 8 is a flowchart illustrating an exemplary process for generating a target report according to some embodiments of the present disclosure. In some embodiments, the process 800 may be performed by a cloud platform. As shown in FIG. 8, the process 800 may include the following operations.
[0467] In 810, in response to receiving a report generation instruction from a user terminal, a report benchmark may be determined. More descriptions regarding the user terminal may be found in FIG. 1A and the relevant descriptions thereof.
[0468] The report generation instruction refers to a relevant instruction used to generate a required report. In some embodiments, the report generation instruction may be determined based on a first terminal and / or a second terminal. A first subject and / or a second subject may input the report generation instruction via a report generation button, etc., on the first terminal and / or the second terminal. The cloud platform may be communicatively connected with the user terminal through a wired connection or a wireless connection and receive the report generation instruction from the first terminal and / or the second terminal.
[0469] The report benchmark refers to relevant benchmark information to which the report needs to conform. For example, the report benchmark may include prioritizing textual content, prioritizing chart content, etc.
[0470] In some embodiments, the cloud platform may determine the report benchmark based on user basic data and / or respiratory treatment data. In some embodiments, the cloud platform may determine, based on a type of user terminal, the report benchmark corresponding to the user terminal. In some embodiments, different types of user terminals may be preset with report benchmarks corresponding to the different types of user terminals. The report benchmark corresponding to the first terminal and the report benchmark corresponding to the second terminal may be different. When receiving the report generation instruction from a type of user terminal, the cloud platform may generate the report benchmark corresponding to the type of the user terminal.
[0471] In some embodiments, the report benchmark may include a content item of a report and / or a report parameter.
[0472] The content item refers to a type of data included in the report. In some embodiments, the content item may include indicator data of a respiratory device, a monitoring parameter of a first monitoring device, and / or a monitoring parameter of a second monitoring device. For example, the content item may include at least one of an Apnea-Hypopnea Index (AHI) , an oxygen saturation, a minute ventilation, etc.
[0473] In some embodiments, the cloud platform may determine the content item based on the respiratory treatment data. In some embodiments, the cloud platform may determine the indicator data included in the respiratory treatment data as the content item. Exemplarily, if the respiratory treatment data includes a monitoring parameter such as the AHI and the oxygen saturation, the cloud platform may determine the AHI and the oxygen saturation as the content items. It should be noted that a count of content items may be preset according to actual needs.
[0474] In some embodiments, the cloud platform may also determine the content item based on basic data, a respiratory trend, and the respiratory treatment data of the first subject through a content item determination model.
[0475] The content item determination model refers to a model used to determine the content item. In some embodiments, the content item determination model may include a machine learning model (e.g., a Neural Network (NN) model or a convolutional neural network (CNN) model) . A model type may be chosen according to a specific circumstance.
[0476] In some embodiments, an input of the content item determination model may include the basic data, the respiratory trend, and the respiratory treatment data of the first subject, and an output of the content item determination model may include at least one content item. In some embodiments, a count of content items output by the content item determination model may be preset. More descriptions regarding the basic data, the respiratory trend, and the respiratory treatment data of the first subject may be found in FIGs. 2 and 3 and the relevant descriptions thereof.
[0477] In some embodiments, the content item determination model may be obtained by training an initial machine learning model based on a plurality of training samples (also referred to as content item training samples) with labels (also referred to as content item labels) in various ways. The training of the content item determination model may be similar to the training of the result prediction model. More descriptions may be found in FIG. 2 and the descriptions thereof.
[0478] In some embodiments, each of the content item training samples may include sample basic data, a sample respiratory trend, and sample respiratory treatment data of a sample first subject. The content item training samples may be determined based on historical data. The content item labels may be content items corresponding to the content item training samples. The content item labels may be obtained based on manual labeling or through statistics of historical data. For example, the cloud platform may determine data content about which the sample first subject is concerned as the content item labels.
[0479] In some embodiments, the cloud platform may count durations and frequencies of the sample first subject viewing different historical content items in historical data corresponding to a content item training sample. The cloud platform may determine one or more historical content items each of which corresponds to a viewing duration and a viewing frequency satisfying a preset condition as one or more content item labels corresponding to the sample first subject. The preset condition may be used to determine whether a content item may be used as the label. For example, the preset condition may include the viewing duration of a historical content item being greater than a first threshold, and / or the viewing frequency of a historical content item being greater than a second threshold. The first threshold and / or the second threshold may be set in advance.
