A Method and System for Intelligent Monitoring of Remote Nursing Data

Abstract

The present invention provides an intelligent monitoring method and system for remote care data, which relates to the technical field of care monitoring. The method includes: generating a prompt message at the start moment of a monitoring cycle; testing a wearable monitoring device to determine whether it is worn correctly; if it is worn correctly, obtaining user precautions from a background server; determining whether the care status information is abnormal according to the user precautions and the monitoring data obtained by the wearable monitoring device; if the care status information is abnormal, generating a warning message. According to the present invention, after the user wears the wearable monitoring device, the user's activities can be monitored, and the type of user activities that need to be monitored can be judged according to the user precautions, so as to judge whether the user's activities are abnormal and whether a dangerous situation may occur based on the monitoring results. A warning message can also be generated to prompt the user's guardians or medical staff to pay attention to the patient, improving the safety of the user's activities and the accuracy of monitoring.

Description

Technical Field

[0001] The present invention relates to the technical field of nursing monitoring, and particularly to a method and system for intelligent monitoring of remote nursing data. Background Art

[0002] In the related art, CN119581034A discloses a remote monitoring nursing warning system based on deep learning. This solution relates to the technical field of remote monitoring and warning, and includes a remote monitoring nursing warning platform. The remote monitoring nursing warning platform is communicatively connected to a data acquisition module, a feature extraction module, a nursing warning evaluation module, a warning grading response module, and a user interaction module. Among them, the modules are electrically connected to each other; the data acquisition module is used to collect the physiological data set of patients in real time through remote monitoring tools. This remote monitoring nursing warning system based on deep learning analyzes and processes a large amount of patient physiological data through the application of deep learning technology, and by learning normal physiological patterns and identifying abnormal patterns, it mines potential rules and patterns from a large amount of data, analyzes the changes in the health status of patients, improves the accuracy of patient nursing monitoring, better captures the subtle changes in the patient's condition, and thus realizes earlier intervention and treatment.

[0003] CN119480154A relates to the technical field of data processing and discloses an esophageal cancer patient nursing remote monitoring system and method that can improve the nursing level. The system includes a data acquisition module, a personalized data transmission module, a remote monitoring center, and a remote service module; the personalized data transmission module is used to predict the potential key data of patients using a collaborative recommendation algorithm, and assign corresponding personalized transmission parameters to the potential key data and non-potential key data in the monitoring data. At the same time, based on 5G communication technology and the corresponding personalized transmission parameters, the potential key data and non-potential key data are respectively transmitted to the remote monitoring center; the remote monitoring center is used to receive the monitoring data and remotely monitor and analyze the nursing status according to the monitoring data. This solution can not only reduce the medical burden of patients, but also enable medical staff to obtain key data in the first time, formulate more appropriate treatment plans for patients, and thus improve the overall medical nursing level.

[0004] Therefore, in the related art, although it is possible to remotely monitor the physiological data of users and judge the health status of users, it is impossible to analyze data such as the daily activity data and movement postures of users, it is difficult to monitor the activities of postoperative users, and it is also difficult to judge whether there are abnormal or dangerous conditions during the activities of users. Summary of the Invention

[0005] The present invention provides a method and system for intelligent monitoring of remote nursing data, which can solve the technical problems that the related art cannot monitor the activities of postoperative users and it is also difficult to judge whether there are abnormal or dangerous conditions during the activities.

[0006] According to a first aspect of the present invention, there is provided an intelligent monitoring method for remote care data, including:

[0007] At the start moment of the monitoring period, generate a prompt message, wherein the prompt message is used to prompt the user to wear the wearable monitoring device;

[0008] After receiving the instruction that the user has finished wearing, test the wearable monitoring device to determine whether it is worn correctly;

[0009] When worn correctly, obtain user precautions from the background server;

[0010] According to the user precautions and the monitoring data obtained by the wearable monitoring device at each monitoring moment, determine whether the care status information is abnormal;

[0011] If the care status information is abnormal, generate a warning message and send it to the background server.

[0012] According to a second aspect of the present invention, there is provided an intelligent monitoring system for remote care data, including:

[0013] A prompt module, configured to generate a prompt message at the start moment of the monitoring period, wherein the prompt message is used to prompt the user to wear the wearable monitoring device;

[0014] A test module, configured to test the wearable monitoring device after receiving the instruction that the user has finished wearing, to determine whether it is worn correctly;

[0015] A precautions module, configured to obtain user precautions from the background server when worn correctly;

[0016] A care status information module, configured to determine whether the care status information is abnormal according to the user precautions and the monitoring data obtained by the wearable monitoring device at each monitoring moment;

[0017] A warning module, configured to generate a warning message and send it to the background server if the care status information is abnormal.

