Intelligent extraction method and system for reference information of patient in recovery period

Abstract

The invention relates to the field of data processing, in particular to an intelligent extraction method and system for reference information of a patient in a recovery period, and the method comprises the steps: recognizing the motion condition and motion intensity of a user through motion data collected by an intelligent wearable device, and accumulating the current motion duration of the patient when the motion intensity reaches a certain value; meanwhile, collecting physical sign data of the user in real time, determining reference physical sign data matched with the patient from a physical sign reference database in combination with the current exercise intensity, and calculating the similarity between the actually measured physical sign data and the reference physical sign data. If the similarity is too low, it can be deduced that the patient possibly feels uncomfortable during movement, at the moment, a question answering process is executed through a pre-constructed question template, and a symptom entity possibly appearing during movement of the patient is found. Finally, the accumulated exercise duration, the symptom entity, the actually-measured sign data and the exercise intensity data serve as key data to be extracted and provided for doctors, and the doctors have related data to refer to and judge the recovery condition of the patient.

Description

[0001] This application is a divisional application filed for the Chinese application with application date of February 20, 2025, application number 202510191037.5, and invention name “A method and system for intelligent extraction of key patient information”. Technical Field

[0002] The present invention relates to the field of data processing, and in particular to a method and system for intelligently extracting reference information of a patient during recovery period. Background Art

[0003] Patients usually have an observation period after receiving treatment. For hospitalized patients, they will be observed by medical staff or professional instruments and equipment to assess their recovery.

[0004] However, for patients recovering at home, the only way to understand their recovery is through regular physical examinations and their own personal accounts. During this time, doctors are unable to observe the patient's physical signs to understand their progress. If the condition worsens, the lack of data collection prevents timely intervention, resulting in a prolonged recovery period or even a difficult recovery. Summary of the Invention

[0005] In view of this, an object of the present invention is to provide a method and system for intelligently extracting reference information of a patient during the recovery period, so as to solve the problems in the background technology.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] The present invention provides an intelligent extraction method for patient recovery period reference information, comprising the steps of:

[0008] Obtaining the patient's identity information, and the patient's measured vital sign data and motion data during a target time period, wherein the measured vital sign data and the motion data are collected by the patient's smart wearable device, and the target time period is a time period of a target duration before the current time point;

[0009] Calculating the patient's current exercise intensity based on the patient's identity information and the exercise data of the target time period; and when the current exercise intensity is greater than a set value, adding the target duration to the cumulative exercise duration of this exercise;

[0010] Determining target reference vital sign data of the patient within the target time period based on the patient's current exercise intensity, the patient's identity information, and a pre-built vital sign reference database, wherein the vital sign reference database includes reference vital sign data of multiple user groups performing exercises of various intensities;

[0011] Calculating the similarity between the measured physical sign data and the reference physical sign data; and executing a question-and-answer process with the patient when the similarity is less than a preset similarity threshold, wherein the question-and-answer process is used to extract symptom entities;

[0012] The key information of the patient's recovery period is constructed based on the patient's cumulative exercise time, the current exercise intensity, the measured physical sign data of the target time period, and the symptom entity.

[0013] In one embodiment of the present application, obtaining the patient's motion data during a target time period includes:

[0014] Acquire the current motion type and three-axis acceleration data collected by the smart wearable device, wherein the three-axis acceleration data includes the three-axis acceleration (X t ,Y t ,Z t );

[0015] The three-axis acceleration data are filtered respectively to obtain three-axis acceleration filtered data, wherein the three-axis acceleration filtered data includes the three-axis acceleration (X) at multiple time points after filtering. t ',Y t ',Z t ');

[0016] The three-axis acceleration (X t ',Y t ',Z t ') to combine and get the combined acceleration a t ;

[0017] The total acceleration a based on multiple time points t Draw the acceleration time domain variation waveform, extract all peak points in the acceleration time domain variation waveform, and calculate the standard deviation σ of all peak points;

[0018] Performing a fast Fourier transform on the acceleration time-domain variation waveform to obtain an acceleration frequency-domain variation waveform, and extracting a main frequency component f0 from the acceleration frequency-domain variation waveform;

[0019] Motion data is constructed based on the standard deviation σ and the main frequency component f0.