[0480] In some embodiments of the present disclosure, through the content item determination model, patterns may be found from a large amount of relevant data using a self-learning ability of the machine learning model, which determines the content item efficiently and accurately and obtains a better result than manual setting.
[0481] The report parameter refers to parameter information related to a form of the report. In some embodiments, the report parameter may be divided into a first report parameter adapted to the first subject (also referred to as a first subject-adapted report parameter) and a second report parameter adapted to the second subject (also referred to as a second subject-adapted report parameter) . The first subject-adapted report parameter may be used to present the content item to the first subject at the first terminal. The second subject-adapted report parameter may be used to present the content item to the second subject at the second terminal. The first subject-adapted report parameter may differ from the second subject-adapted report parameter.
[0482] In some embodiments, the cloud platform may determine the report parameter based on the user basic data.
[0483] In some embodiments, the cloud platform may determine the first subject-adapted report parameter based on the basic data of the first subject. In some embodiments, the cloud platform may construct a second vector to be matched based on the basic data of the first subject and determine the report parameter by querying a second vector database. The second vector database may be constructed based on historical data, or the second vector database may be artificially modified and supplemented. The second vector database may include one or more second reference vectors constructed based on historical basic data of the first subject and a reference report parameter corresponding to each of the second reference vectors. The report parameter may be determined through the second vector database in a manner similar to the manner in which the auxiliary treatment plan is determined through the first vector database. More descriptions may be found in FIG. 8 and the relevant descriptions thereof.
[0484] In some embodiments, the cloud platform may determine the second subject-adapted report parameter based on basic data of the second subject. In some embodiments, the cloud platform may obtain and analyze historical report data corresponding to the second subject based on the basic data of the second subject to determine the report parameter. For example, the cloud platform may determine a historical report parameter that is used most frequently by the second subject in the historical report data as a current report parameter. In some embodiments, the second subject may also change or manually select the determined report parameter to meet a usage requirement.
[0485] Merely by way of example, the first subject-adapted report parameter may include a higher degree of visualization, more prominent key information, a straightforward and easy-to-understand language style, and a style and a layout that may be changed according to basic information of the first subject, etc. The second subject-adapted report parameter may include a lower degree of visualization, more content items, and a professional and comprehensive language style, and there may not be too many changes in the style and layout, etc. As another example, if the basic data of the first subject shows that the first subject is relatively old, the first subject-adapted report may be determined to be a report template with a relatively large display font, a simple style, and a high degree of visualization for display. The report template may be preset by the system or artificially.
[0486] In some embodiments of the present disclosure, the first subject-adapted report parameter may be determined based on the basic data of the first subject, and the generated report may be used to be sent to the first terminal to be displayed to the first subject, so that compared to the second subject-adapted report parameter, the first subject-adapted report parameter may have a higher degree of visualization, and the key data is more prominent, which is convenient for non-professional first subjects to understand. The second subject-adapted report parameter may be determined based on the basic data of the second subject, the generated report may be sent to the second terminal to be displayed to the subject, so that the report may be more comprehensive and professional.
[0487] In some embodiments, the report parameter may include at least one of a report order, a report format, a language style, etc., of the one or more content items. The first subject-adapted report parameter may be different from the second subject-adapted parameter in at least one of the report order, a level of detail, the report format, or the language style.
[0488] The report order of content items refers to an order in which a plurality of content items are reported. In some embodiments, the report order of the content items may be preset by the system or artificially. For example, the cloud platform may rank a data type that the first subject focus on at a top of the plurality of content items. More descriptions regarding the data type that is focused on may be found in FIG. 2 and the relevant descriptions thereof.
[0489] In some embodiments, the cloud platform may determine the report order of the content items based on a priority of the content item. For example, the cloud platform may set a content item with a relatively high priority to be reported at a top position.
[0490] In some embodiments, the cloud platform may determine the priority of the content item based on a user input.