[0018] By adopting the above technical solutions, the present invention can achieve the following technical effects:

[0019] According to the present invention, after the user wears the wearable monitoring device, the user's activities can be monitored, and the type of user activities to be monitored can be determined according to the user's precautions. Then, based on the monitoring results, it can be determined whether the user's activities are abnormal and whether a dangerous situation may occur. A warning message can also be generated to prompt the caregiver or medical staff of the user to pay attention to the patient, improving the safety of the user's activities, as well as the accuracy and flexibility of the monitoring. When determining the wearing correctness score, when the user first wears the motion monitoring device, the activity range of the motion monitoring device can be determined through the activities of the user's limbs, and then the reference activity surface can be determined. When the user wears the motion monitoring device subsequently, the position error between the motion monitoring device and the reference activity surface can be determined, so as to determine the wearing correctness score of the motion monitoring device, and thus determine whether the motion monitoring device is worn correctly, providing an accurate data basis for using the motion monitoring device to monitor the user's actions and postures subsequently. When training the condition determination model, the error of the training category information can be determined through the cross-entropy loss function, and the error of the training duration judgment threshold and the training heart rate judgment threshold in two cases where the training category information is correct and incorrect can be determined through the conditional function, so as to improve the training intensity and training efficiency. Further, the reference value of the data of the sample user can be determined through the relative gap between the total recovery duration of the sample user and the postoperative recovery duration, so as to improve the objectivity of the loss function, enhance the pertinence of the training, and effectively improve the performance of the condition determination model. When determining the motion status score, the probability that the user is performing strenuous exercise can be determined through the relative gap between the average heart rate during the detection period and the heart rate at rest, and the probability that the user is performing strenuous exercise can also be determined through the maximum frequency of large-amplitude movements of the user's limbs. Thus, the probability that the user is performing strenuous exercise can be comprehensively determined from two aspects of heart rate and movement frequency, improving the accuracy and objectivity of the motion status score. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 Exemplarily shows a schematic flowchart of a remote care data intelligent monitoring method according to an embodiment of the present invention;

[0021] Figure 2 Exemplarily shows a block diagram of a remote care data intelligent monitoring system according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] The technical solutions of the present invention will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.

[0023] Figure 1 Exemplarily shows a schematic flowchart of a remote care data intelligent monitoring method according to an embodiment of the present invention, and the method includes:

[0024] Step S101, at the start moment of the monitoring period, generate a prompt message, where the prompt message is used to prompt the user to wear the wearable monitoring device.

[0025] Step S102, after receiving the instruction that the user has finished wearing, test the wearable monitoring device to determine whether it is worn correctly.

[0026] Step S103, when worn correctly, obtain user precautions from the background server.

[0027] Step S104, based on the user precautions and the monitoring data obtained by the wearable monitoring device at each monitoring moment, determine whether the nursing status information is abnormal.

[0028] Step S105, if the nursing status information is abnormal, generate a warning message and send it to the background server.

[0029] According to the remote nursing data intelligent monitoring method of an embodiment of the present invention, after the user finishes wearing the wearable monitoring device, the user's activities can be monitored, and the type of user activities that need to be monitored can be judged according to the user precautions. Thus, based on the monitoring results, it can be judged whether the user's activities are abnormal and whether a dangerous situation may occur. A warning message can also be generated to prompt the user's guardians or medical staff to pay attention to the patient, improving the safety of the user's activities as well as the accuracy and flexibility of the monitoring.

[0030] According to an embodiment of the present invention, in step S101, the duration of each monitoring period can be one day. A certain moment in a day (for example, 9:00 am) can be set as the start moment of the monitoring period. At the start moment of the monitoring period, a prompt message can be generated to prompt the user to wear the wearable monitoring device properly, so that when the user is performing activities, information such as the user's movements and postures can be monitored to determine whether there are risks in the user's activities. For example, the user is a user who needs to perform rehabilitation activities after surgery. This user cannot perform strenuous exercise, cannot sit for a long time, and should perform simple activities such as taking a walk. The wearable monitoring device can monitor whether the user's activities belong to activities that are not conducive to rehabilitation such as strenuous exercise and sitting for a long time. The wearable monitoring device can include monitoring devices for multiple parts. For example, the wearable monitoring device includes a motion monitoring device for limb wearing and a heart rate monitoring device for monitoring the heart rate. The motion monitoring device can include a neck monitoring device, a waist monitoring device, a knee monitoring device, an elbow monitoring device, etc. Inertial measurement units can be set in each monitoring device to monitor the position of each monitoring device, and then determine the relative position relationship between the various monitoring devices, thereby determining the user's posture. The heart rate monitoring device can be used to detect the user's heart rate to determine whether the user is performing strenuous exercise.

[0031] According to an embodiment of the present invention, in step S102, after the user finishes wearing, a wearing completion instruction can be input, and the wearable monitoring device can determine whether the user is wearing correctly. After determining that the user is wearing correctly, the device starts to monitor the user's movements and postures.

[0032] According to an embodiment of the present invention, after receiving the instruction that the user has finished wearing, the wearable monitoring device is tested to determine whether it is worn correctly, including: when the user first wears the motion monitoring device, obtaining the reference relative position relationship of each motion monitoring device; according to the reference relative position relationship and the verified relative position relationship of each motion monitoring device when the instruction that the user has finished wearing is received, determining whether the motion monitoring device is worn correctly; according to the verified heart rate data detected by the heart rate monitoring device within the first verification time period, determining whether the heart rate monitoring device is worn correctly; when both the motion monitoring device and the heart rate monitoring device are worn correctly, determining that the wearable monitoring device is worn correctly.