[0020] In one embodiment of the present application, calculating the patient's current exercise intensity based on the patient's identity information and the exercise data of the target time period includes:

[0021] Determining a target weight sequence based on the current exercise type, the patient's identity information, and a pre-constructed exercise type-weight sequence relationship table, wherein the exercise type-weight sequence relationship table includes weight sequences for various user groups when performing various exercise types;

[0022] The motion data of the target time period is weighted based on the target weight sequence to obtain a current motion intensity S, wherein the mathematical expression of the current motion intensity S is:

[0023] S=w1×σ+w2×f0

[0024] Where w1 is the first weight and w2 is the second weight.

[0025] In one embodiment of the present application, the process of constructing the exercise type-weight sequence relationship table includes:

[0026] Acquire a plurality of sample data, wherein the sample data includes user information of a normal user, exercise type, measured physical sign sample data and exercise sample data sampled when the normal user exercises within a target duration;

[0027] Labeling the exercise intensity of multiple sample data i based on the measured physical sign sample data, and normalizing the measured physical sign sample data to obtain multiple sample data i containing labels, wherein the measured physical sign sample data includes heart rate and blood pressure;

[0028] Dividing a plurality of sample data containing labels into a plurality of data units based on the user information;

[0029] Classifying the data unit based on the motion type to obtain a plurality of data subunits;

[0030] Clustering multiple sample data within the data subunit based on motion intensity to obtain multiple intensity clusters;

[0031] Calculate the mean value of motion intensity, the mean value of the standard deviation of the acceleration peak value, and the mean value of the main frequency component of the acceleration frequency domain variation waveform in each intensity cluster; and construct a mathematical expression for motion intensity based on the mean value of motion intensity, the mean value of the standard deviation of the acceleration peak value, and the mean value of the main frequency component of the acceleration frequency domain variation waveform:

[0032] S i =w1×σ i +w2×f 0i

[0033] Where S i is the mean motion intensity of the ith intensity cluster, σ i is the standard deviation of the acceleration peak of the i-th intensity cluster, f 0iis the mean value of the main frequency component of the acceleration frequency domain variation waveform of the i-th intensity cluster;

[0034] The mathematical expressions of any two exercise intensities are combined into an equation system, and the first weight w1 and the second weight w2 are solved. When any data subunit has multiple different first weights w1 or multiple different second weights w2, the multiple different first weights w1 or multiple different second weights w2 are averaged to obtain the updated first weight w1 and the updated second weight w2;

[0035] The motion type-weight sequence relationship table is constructed based on the first weight w1 and the second weight w2 of each data subunit.

[0036] In one embodiment of the present application, the process of constructing the vital sign reference database includes:

[0037] Clustering multiple sample data in each data subunit based on heart rate and blood pressure to obtain multiple physical sign clusters;

[0038] Eliminating sample data with abnormal exercise intensity from the physical sign cluster to obtain a screened physical sign cluster;

[0039] Calculating the exercise intensity mean A1, exercise intensity standard deviation σ1, heart rate mean A2, heart rate standard deviation σ2, blood pressure mean A3 and blood pressure standard deviation σ3 of the plurality of sample data in the screened physical sign cluster;

[0040] Based on the exercise intensity mean A1 and the exercise intensity standard deviation σ1, the exercise intensity range of the sign cluster is constructed (A1-n×σ1, A1+n×σ1); based on the heart rate mean A2 and the heart rate standard deviation σ2, the heart rate range of the sign cluster is constructed (A2-n×σ2, A2+n×σ2); and based on the blood pressure mean A3 and the blood pressure standard deviation σ3, the blood pressure range of the sign cluster is constructed (A3-n×σ3, A3+n×σ3), where n is a range adjustment parameter;

[0041] A physical sign reference database is constructed based on heart rate ranges and blood pressure ranges in multiple exercise intensity ranges.

[0042] In one embodiment of the present application, the measured vital sign data includes measured heart rate and measured blood pressure at multiple time points, and calculating the similarity between the measured vital sign data and the reference vital sign data includes:

[0043] Normalize the measured heart rate and blood pressure at multiple time points respectively, and calculate the normalized measured heart rate H at multiple time points t The mean value H and standard deviation σ H , and calculate the normalized measured blood pressure B at multiple time points t The mean value B and standard deviation σ B;

[0044] Construct the heart rate distribution range of the target time period (Hm×σ H ,H+m×σ H ) and blood pressure distribution range (Bm×σ B ,B+m×σ B ), where m is the range adjustment parameter;

[0045] The similarity is calculated based on the heart rate reference range and the blood pressure reference range in the reference vital sign data, wherein the similarity includes the heart rate similarity D1 and the blood pressure similarity D2. The mathematical expressions of the heart rate similarity D1 and the blood pressure similarity D2 are:

[0046]

[0047] Where, L H is the length of the heart rate reference range, l H L is the length of the overlapping interval between the heart rate reference range and the heart rate distribution range, B is the length of the blood pressure reference range, l B It is the length of the overlapping interval between the blood pressure reference range and the blood pressure distribution range.