[0491] In some embodiments, the cloud platform may also determine the priority of the content item based on the content item determination model. In this embodiment, the output of the content item determination model may include at least one content item and a priority corresponding to the at least one content item. Accordingly, when training the content item determination model, the content item and the priority corresponding to the content item may be used as a content item label. The cloud platform may obtain the content items and the priorities corresponding to the content items by sorting one or more content items according to the viewing durations and viewing frequencies. The longer the view duration is, and / or the greater the viewing frequency is, the higher the ranking of the content item corresponding to the view duration and / or the viewing frequency may be, and the higher the priority corresponding to the content item may be. The cloud platform may obtain the content item label by marking, according to concern degrees of the first subject of each data item of the basic data, the respiratory trend, or the respiratory treatment data of the first subject, etc., the priorities of the content items corresponding to the basic data, the respiratory trend, and the respiratory treatment data of the first subject. The first subject may input the concern degree of each data item through the first terminal.
[0492] In some embodiments, the report format may include various items such as a layout, a font size, a style, a degree of visualization, etc., of the content item. The degree of visualization refers to a degree to which data is represented using a visual mean such as a graph or a chart. The report format for the first subject (e.g., a young patient) may include a high degree of visualization, prominent key information, and varied styles and layouts. The report format for the second subject may include a low degree of visualization, a single style and layout, and a comprehensive distribution of information.
[0493] The language style of the content item may be divided into a simple language style, a professional language style, etc. The content item of a report adapted to the first subject (also referred to as a first subject-adapted content item) may be described in the simple language style. The content item of a report adapted to the second subject (also referred to as a second subject-adapted content item) may be described in the professional language style.
[0494] In some embodiments of the present disclosure, the content item may be determined based on the respiratory treatment data, and the report parameter may be determined based on the user basic data, which makes the determined report parameter more in line with the actual needs of the user. For example, the report parameter may be determined based on the basic data of the first subject, which makes it easier for the first subject to understand key information. The report parameter may be determined based on the basic data of the second subject, which makes it easier for the second subject to analyze in conjunction with the comprehensive data.
[0495] In 820, a target report meeting the report benchmark may be generated.
[0496] The target report refers to a report that needs to be sent to the user terminal. The target report may include a first report adapted to the first subject (also referred to as a first subject-adapted report) sent to the first terminal and / or a second report adapted to the second subject (also referred to as a second subject-adapted report) sent to the second terminal.
[0497] In some embodiments, the cloud platform may generate the target report based on the report benchmark through a technology such as information extraction or deep learning. The cloud platform may send the generated target report to the first terminal and / or the second terminal.
[0498] In some embodiments, the cloud platform may determine a comprehensive recommendation of the second subject and a diagnostic analysis corresponding to the at least one content item through a report analysis model based on the basic data, the respiratory trend, and the respiratory treatment data of the first subject, and the at least one content item. In some embodiments, the cloud platform may generate the target report by lay outing the at least one content item, the diagnostic analysis corresponding to the at least one content item, and the comprehensive recommendation of the second subject according to the report parameter.
[0499] The report analysis model refers to a model used to determine the diagnostic analysis corresponding to the content item and the comprehensive recommendation of the second subject. In some embodiments, the report analysis model may be a language model (e.g., a large language model) .
[0500] The large language model (LLM) refers to a machine learning model with a large parameter scale that is obtained by training based on a deep learning technique, large-scale data, and computational resources. The LLM may be primarily oriented toward natural language processing but may evolve to process other forms of data. An exemplary large language model may include a bidirectional encoder representation from transformers (BERT) model, a generative pre-trained transformer (GPT) model, an XL-Net model, a ChatGLM-6B model, etc.
[0501] In some embodiments, the report analysis model may be a language model based on a pre-trained and fine-tuned mode. The language model based on a pre-trained and fine-tuned mode refers to a language model that includes two phases of processing: general feature extraction and task output, such as BERT. The two phases of the language model based on a pre-trained and fine-tuned mode may be trained separately. The phase of general feature extraction may be pre-trained and the phase of task output may be trained (fine-tuned) based on an application task. The general feature extraction may be performed using a trained feature extraction model (e.g., the pre-trained part of BERT) . A result of the task output may be determined through a task processing layer based on a feature extracted in the phase of general feature extraction. The task processing layer may be a neural network or other structures. The task processing layer may be obtained by training (fine-tuning) . Training data of the task processing layer may be formed by training a feature extraction model based on original training data, and the task processing layer may be trained based on a label through supervised learning.