[0033] According to an embodiment of the present invention, when the user first wears the motion monitoring device, a professional such as a medical staff can assist the user in wearing, and record the reference relative position relationship of each motion monitoring device after correct wearing as the correct relative position relationship for comparison with the relative position relationship determined by the user himself when wearing later. If the relative position relationship of each motion monitoring device determined by the user himself when wearing is the same as or close to the reference relative position relationship, it means that the user is wearing correctly.

[0034] According to an embodiment of the present invention, when the user first wears the motion monitoring device, the user may be in a certain posture. However, when the user wears the motion monitoring device later, the user may be in another posture. The user's wearing method may be correct, but due to the difference in postures, the obtained relative position relationship may not be consistent with the reference relative position relationship, resulting in a misjudgment that the user is not wearing correctly. To reduce the possibility of such misjudgment, the activity range of each motion monitoring device can be set based on the reference relative position relationship, and when the relative position relationship obtained when the user wears later is within or close to the activity range, it is determined that the user is wearing correctly.

[0035] According to an embodiment of the present invention, determining whether the motion monitoring devices are worn correctly based on the reference relative position relationship and the verified relative position relationships of the respective motion monitoring devices when receiving the instruction that the user has completed wearing includes: setting a reference motion monitoring device, and determining the reference distances between other motion monitoring devices and the reference motion monitoring device according to the reference relative position relationship; determining the reference activity surfaces of other motion monitoring devices according to the reference distances; determining whether other motion monitoring devices are on the corresponding reference activity surfaces according to the verified relative position relationships; if the i-th motion monitoring device is on the corresponding reference activity surface, the wearing correctness score of the i-th motion monitoring device is 1; otherwise, determining the reference point closest to the i-th motion monitoring device on the reference activity surface corresponding to the i-th motion monitoring device; determining the wearing correctness score of the i-th motion monitoring device according to the reference point and the position of the i-th motion monitoring device; and determining that the motion monitoring devices are worn correctly when the minimum value of the wearing correctness scores of the respective motion monitoring devices is greater than or equal to a preset correctness score threshold.

[0036] According to an embodiment of the present invention, a reference motion monitoring device can be set. For example, the neck monitoring device and the waist monitoring device can be set as the reference motion monitoring devices, and the reference distances between other motion monitoring devices and the reference motion monitoring device are determined. For example, the reference distance between the elbow monitoring device and the neck monitoring device can be determined, and the reference distance between the knee monitoring device and the waist monitoring device can be determined. Based on this reference distance, the reference activity surfaces of the elbow monitoring device and the knee monitoring device can be set. For example, when the user first wears the motion monitoring devices, a spatial coordinate system is established with the neck monitoring device as the coordinate origin. The user can move the arms and legs to the maximum extent, and during the movement, the position coordinates of the elbow monitoring device and the knee monitoring device in the coordinate system are recorded. Then, based on these coordinates, fitting is performed to obtain the activity surfaces of the elbow monitoring device and the knee monitoring device, that is, the surfaces where the activity trajectories of the elbow monitoring device and the knee monitoring device are located. Further, the position coordinates of the elbow monitoring device and the knee monitoring device can be recorded when the movement amplitudes of the arms and legs reach the maximum in multiple directions, and based on these position coordinates, fitting is performed to obtain the activity range curve. The activity range curve is a closed curve on the activity surface, and the activity surface within the activity range curve is the reference activity surface, which can be used to describe the activity range of the elbow monitoring device and the knee monitoring device.

[0037] According to an embodiment of the present invention, when the user wears the motion monitoring device subsequently, it is possible to determine whether the user wears it correctly based on the verified relative position relationship of the motion detection devices and the reference activity surface. For example, after the user wears the motion monitoring device, a coordinate system can be established with the neck detection device as the coordinate origin, and the reference activity surfaces of each other motion detection device (for example, the elbow monitoring device and the knee monitoring device) can be obtained from the database, and the verified relative position relationship between the other motion monitoring devices and the reference motion monitoring device can be obtained, so as to determine the relationship between the position coordinates of the other motion monitoring devices in the coordinate system and the reference activity surface.

[0038] According to an embodiment of the present invention, if the i-th motion monitoring device is on the corresponding reference activity surface, the wearing correctness score of the i-th motion monitoring device is 1, that is, it is determined that the motion detection device is worn correctly.

[0039] According to an embodiment of the present invention, during the actual wearing process, there may be certain errors. For example, if the wearing position is slightly deviated from the initial wearing position, the position coordinates of the motion monitoring device in the coordinate system are not on the reference activity surface. However, in this case, the motion detection device can still achieve the purpose of monitoring the user's actions and postures. Therefore, although there is a slight deviation, the user is not wearing it incorrectly. To adapt to this situation, the position relationship between the reference activity surface and the motion detection device can be solved. If this position relationship indicates that the position deviation between the reference activity surface and the motion detection device is small, it can also be determined that the motion detection device is worn correctly. For example, the reference point closest to the motion monitoring device can be determined on the reference activity surface, and the wearing correctness score of the motion monitoring device can be determined based on the position error between the reference point and the motion monitoring device.

[0040] According to an embodiment of the present invention, determining the wearing correctness score of the i-th motion monitoring device according to the reference point and the position where the i-th motion monitoring device is located includes: determining the wearing correctness score of the i-th motion monitoring device according to formula (1) ,

[0041] (1)

[0042] Wherein, is the vector between the position where the i-th motion monitoring device is located and the position where the reference motion monitoring device is located, is the vector between the reference point corresponding to the i-th motion monitoring device and the position where the reference motion monitoring device is located.