[0048] In one embodiment of the present application, executing a question-and-answer process with the patient includes:

[0049] Sending a pre-built question template to the patient, wherein the question template is built based on possible symptom entities;

[0050] When feedback information is received from the patient, the question-and-answer process is completed; when no feedback information is received from the patient, the pre-built question template is sent to the patient until feedback information is received from the patient, or the number of times the question template is sent exceeds a preset threshold, the question-and-answer process is completed.

[0051] In one embodiment of the present application, when the similarity is less than a preset similarity threshold, executing a question-and-answer process with the patient includes:

[0052] When the heart rate similarity D1 is less than a preset similarity threshold or the blood pressure similarity D2 is less than a preset similarity threshold, a question-and-answer process with the patient is executed.

[0053] In one embodiment of the present application, it further includes:

[0054] The patient's key recovery period information is sent to a target object.

[0055] The present application also provides an intelligent extraction system for patient recovery period reference information, comprising:

[0056] an acquisition module, configured to acquire the patient's measured vital sign data and motion data during a target time period, wherein the measured vital sign data and motion data are collected by the patient's smart wearable device, and the target time period is a time period of a target duration before the current time point;

[0057] an intensity calculation module, configured to calculate the patient's current exercise intensity based on the exercise data of the target time period; and when the current exercise intensity is greater than a set value, add the target duration to the cumulative exercise duration of the current exercise;

[0058] a reference vital sign determination module, configured to determine target reference vital sign data of the patient within the target time period based on the patient's current exercise intensity, the patient's identity information, and a pre-built vital sign reference database, wherein the vital sign reference database includes reference vital sign data of multiple user groups performing exercises of various intensities;

[0059] a question-and-answer module, configured to calculate the similarity between the measured physical sign data and the reference physical sign data; and when the similarity is less than a preset similarity threshold, executing a question-and-answer process with the patient, wherein the question-and-answer process is used to extract symptom entities;

[0060] An information extraction module is used to construct key information of the patient's recovery period based on the patient's cumulative exercise time, the current exercise intensity, the measured physical sign data of the target time period, and the symptom entity.

[0061] The beneficial effects of the present invention are as follows: the present invention provides an intelligent extraction method and system for patient recovery period reference information, which identifies the user's exercise status and exercise intensity through the exercise data collected by the smart wearable device, and accumulates the patient's current exercise time when the exercise intensity reaches a certain value. At the same time, the user's vital sign data is collected in real time, and the reference vital sign data that matches the patient is determined from the vital sign reference database in combination with the current exercise intensity, and the similarity between the measured vital sign data and the reference vital sign data is calculated. If the similarity is too low, it can be inferred that the patient may be uncomfortable during exercise. At this time, the pre-built question template is used to execute the question-and-answer process to find the symptom entities that may appear in the patient during exercise. Finally, the accumulated exercise time, symptom entity, measured vital sign data and exercise intensity data are all extracted as key data to provide to the doctor, so that the doctor has relevant data to refer to and judge the patient's recovery. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0063] Figure 1 This is a diagram of an application scenario of a method for intelligently extracting reference information of a patient's recovery period, as shown in one embodiment of the present application;

[0064] Figure 2 This is a flow chart of a method for intelligently extracting patient recovery period reference information shown in one embodiment of the present application;

[0065] Figure 3 A flowchart for constructing the exercise type-weight sequence relationship table and the physical sign reference database in this application;

[0066] Figure 4 This is a structural diagram of an intelligent extraction system for patient recovery period reference information shown in one embodiment of the present application. DETAILED DESCRIPTION

[0067] The following describes the embodiments of the present invention through specific examples. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments. The details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other unless they conflict.

[0068] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention. Therefore, the drawings only show the layers related to the present invention and are not drawn according to the number, shape and size of the layers in actual implementation. In actual implementation, the type, quantity and proportion of each layer can be changed arbitrarily, and the layer layout type may also be more complicated.