[0502] In some embodiments, an input of the report analysis model may include the basic data, the respiratory trend, and the respiratory treatment data of the first subject, and the at least one content item, and an output of the report analysis model may include the comprehensive recommendation of the second subject and the diagnostic analysis corresponding to the at least one content item.
[0503] In some embodiments, the cloud platform may obtain a pre-trained model and obtain the report analysis model by fine-tuning the pre-trained model.
[0504] In some embodiments, the pre-trained model may be a language model or a large language model obtained after a pre-training phase. The pre-training phase refers to a phase in which the language model is trained based on large-scale data through an unsupervised learning manner.
[0505] The unsupervised learning manner refers to a training manner in which the language model learns the rules of language and a way to represent language by reading a large amount of unlabeled text. For example, the unsupervised learning manner may include masked language modeling (MLM) , autoregressive language modeling (ALM) , etc. In the pre-training process, the pre-trained model may learn contextual information of language, such as a grammatical structure of a sentence, or a relationship between words. Such information may help the pre-trained model understand the language better and perform better in subsequent tasks. At the same time, the pre-training may reduce an amount of labeled data required by a certain task and reduce training costs.
[0506] In some embodiments, the pre-trained model may be a language model or a large language model obtained by pre-training based on a pre-trained dataset. The pre-trained dataset may be a general text corpus. The pre-trained dataset may consist of a myriad of text sources, including books, articles, or websites. The data in the pre-trained dataset is carefully curated to ensure a comprehensive reflection of human knowledge, linguistic nuances, and cultural perspectives. The pre-trained dataset may be usually a large-scale dataset containing rich features and samples.
[0507] The pre-trained model may be suitable for various scenarios. In practical applications, a language model or a large language model dedicated to a specific task may be obtained by fine-tuning training.
[0508] In some embodiments, the cloud platform may obtain an existing pre-trained model via the network. For example, BERT, GPT, or XLNet may be used as the pre-trained model. In some embodiments, the cloud platform may obtain the pre-trained model by pre-training. For example, the cloud platform may train GPT based on pre-training data or further enhance the language expression ability of GPT using a technology such as supervised fine-tuning, feedback self-help, human feedback reinforcement learning, etc.
[0509] In some embodiments, the cloud platform may obtain the report analysis model by fine-tuning the pre-trained model based on a report analysis training sample with a report analysis label through a preset fine-tuning algorithm. The preset fine-tuning algorithm refers to an algorithm that fine-tunes a model parameter.
[0510] In some embodiments, the preset fine-tuning algorithm may include a Low-Rank Adaptation (LoRA) algorithm. The LoRA algorithm may freeze a weight of the pre-trained model and inject a trainable layer (e.g., a rank-decomposition matrix) into each Transformer block. Because there is no need to determine weight gradients for most models, a count of parameters needed to be trained may be greatly reduced and a memory requirement of the GPU is lowered.
[0511] In some embodiments, the process of model fine-tuning through LoRA may include: constructing a newly added linear layer in the pre-trained model, inputting a report analysis training set (including the report analysis training sample with the report analysis label) into an original structural layer and the newly added linear layer in the pre-trained model, respectively, and training the pre-trained model based on the report analysis training set; in the training process, keeping a first weight matrix of the original structural layer unchanged and training a second weight matrix of the newly added linear layer to obtain a trained second weight matrix; and obtaining a trained report analysis model based on the first weight matrix and the trained second weight matrix. A sum result of the first weight matrix and the second weight matrix may be an original weight matrix of the pre-trained model without the newly added linear layer.
[0512] In some embodiments, the newly added linear layer may include a first linear layer and a second linear layer. The first linear layer may be configured to perform dimensionality reduction processing on the second weight matrix to obtain a low-rank matrix. The second linear layer may be configured to perform dimensionality increasing processing on the low-rank matrix to restore the original dimensionality of the second weight matrix.
[0513] In some embodiments, the process of training the second weight matrix of the newly added linear layer may include the following operations S21 -S24.