[0043] According to an embodiment of the present invention, in formula (1), is the modulus of the vector between the location of the motion monitoring device and the reference point, which can represent the position error between the location of the motion monitoring device and the reference activity surface. Then it is the relative position error between the location of the motion monitoring device and the reference activity surface. It can represent the wearing correctness score of the motion monitoring device. The closer this score is to 1, the smaller the relative position error, and the closer the location of the motion monitoring device is to the reference activity surface.

[0044] According to an embodiment of the present invention, when the minimum value of the wearing correctness scores of each motion monitoring device is greater than or equal to a preset correctness score threshold, it is determined that the motion monitoring device is worn correctly. In other words, when the wearing correctness scores of all motion monitoring devices are higher than the preset correctness score threshold (for example, 0.9), it can be determined that the motion monitoring device is worn correctly.

[0045] In this way, when the user first wears the motion monitoring device, the activity range of the motion monitoring device can be determined through the activities of the user's limbs, and then the reference activity surface can be determined. And when the user wears the motion monitoring device subsequently, the position error between the motion monitoring device and the reference activity surface can be determined, so as to determine the wearing correctness score of the motion monitoring device, and thus judge whether the motion monitoring device is worn correctly, providing an accurate data basis for subsequent use of the motion monitoring device to monitor the user's movements and postures.

[0046] According to an embodiment of the present invention, in addition to the motion detection device, it can also be determined whether the heart rate monitoring device is worn correctly. Verification heart rate data can be obtained within the first verification time period. If the verification heart rate data can be obtained normally and there is no situation where data cannot be obtained, it can be determined that the heart rate monitoring device is worn correctly. If both the motion monitoring device and the heart rate monitoring device are worn correctly, it is determined that the wearable monitoring device is worn correctly.

[0047] According to an embodiment of the present invention, in step S103, if the wearable monitoring device is worn correctly, user precautions can be obtained from the background server. The user can be a patient in the postoperative rehabilitation period. Various precautions of the user during the rehabilitation period can be stored in the hospital's background server. The user precautions can be filled in by medical staff and saved in the background server. The user precautions are in the form of text content, rather than code form, so as to facilitate the medical staff to fill in.

[0048] According to an embodiment of the present invention, in step S104, the user precautions can be analyzed, and the monitoring data of the wearable monitoring device can be analyzed, so as to determine whether the user's movements, postures, etc. match the items that the user is not suitable for described in the user precautions, and thus determine whether the nursing status information is abnormal.

[0049] According to an embodiment of the present invention, determining whether the nursing status information is abnormal based on the user's precautions and the monitoring data obtained by the wearable monitoring device at each monitoring moment includes: inputting the user's precautions into a trained natural language processing model to obtain semantic information of the user's precautions; processing the semantic information of the user's precautions and the user's postoperative recovery duration through a trained condition determination model to determine the achievement condition judgment threshold of the user's precautions; and determining whether the nursing status information is abnormal according to the monitoring data and the achievement condition judgment threshold.

[0050] According to an embodiment of the present invention, the natural language processing model can be a recurrent neural network model, etc. The present invention does not limit the type of the natural language processing model. The natural language processing model can obtain the semantic information of the user's precautions. If there are multiple user's precautions, the semantic information of multiple user's precautions can be obtained through the natural language processing model, and the semantic information is information in vector form.

[0051] According to an embodiment of the present invention, the condition determination model can be a BP neural network model, which can process the semantic information of the user's precautions and the user's postoperative recovery duration to obtain the achievement condition judgment threshold of the user's precautions. For example, if the user exercises vigorously for 10 minutes, it can be determined that the user meets the condition of "vigorous exercise" in the user's precautions, and 10 minutes is the achievement condition judgment threshold of this user's precautions.

[0052] According to an embodiment of the present invention, the training steps of the condition determination model include: inputting the semantic information of the user's precautions and the training recovery duration of the sample user into the condition determination model to obtain the training duration judgment threshold and the training heart rate judgment threshold; determining the training category information of the training duration judgment threshold; determining the loss function of the condition determination model according to the training duration judgment threshold, the training heart rate judgment threshold, the training category information, and the training annotation information; and training the condition determination model according to the loss function of the condition determination model to obtain the trained condition determination model.

[0053] According to an embodiment of the present invention, the training category information of the training duration judgment threshold can be used to indicate whether the threshold is an upper limit threshold or a lower limit threshold. In the example, if the threshold is an upper limit threshold, and the duration of a certain exercise performed by the user is lower than the duration represented by the upper limit threshold, it can indicate that the user has not performed abnormal exercise, that is, the nursing status information is normal. For example, the upper limit threshold for strenuous exercise is 10 minutes, and the duration of the user's strenuous exercise is 5 minutes, which can indicate that the user has not performed strenuous exercise and the nursing status information is normal. If the threshold is a lower limit threshold, and the duration of a certain exercise performed by the user is lower than the lower limit threshold, it can indicate that the nursing status information of the user is abnormal. For example, within a monitoring period, the lower limit threshold for the total duration of the user's walking is 1 hour. If the total duration of the user's walking within a monitoring period does not reach 1 hour, the nursing status information is abnormal.