[0069] In the following description, numerous details are set forth to provide a more thorough explanation of the embodiments of the present invention; however, it is apparent to one skilled in the art that the embodiments of the present invention may be practiced without these specific details.

[0070] Figure 1 This is an application scenario diagram of an intelligent extraction method for patient recovery period reference information shown in an embodiment of the present application, such as Figure 1 As shown, in this application, smart wearable devices, such as smart bracelets, are used to collect the user's motion data (such as three-axis acceleration, body tilt angle, etc.) and vital signs data (such as heart rate and blood pressure). Limited by the data processing capabilities of smart bracelets, this application uploads the collected data to the cloud for calculation and data processing. In order to identify the user's exercise intensity, cumulative exercise time, and extract the user's exercise intensity, cumulative exercise time, measured vital signs data and other data and provide them to the target object as basic data for judging the user's physical recovery.

[0071] In addition, the information of the relevant personnel involved in this application has been obtained with full consent and authorization, and the collection, use and processing of relevant information must comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0072] Figure 2 FIG. 1 is a flow chart of a method for intelligently extracting patient recovery period reference information in one embodiment of the present application. Figure 2 As shown, the intelligent extraction method of patient recovery period reference information of this embodiment may include the following steps:

[0073] S210, obtaining the patient's identity information, and the patient's measured vital sign data and motion data during a target time period, wherein the measured vital sign data and the motion data are collected by the patient's smart wearable device, and the target time period is a time period of a target duration before the current time point;

[0074] Specifically, identity information mainly includes age and gender; the physical sign data of people of different ages and genders when exercising are relatively large, so they need to be distinguished.

[0075] The measured vital sign data include the measured heart rate and the measured blood pressure.

[0076] Motion data mainly includes motion type data and three-axis acceleration data collected by smart wearable devices. The acquisition process is as follows:

[0077] S211, obtaining the current motion type and three-axis acceleration data collected by the smart wearable device, wherein the three-axis acceleration data includes the three-axis acceleration (X t ,Y t ,Z t );

[0078] The three-axis acceleration data generated by users in different sports types corresponds to different sports intensities. For example, when cycling and running, similar three-axis acceleration data often correspond to different sports intensities. Therefore, it is necessary to use the sports type recognition function of the smart bracelet to identify the user's current sports type.

[0079] Smart wristbands are typically equipped with a variety of sensors, such as accelerometers, gyroscopes, and heart rate sensors. These sensors can monitor the user's various physiological parameters and movement status in real time. For example, an accelerometer can detect changes in the user's movement direction and speed, while a gyroscope can sense the direction and rotation of the wrist. Existing technology can identify a user's movement type by analyzing data such as cadence, stride length, body tilt angle, and arm swing patterns.

[0080] Three-axis acceleration is an important basic data for user exercise intensity.

[0081] S212: Filter the three-axis acceleration data to obtain three-axis acceleration filtered data, wherein the three-axis acceleration filtered data includes the three-axis accelerations (X t ',Y t ',Z t ');

[0082] This application uses low-pass filtering to smooth the signal, with the goal of removing noise and smoothing the signal to facilitate the extraction of useful motion information.

[0083] S213, the three-axis acceleration (X t ',Y t ',Z t ') to combine and get the combined acceleration a t ;

[0084]

[0085] The combined acceleration is one-dimensional data. Converting three-dimensional data into one-dimensional data can effectively simplify the data dimension and facilitate subsequent data processing.

[0086] S214, based on the combined acceleration a at multiple time points t Draw the acceleration time domain variation waveform, extract all peak points in the acceleration time domain variation waveform, and calculate the standard deviation σ of all peak points;

[0087] The peak point reflects the characteristics of the waveform, including amplitude and frequency. Calculating the standard deviation of the peak point amplitude can reflect the stability of acceleration changes.

[0088] S215, performing a fast Fourier transform on the acceleration time-domain variation waveform to obtain an acceleration frequency-domain variation waveform, and extracting a main frequency component f0 from the acceleration frequency-domain variation waveform;

[0089] By extracting the main frequency component of the acceleration frequency domain change waveform, the main frequency of the acceleration change can be obtained to reflect the speed of the acceleration change.

[0090] S216 , constructing motion data based on the standard deviation σ and the main frequency component f0.

[0091] This application mainly uses the stability of acceleration changes and the speed of acceleration changes to reflect the intensity of exercise, so the standard deviation σ and the main frequency component f0 are mainly extracted.