[0514] In S21, at least two sub-weight matrices may be obtained by performing the dimensionality reduction processing on the second weight matrix using the first linear layer. The sub-weight matrix refers to the low-rank matrix obtained by performing the dimensionality reduction processing based on a weight update matrix with complete rank (e.g., the second weight matrix) . For example, the first linear layer may generate the at least two sub-weight matrices (e.g., a sub-weight matrix A and a sub-weight matrix B) by performing the dimensionality reduction processing on the second weight matrix, and a product of the at least two sub-weight matrices (e.g., the sub-weight matrix A and the sub-weight matrix B) may be the second weight matrix.
[0515] In S22, at least two trained sub-weight matrices may be obtained by iteratively updating the at least two sub-weight matrices based on the report analysis training set. For example, the sub-weight matrix A and the sub-weight matrix B may be iteratively updated, respectively, the iteration may end when a preset condition is satisfied, and a trained sub-weight matrix A and a trained sub-weight matrix B may be obtained.
[0516] In S23, the trained second weight matrix may be obtained by performing the dimensionality increasing processing on the at least two trained sub-weight matrices using the second linear layer. For example, the trained sub-weight matrix A and the trained sub-weight matrix B may be input into the second linear layer for dimensionality increasing processing to restore the original dimensionality of the second weight matrix to obtain the trained second weight matrix.
[0517] In S24, the trained report analysis model may be obtained based on the first weight matrix and the trained second weight matrix. For example, the first weight matrix and the trained second weight matrix may be added together to obtain a trained weight matrix, that is designated as the trained report analysis model.
[0518] In some embodiments of the present disclosure, when the report analysis model is trained based on the report analysis training set, the report analysis model may be trained by keeping the first weight matrix of the original structural layer in the pre-trained model unchanged and only training and updating the low-rank sub-weight matrix obtained by performing the dimensionality reduction processing on the second weight matrix, which eliminates the need to determine the gradient or maintain the optimizer state of most parameters and only needs to optimize the injected much smaller low-rank sub-weight matrix, thereby greatly reducing the count of parameters to be trained, lowering the memory requirement of the GPU, and saving arithmetic cost and time consumption. The pre-trained model may be fine-tuned, which may make the trained report analysis model more in line with the actual needs and take into account the respiratory treatment data obtained by the first subject using the respiratory device when answering the questions.
[0519] In some embodiments, the report analysis training sample may include a sample content item, sample respiratory treatment data, sample basic data of a sample first subject, and a sample respiratory trend. The report analysis training sample may be obtained based on historical data. The report analysis label may be the comprehensive recommendation of the second subject corresponding to the report analysis training sample and the diagnostic analysis corresponding to each content item in the report analysis training sample. The report analysis label may be determined based on a historical report analysis in historical data or by manual labeling (e.g., second subject labeling) .
[0520] In some embodiments, the preset fine-tuning algorithm may also include algorithms such as an adaptive fine-tuning or multi-task learning algorithm, which is not limited herein.
[0521] In some embodiments of the present disclosure, the report benchmark may be determined and the target report may be generated, which may make the target report sent to the user terminal conform to the report benchmark and better meet the actual needs of the user. The target report is generated using the large language model, which may obtain a target report that is more adaptable to the application scenario and meet the needs of the user more quickly in the case of limited training data.
[0522] FIG. 9 is schematic diagram illustrating an exemplary process of generating reminder information according to some embodiments of the present disclosure.
[0523] As illustrated in FIG. 9, in some embodiments, a cloud platform may generate a first determination result 930 based on a relationship between respiratory treatment data 910 and a first reminder indicator 920, and generate reminder information 940 based on the first determination result 930. More descriptions regarding the respiratory treatment data may be found in FIG. 2 and related descriptions thereof.
[0524] The first reminder indicator refers to data used to determine whether the respiratory treatment data of the first subject is abnormal.
[0525] In some embodiments, the first reminder indicator may be a difference threshold corresponding to a difference between the respiratory treatment data and reference respiratory treatment data. When the difference between the respiratory treatment data and the reference respiratory treatment data exceeds the difference threshold, it may indicate that the respiratory treatment data is abnormal. The reference respiratory treatment data refers to a reference value (e.g., a standard value or a desired value) of the respiratory treatment data. The reference respiratory treatment data may be preset by the system or preset manually. Different indicator data items in the respiratory treatment data may correspond to different reference respiratory treatment data and different first reminder indicators. For example, the first reminder indicator may include a difference threshold of the blood oxygen concentration, a difference threshold of the heart rate, a difference threshold of the respiratory frequency, a difference threshold of the blood pressure, etc.