[0054] According to an embodiment of the present invention, based on the training duration judgment threshold, the training heart rate judgment threshold, the training category information, and the training annotation information, the loss function of the conditional determination model is determined, including: determining the loss function LOSS of the conditional determination model according to formula (2),

[0055] (2)

[0056] Wherein, is the probability that the training category information of the training duration judgment threshold determined according to the training annotation information of the j-th sample user is of the upper limit type, is the probability that the training category information of the training duration judgment threshold determined by the conditional determination model is of the upper limit type, is the training duration judgment threshold of the j-th sample user, is the annotation duration judgment threshold determined according to the training annotation information of the j-th sample user, is the training heart rate judgment threshold of the j-th sample user, is the annotation heart rate judgment threshold determined according to the training annotation information of the j-th sample user, if is a conditional function, is the total recovery duration of the j-th sample user, is the postoperative recovery duration of the j-th sample user, 、 and are preset weights, m is the number of sample users, j ≤ m, and both j and m are positive integers.

[0057] According to an embodiment of the present invention, in formula (2), is the cross-entropy loss function regarding the training category information. Among them, if the training category information determined based on the training annotation information is of the upper limit type, then , ,and this cross-entropy loss function is During the training process, it can make close to 1, thereby reducing the cross-entropy loss function to improve the accuracy of the training class information output by the conditional determination model. If the training class information determined based on the training annotation information is of the lower limit type, then , , the cross-entropy loss function is During the training process, it can make close to 0, thereby reducing the cross-entropy loss function to improve the accuracy of the training class information output by the conditional determination model. Among them, the training annotation information is the correct annotation made by medical staff, which can represent the correct duration judgment threshold, the correct heart rate judgment threshold, and the correct threshold type.

[0058] According to an embodiment of the present invention, is a conditional function, indicating that in the case of , the conditional function value is 1, otherwise it is . Among them, if , then the gap between the training class information and the annotation is too large and incorrect, indicating that the training class information is incorrect. In this case, the conditional function value can be directly determined to be 1 to increase the loss function value and enhance the training intensity. In the case of , it indicates that the training class information is correct. A less than 1 can be used as the conditional function value, and the meaning of this value is the relative error between the training duration judgment threshold and the annotation duration judgment threshold. During the training process, this conditional function value can be reduced to reduce the relative error between the training duration judgment threshold and the annotation duration judgment threshold and improve the accuracy of the training duration judgment threshold.

[0059] According to an embodiment of the present invention, is a conditional function, indicating that in the case of , the conditional function value is 1, otherwise it is . Among them, if , then the gap between the training class information and the annotation is too large and incorrect, indicating that the training class information is incorrect. In this case, the conditional function value can be directly determined to be 1 to increase the loss function value and enhance the training intensity. In the case of , it indicates that the training class information is correct. A As a conditional function value, the meaning of this value is the relative error between the marked heart rate judgment threshold and the training heart rate judgment threshold. If the category error between the training duration judgment threshold in the training annotation information and the training duration judgment threshold generated by the conditional determination model is large, that is, the training category information is incorrect, it means that the conditional determination model fails to accurately understand the meaning represented by the semantic information of the user's precautions. For example, it fails to understand the meaning of strenuous exercise, resulting in the category information of the training duration judgment threshold being determined as the lower limit threshold. In this case, it is also difficult for the conditional determination model to generate an accurate heart rate judgment threshold, that is, it cannot understand the difference between strenuous exercise and walking, thus unable to distinguish the heart rate gap between strenuous exercise and walking, and thus unable to generate an accurate heart rate judgment threshold. In this case, it is directly considered that the training category information is incorrect, and the upper limit 1 of the relative error is used as the conditional function value.

[0060] According to an embodiment of the present invention, the above three items can be weighted and summed as the error of the output of the conditional determination model. It represents the relative gap between the total recovery duration of the sample user and the postoperative recovery duration of the sample user. The smaller the gap between the postoperative recovery duration and the total recovery duration, the closer the sample user is to complete recovery, the smaller the restrictive effect of the precautions on it, and the higher the probability that the sample user violates the user's precautions. Then the reference value of the training annotation information of this sample user is lower. Therefore, a lower weight can be assigned to the data of this sample user through this relative gap. On the contrary, the larger the relative gap between the total recovery duration of the sample user and the postoperative recovery duration of the sample user, the farther the sample user is from complete recovery, the greater the restrictive effect of the precautions on it, and the greater the reference value of the training annotation information of this sample user. A higher weight can be assigned through this relative gap.

[0061] According to an embodiment of the present invention, by weighted summing the error of the output of the conditional determination model through the above weights, the loss function of the conditional determination model can be obtained. The conditional determination model can be trained by the method of backpropagation, and after multiple trainings, the trained conditional determination model can be obtained.

[0062] In this way, the error of the training category information can be determined through the cross-entropy loss function, and the errors of the training duration judgment threshold and the training heart rate judgment threshold in the two cases of correct and incorrect training category information can be determined through the conditional function, so as to improve the training intensity and training efficiency. Further, the reference value of the data of the sample user can also be determined through the relative gap between the total recovery duration of the sample user and the postoperative recovery duration, thereby improving the objectivity of the loss function, enhancing the pertinence of training, and effectively improving the performance of the conditional determination model.

[0063] According to an embodiment of the present invention, after obtaining the achievement condition judgment threshold of the user's precautions through the trained conditional determination model, it is possible to determine whether the user reaches the achievement condition judgment threshold through the monitoring data to determine whether the nursing status information is abnormal.

[0064] According to an embodiment of the present invention, determining whether the nursing status information is abnormal according to the monitoring data and the achievement condition judgment threshold includes: setting a first activity range in each reference activity surface; determining the intersection position of the connection line between the position where the motion monitoring device is located and the reference motion monitoring device on the reference activity surface; using the moment when the intersection position of at least one motion monitoring device first leaves the first activity range as the start time of timing; when the intersection positions of all motion monitoring devices do not cross the boundary of the first activity range within a first preset time period, recording the end time of timing; determining whether the nursing status information is abnormal according to the number of times the intersection of each motion monitoring device crosses the boundary of the first activity range during the detection time period between the start time of timing and the end time of timing, the heart rate at multiple moments during the detection time period, and the achievement condition judgment threshold.

[0065] According to an embodiment of the present invention, taking the item of "no strenuous exercise" in the user's precautions as an example, a first activity range can be set in the reference activity surface. The first activity range can represent a range with a relatively small amplitude of limb movement. For example, the activity range of the knee monitoring device during the leg-lifting action of normal walking and the activity range of the elbow detection device when normally picking up an item.

[0066] According to an embodiment of the present invention, since the position where the motion detection device is located may not be on the reference activity surface, the connection line between the position where the motion monitoring device is located and the reference motion monitoring device can be determined, and the intersection position of this connection line and the reference activity surface can be determined. This position can be used to represent the projection of the motion detection device on the reference activity surface. Also, through the relationship between this intersection position and the first activity range, it can be determined whether the user's movement amplitude is too large. For example, if the intersection position is outside the first activity range, it can indicate that the user's movement amplitude is large.

[0067] According to an embodiment of the present invention, if the user only makes large-amplitude movements occasionally, for example, stretching the body may cause a large amplitude of the arm to be lifted, but not strenuous exercise. Therefore, it is possible to determine whether the user has a behavior of making large-amplitude movements with a high frequency to determine whether the user is doing strenuous exercise. The moment when the intersection position of at least one motion monitoring device first leaves the first activity range is used as the start time of timing, and the moment when the intersection positions of all motion monitoring devices do not cross the boundary of the first activity range within the first preset time period (for example, the limbs maintain a posture with a small movement amplitude for the first preset time period) is used as the end time of timing. It is possible to determine the number of times the intersection of each motion monitoring device crosses the boundary of the first activity range within the detection time period between the start time and the end time of timing, so as to determine whether the frequency of making large-amplitude movements is high. If the user is only stretching the body, the duration of the detection time period is short, and the frequency of making large-amplitude movements is low, and it will not be determined as strenuous exercise.

[0068] According to an embodiment of the present invention, based on the number of times the intersection of each motion monitoring device crosses the boundary of the first activity range within the detection time period between the start time and the end time of timing, the heart rate at multiple moments within the detection time period, and the achievement condition judgment threshold, it is determined whether the nursing status information is abnormal, including: determining the exercise status score S according to formula (3),

[0069] (3)

[0070] wherein, is the heart rate at the k-th moment within the detection time period, is the heart rate threshold, is the maximum value of the number of times the intersection of each motion monitoring device crosses the boundary of the first activity range, N is the number of moments within the detection time period, and if is a conditional function; if the exercise status score is greater than or equal to the strenuous exercise judgment threshold and the duration of the detection time period is greater than or equal to the strenuous exercise duration, it is determined that the nursing status information is abnormal.

[0071] According to an embodiment of the present invention, in formula (3), is a conditional function, indicating that in the case of , the value of the conditional function is , otherwise it is 0. is the average heart rate at multiple moments within the detection time period. In the case of strenuous exercise, the average heart rate is high. is the heart rate threshold. For example, it is the average value of the heart rate measured at multiple moments when the user is relatively calm, or it can be slightly higher than the average value of the heart rate measured at multiple moments when the user is relatively calm. For example, the heart rate threshold can be set to 1.1 times the average value of the heart rate measured at multiple moments when the user is relatively calm, etc. If indicates that the user's heart rate is relatively high, higher than the heart rate level during calmness, and the user may be engaged in strenuous exercise. is the relative gap between the average heart rate at multiple moments within the detection time period and the heart rate threshold. The larger this relative gap, the higher the user's heart rate and the higher the likelihood of the user being engaged in strenuous exercise. If it indicates that the user's heart rate has not changed significantly compared to that during calmness, then the user is not engaged in strenuous exercise. Therefore, the conditional function value is 0.

[0072] According to an embodiment of the present invention, represents the maximum frequency of large-amplitude movements of the user's limbs. The larger this maximum frequency, the higher the probability that the user is engaged in strenuous exercise. Multiply the probability that the user is engaged in strenuous exercise determined based on the heart rate above and the probability that the user is engaged in strenuous exercise determined based on the movement frequency to obtain a motion status score. The higher the motion status score, the higher the probability that the user is engaged in strenuous exercise. If the motion status score is greater than or equal to the strenuous exercise judgment threshold and the duration of the detection time period is greater than or equal to the strenuous exercise duration (for example, 10 minutes), then it is determined that the nursing status information is abnormal.