[0526] The first determination result refers to a determination result indicating whether the respiratory treatment data is abnormal. The first determination result may include a determination result indicating ...
Claims
1.A method for respiratory treatment management, implemented by a cloud platform, the method comprising:obtaining, based on a user terminal, at least one of user basic data, basic data of one or more respiratory devices, usage data of the one or more respiratory devices, or respiratory treatment data of a first subject; wherein the user basic data includes basic data of the first subject or basic data of a second subject, the second subject provides guidance or a suggestion regarding respiratory treatment to the first subject, and the respiratory treatment data is obtained from a first monitoring device; andgenerating target information based on at least one of the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data; andsending the target information to the user terminal.2.The method of claim 1, wherein the generating target information based on at least one of the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data includes:generating a display interface based on the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data; whereinthe display interface includes at least one of statistical information of the usage data of the one or more respiratory devices, statistical information of the respiratory treatment data, respiratory trend of the first subject, a recommended treatment plan of the first subject, and a supervisory relationship between the first subject and the second subject.3.The method of claim 2, wherein the display interface includes a plurality of sub-areas, the plurality of sub-areas correspond to different treatment parameters, and a display content of each of the plurality of sub-areas includes: a type of a treatment parameter and abnormal statistical data of the treatment parameter.4.The method of claim 2, wherein the respiratory trend of the first subject is determined through operations including:predicting, based on the basic data of the first subject, the respiratory treatment data, and physiological data of the first subject, the respiratory trend of the first subject through a first prediction model; wherein the first prediction model is a trained machine learning model, the physiological data is acquired by a second monitoring device, and the second monitoring device is built into the user terminal or connected with a user terminal network.5.The method of claim 2, wherein the user terminal includes a second terminal or a first terminal, and the recommended treatment plan of the first subject is determined through operations including:determining a first candidate treatment plan;determining, based on the respiratory trend of the first subject and the first candidate treatment plan, a predicted treatment result corresponding to the first candidate treatment plan through a second prediction model, wherein the second prediction model is a trained machine learning model;determining a second candidate treatment plan based on the predicted treatment result and the first candidate treatment plan;sending the second candidate treatment plan and a predicted treatment result corresponding to the second candidate treatment plan to the second terminal; andin response to receiving a first recommended instruction fed back by the second terminal, generating the recommended treatment plan.6.The method of claim 1, wherein the generating target information based on the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data includes:in response to receiving a report generation instruction from the user terminal,determining a report benchmark including a report content item and a report parameter; wherein the report benchmark is determined based on the user basic data or the respiratory treatment data; andgenerating a target report meeting the report benchmark.7.The method of claim 1, wherein the generating target information based on the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data includes:generating a first result based on a relationship between the respiratory treatment data and a first index; wherein the first index is determined based on at least one of the basic data, posture data, and environmental data of the first subject; andgenerating reminder information based on the first result.8.The method of claim 7, wherein the user terminal includes a second terminal and a first terminal, and the generating reminder information based on the first result includes:determining a type of a reason causing an abnormality of the respiratory treatment data;in response to a determination that the type of the reason causing the abnormality of the respiratory treatment data meets a condition, generating first reminder information based on the first result and sending the first reminder information to the first terminal; andin response to a determination that the type of the reason causing the abnormality of the respiratory treatment data does not meet the condition, generating second reminder information based on the first result and sending the second reminder information to the second terminal and the first terminal.9.The method of claim 1, wherein the generating target information based on the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data includes:generating, based on the user basic data, the basic data of the one or more respiratory devices, and historical respiratory treatment data of the first subject, a recommended setting parameter for the one or more respiratory devices; andsending, through the user terminal, a control instruction to the one or more respiratory devices, wherein the control instruction includes the recommended setting parameter.10.The method of claim 9, wherein the generating, based on the user basic data, the basic data of the one or more respiratory devices, and historical treatment data of the first subject, a recommended setting parameter for the one or more respiratory devices includes:determining, based on the user basic data, the basic data of the one or more respiratory devices, and the historical respiratory treatment data, at least one candidate setting parameter through a third prediction model; wherein the third prediction model is a trained machine learning model;sending the at least one candidate setting parameter to the second terminal through a network; andgenerating the recommended setting parameter based on a second recommendation instruction fed back by the second terminal.11.The method of claim 10, wherein the one or more respiratory devices include a respiratory primary device and a respiratory auxiliary device, and the recommended setting parameter includes a first setting parameter of the respiratory primary device, a second setting parameter of the respiratory auxiliary device, and an associated setting parameter of the respiratory primary device and the respiratory auxiliary device.12.The method of claim 1, wherein the generating target information based on the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data includes:determining a quality of the respiratory treatment data; andin response to a determination that the quality does not meet a quality condition, correcting the respiratory treatment data based on historical respiratory treatment data of the first subject to generate corrected respiratory treatment data.13.The method of claim 12, wherein the correcting the respiratory treatment data based on historical respiratory treatment data of the first subject includes:generating the corrected respiratory treatment data by processing a historical treatment sequence based on a correction model, wherein the correction model is a sequence model, and the historical treatment sequence is constructed based on the historical respiratory treatment data according to a preset time rule.14.The method of claim 1, wherein the generating target information based on the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data includes:determining, based on the historical respiratory treatment data and a respiratory trend of the first subject, acquisition parameters of different treatment parameters.15.The method of claim 14, wherein the determining, based on the historical respiratory treatment data and a respiratory trend of the first subject, acquisition parameters of different treatment parameters includes:predicting, based on the historical respiratory treatment data, a predicted acquisition data sequence of the different treatment parameters through a fourth prediction model, wherein the fourth prediction model is a sequence model;determining priorities of the different treatment parameters based on the predicted acquisition data sequence; anddetermining the acquisition parameters of the different treatment parameters based on the priorities of the different treatment parameters.16.A system for respiratory treatment management, comprising a cloud platform, a user terminal, one or more respiratory devices, and a first monitoring device, the one or more respiratory devices and the first monitoring device connected with the user terminal, and the user terminal connected with the cloud platform, whereinthe one or more respiratory devices are configured to acquire basic data of the one or more respiratory devices and usage data of the one or more respiratory devices, and transmit the basic data of the one or more respiratory devices and the usage data of the one or more respiratory devices to the user terminal through a network;the first monitoring device is configured to acquire respiratory treatment data of a first subject and transmit the respiratory treatment data to the user terminal;the user terminal is configured to acquire user basic data and transmit the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, the respiratory treatment data, and the user basic data to the cloud platform; andthe cloud platform is configured to generate target information based on at least one of the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data and send the target information to the user terminal.17.The system of claim 16, further comprising a second monitoring device, whereinthe second monitoring device is configured to acquire physiological data of the first subject, andthe second monitoring device is connected with the user terminal or built into the user terminal.18.The system of claim 16, wherein the cloud platform is configured to:generate a display interface based on the user basic data, the basic data of the one or more respiratory devices, the usage data of the one or more respiratory devices, or the respiratory treatment data; whereinthe display interface includes at least one of statistical information of the usage data of the one or more respiratory devices, statistical information of the respiratory treatment data, a respiratory trend of the first subject, a recommended treatment plan of the first subject, and a supervisory relationship between the first subject and the second subject.19.The system of claim 16, wherein the cloud platform is configured to:generate, based on the user basic data, the basic data of the one or more respiratory devices, and historical respiratory treatment data of the first subject, a recommended setting parameter for the one or more respiratory devices; andsend, through the user terminal, a control instruction to the one or more respiratory devices, wherein the control instruction includes the recommended setting parameter.20.A non-transitory computer-readable storage medium storing computer instructions, wherein after reading the computer instructions in the storage medium, a computer executes the method for respiratory treatment management of any one of claims 1-15.
Citation Information
Patent Citations
Radiotherapy-based plan parameter prediction device
CN116072263A
System and method for determining use of respiratory therapy system
CN116528751A
Fishpond monitoring equipment data automatic processing method, device, equipment and medium
CN116756136A
Predicting a respiratory event based on trend information, and applications thereof
US10709414B1
Methods and systems for treating cancer and predicting and optimizing treatment outcomes in individual cancer patients
US20210249132A1
Cited By
Medical subject knowledge service-oriented data processing method and electronic equipment
CN121615743A