[0073] In this way, the probability that the user is engaged in strenuous exercise can be determined by the relative gap between the average heart rate within the detection time period and the heart rate during calmness, and the probability that the user is engaged in strenuous exercise can also be determined by the maximum frequency of large-amplitude movements of the user's limbs. Thus, the probability that the user is engaged in strenuous exercise is comprehensively determined from two aspects of heart rate and movement frequency, improving the accuracy and objectivity of the motion status score.

[0074] According to an embodiment of the present invention, in step S105, if the nursing status information is abnormal, a warning message can be generated and sent to the background server, and the background server can convey the warning information to medical staff or the user's guardian, or can also convey it to the user himself, thereby prompting the user not to continue with abnormal actions and improving the user's rehabilitation quality.

[0075] The intelligent monitoring method for remote care data according to an embodiment of the present invention can monitor the activities of a user after the user wears the wearable monitoring device, and can determine the type of user activities to be monitored according to the user's precautions, so as to judge whether the user's activities are abnormal and whether a dangerous situation may occur based on the monitoring results. A warning message can also be generated to prompt the caregiver or medical staff of the user to pay attention to the patient, improving the safety of the user's activities as well as the accuracy and flexibility of the monitoring. When determining the wearing correctness score, when the user first wears the motion monitoring device, the activity range of the motion monitoring device can be determined through the activities of the user's limbs, and then the reference activity surface can be determined. When the user wears the motion monitoring device subsequently, the position error between the motion monitoring device and the reference activity surface can be determined, so as to determine the wearing correctness score of the motion monitoring device, thereby judging whether the motion monitoring device is worn correctly, providing an accurate data basis for subsequently using the motion monitoring device to monitor the user's movements and postures. When training the condition determination model, the cross-entropy loss function can be used to determine the error of the training category information, and the condition function can be used to determine the error of the training duration judgment threshold and the training heart rate judgment threshold in two cases where the training category information is correct and incorrect, so as to improve the training intensity and training efficiency. Further, the reference value of the data of the sample user can be determined through the relative gap between the total recovery duration of the sample user and the postoperative recovery duration, thereby improving the objectivity of the loss function, enhancing the pertinence of the training, and effectively improving the performance of the condition determination model. When determining the motion condition score, the probability that the user is performing strenuous exercise can be determined through the relative gap between the average heart rate during the detection time period and the heart rate at rest, and the probability that the user is performing strenuous exercise can also be determined through the maximum frequency of large-amplitude movements of the user's limbs. Thus, the probability that the user is performing strenuous exercise can be comprehensively determined from two aspects of heart rate and movement frequency, improving the accuracy and objectivity of the motion condition score.

[0076] Figure 2 Exemplarily shown is a block diagram of a remote care data intelligent monitoring system according to an embodiment of the present invention. The system includes:

[0077] A prompt module, configured to generate a prompt message at the start moment of the monitoring cycle, where the prompt message is used to prompt the user to wear the wearable monitoring device;

[0078] A test module, configured to test the wearable monitoring device after receiving the instruction that the user has completed wearing, to determine whether it is worn correctly;

[0079] A precautions module, configured to obtain the user's precautions from the background server when the wearing is correct;

[0080] A nursing status information module, configured to determine whether the nursing status information is abnormal according to the user precautions and the monitoring data obtained by the wearable monitoring device at each monitoring moment;

[0081] A warning module, configured to generate a warning message and send it to the background server if the nursing status information is abnormal.

[0082] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the drawings are only examples and do not limit the present invention. The object of the present invention has been fully and effectively achieved. The function and structural principle of the present invention have been shown and described in the embodiments. Without departing from the above principle, the embodiments of the present invention can have any deformation or modification.

[0083] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent monitoring method for remote care data, characterized in that, Including: At the start moment of the monitoring period, a prompt message is generated, where the prompt message is used to prompt the user to wear the wearable monitoring device; After receiving the instruction that the user has completed wearing, the wearable monitoring device is tested to determine whether it is worn correctly; When worn correctly, obtain the user's precautions from the background server; Based on the user's precautions and the monitoring data obtained by the wearable monitoring device at each monitoring moment, determine whether the nursing status information is abnormal; If the nursing status information is abnormal, generate a warning message and send it to the background server; Among them, based on the user's precautions and the monitoring data obtained by the wearable monitoring device at each monitoring moment, determining whether the nursing status information is abnormal includes: Input the user's precautions into the trained natural language processing model to obtain the semantic information of the user's precautions; Through the trained condition determination model, process the semantic information of the user's precautions and the user's postoperative recovery duration to determine the achievement condition judgment threshold of the user's precautions; Based on the monitoring data and the achievement condition judgment threshold, determine whether the nursing status information is abnormal; Among them, the training steps of the condition determination model include: Input the semantic information of the user's precautions and the training recovery duration of the sample user into the condition determination model to obtain the training duration judgment threshold and the training heart rate judgment threshold; Determine the training category information of the training duration judgment threshold; According to the formula Determine the loss function LOSS of the condition determination model, where, is the probability that the training category information of the training duration judgment threshold determined according to the training annotation information of the j-th sample user is the upper limit type, is the probability that the training category information of the training duration judgment threshold determined by the condition determination model is the upper limit type, is the training duration judgment threshold of the j-th sample user, is the annotation duration judgment threshold determined according to the training annotation information of the j-th sample user, is the training heart rate judgment threshold of the j-th sample user, is the annotation heart rate judgment threshold determined according to the training annotation information of the j-th sample user, if is a conditional function, is the total recovery duration of the j-th sample user, is the postoperative recovery duration of the j-th sample user, 、 and are preset weights, m is the number of sample users, j ≤ m, and both j and m are positive integers; Based on the loss function of the condition determination model, train the condition determination model to obtain the trained condition determination model.

2. The intelligent monitoring method for remote care data according to claim 1, wherein The wearable monitoring device includes a motion monitoring device for limb wearing and a heart rate monitoring device for monitoring heart rate; Among them, after receiving the instruction that the user has completed wearing, testing the wearable monitoring device to determine whether it is worn correctly includes: When the user first wears the motion monitoring device, obtain the reference relative position relationship of each motion monitoring device; Based on the reference relative position relationship and the verification relative position relationship of each motion monitoring device when receiving the instruction that the user has completed wearing, determine whether the motion monitoring device is worn correctly; Based on the verified heart rate data detected by the heart rate monitoring device within the first verification time period, determine whether the heart rate monitoring device is worn correctly; When both the motion monitoring device and the heart rate monitoring device are worn correctly, determine that the wearable monitoring device is worn correctly.

3. The intelligent monitoring method for remote care data according to claim 2, wherein Based on the reference relative position relationship and the verification relative position relationship of each motion monitoring device when receiving the instruction that the user has completed wearing, determining whether the motion monitoring device is worn correctly includes: Set a reference motion monitoring device, and based on the reference relative position relationship, determine the reference distance between other motion monitoring devices and the reference motion monitoring device; Based on the reference distance, determine the reference activity surface of other motion monitoring devices; Based on the verification relative position relationship, determine whether other motion monitoring devices are on the corresponding reference activity surface; If the i-th motion monitoring device is on the corresponding reference activity surface, the wearing correctness score of the i-th motion monitoring device is 1; Otherwise, determine the reference point closest to the i-th motion monitoring device on the reference activity surface corresponding to the i-th motion monitoring device; Determine the wearing correctness score of the i-th motion monitoring device according to the reference point and the location of the i-th motion monitoring device; When the minimum value of the wearing correctness scores of all motion monitoring devices is greater than or equal to the preset correctness score threshold, determine that the motion monitoring devices are worn correctly.

4. The intelligent monitoring method for remote care data according to claim 3, wherein Determine the wearing correctness score of the i-th motion monitoring device according to the reference point and the location of the i-th motion monitoring device, including: According to the formula Determine the wearing correctness score of the i-th motion monitoring device , where is the vector between the position of the i-th motion monitoring device and the position of the reference motion monitoring device, is the vector between the reference point corresponding to the i-th motion monitoring device and the position of the reference motion monitoring device.

5. The intelligent remote care data monitoring method according to claim 3, characterized in that Determine whether the nursing status information is abnormal according to the monitoring data and the achievement condition judgment threshold, including: Set a first activity range in each reference activity surface; Determine the intersection position of the line connecting the position of the motion monitoring device and the reference motion monitoring device on the reference activity surface; Take the moment when the intersection position of at least one motion monitoring device first leaves the first activity range as the start time of timing; When the intersection positions of all motion monitoring devices do not cross the boundary of the first activity range within the first preset time period, record the end time of timing; Determine whether the nursing status information is abnormal according to the number of times the intersection of each motion monitoring device crosses the boundary of the first activity range during the detection time period between the start time of timing and the end time of timing, the heart rates at multiple moments during the detection time period, and the achievement condition judgment threshold.

6. The intelligent monitoring method for remote care data according to claim 5, wherein Determine whether the nursing status information is abnormal according to the number of times the intersection of each motion monitoring device crosses the boundary of the first activity range during the detection time period between the start time of timing and the end time of timing, the heart rates at multiple moments during the detection time period, and the achievement condition judgment threshold, including: According to the formula Determine the motion status score S, where is the heart rate at the k-th moment within the detection time period, is the heart rate threshold, is the maximum value of the number of times the intersection of each motion monitoring device enters and exits the boundary of the first activity range, N is the number of moments within the detection time period, and if is the conditional function; If the motion status score is greater than or equal to the strenuous exercise judgment threshold and the duration of the detection time period is greater than or equal to the strenuous exercise duration, determine that the nursing status information is abnormal.

7. A remote care data intelligent monitoring system for performing the method according to any one of claims 1-6, characterized in that, Including: A prompt module for generating a prompt message at the start moment of the monitoring cycle, where the prompt message is used to prompt the user to wear the wearable monitoring device; A test module for testing the wearable monitoring device after receiving the instruction that the user has completed wearing to determine whether it is worn correctly; A precautions module for obtaining user precautions from the background server when worn correctly; A nursing status information module for determining whether the nursing status information is abnormal according to the user precautions and the monitoring data obtained by the wearable monitoring device at each monitoring moment; A warning module for generating a warning message and sending it to the background server if the nursing status information is abnormal.

Images