[0092] This application uses a real-time acquisition method. That is, once the data extraction process is triggered, the measured vital sign data and motion data at multiple time points in the target time period before the current time point are used as basic data for analysis, processing, and extraction. The trigger condition can be a timed trigger or an active trigger.

[0093] S220, calculating the patient's current exercise intensity based on the patient's identity information and the exercise data of the target time period; and when the current exercise intensity is greater than a set value, adding the target duration to the cumulative exercise duration of the current exercise;

[0094] In this application, the patient's exercise intensity is identified by using exercise type data and three-axis acceleration data. In the prior art, data such as heart rate and blood pressure may be introduced to determine the patient's exercise intensity, but this application is aimed at patients in the recovery period. Therefore, during exercise, abnormal heart rate and blood pressure data are very likely to appear. Introducing heart rate and blood pressure will lead to problems in judging exercise intensity. Therefore, only using exercise type data and three-axis acceleration data to identify the patient's exercise intensity, the specific process is as follows:

[0095] S221, determining a target weight sequence based on the current exercise type, the patient's identity information, and a pre-constructed exercise type-weight sequence relationship table, wherein the exercise type-weight sequence relationship table includes weight sequences for various user groups when performing various exercise types;

[0096] S222: Weighting the exercise data of the target time period based on the target weight sequence to obtain a current exercise intensity S, wherein the mathematical expression of the current exercise intensity S is:

[0097] S=w1×σ+w2×f0

[0098] Where w1 is the first weight and w2 is the second weight.

[0099] In order to more accurately describe the patient's exercise intensity, for different exercise types and user groups, this application constructs different weight sequences in advance to perform weighted summation on the standard deviation σ and the main frequency component f0 to obtain the current exercise intensity.

[0100] Figure 3 This is a flow chart for constructing the exercise type-weight sequence relationship table and the physical sign reference database in this application, such as Figure 3 As shown, specifically, the process of constructing the motion type-weight sequence relationship table in this application includes:

[0101] (1) obtaining a plurality of sample data, wherein the sample data includes user information of a normal user, exercise type, measured physical sign sample data and exercise sample data sampled when the normal user exercises within a target duration;

[0102] The sample data is generated by normal users wearing smart bracelets while exercising, and is collected after authorization by the normal users.

[0103] (2) labeling the exercise intensity of multiple sample data i based on the measured physical sign sample data, and normalizing the measured physical sign sample data to obtain multiple sample data i containing labels, wherein the measured physical sign sample data includes heart rate and blood pressure;

[0104] Since the sample data is generated by normal users, their heart rate and blood pressure are of reference value. Using heart rate and blood pressure to label exercise intensity is more reasonable. However, considering individual differences between users, when standardizing, it is possible to use normal user feedback for labeling. For example, exercise intensity can range from 0-100, allowing users to evaluate the intensity of their exercise.

[0105] (3) dividing the plurality of sample data containing labels into a plurality of data units based on the user information;

[0106] (4) classifying the data unit based on the motion type to obtain a plurality of data subunits;

[0107] Factors that influence the relationship between vital signs like heart rate and blood pressure and exercise intensity primarily include age, gender, and exercise type. Therefore, this application first uses user information to divide multiple labeled sample data into multiple data units, so that each data unit only contains sample data generated by a user group of a specific age and gender. The data units are then further divided so that each data subunit only contains sample data generated when performing a specific type of exercise.

[0108] (5) clustering multiple sample data within the data subunit based on motion intensity to obtain multiple intensity clusters;

[0109] This application only clusters the exercise intensity, that is, clusters the one-dimensional data, and can use the K-means clustering algorithm to divide it into a fixed number of intensity clusters, such as 2-5 clusters.

[0110] (6) Calculate the mean value of the motion intensity, the mean value of the standard deviation of the acceleration peak value, and the mean value of the main frequency component of the acceleration frequency domain change waveform in each intensity cluster; and construct a mathematical expression for the motion intensity based on the mean value of the motion intensity, the mean value of the standard deviation of the acceleration peak value, and the mean value of the main frequency component of the acceleration frequency domain change waveform:

[0111] S i=w1×σ i +w2×f 0i

[0112] Where S i is the mean motion intensity of the ith intensity cluster, σ i is the standard deviation of the acceleration peak of the i-th intensity cluster, f 0i is the mean value of the main frequency component of the acceleration frequency domain variation waveform of the i-th intensity cluster;

[0113] For each intensity cluster, the distribution of its motion intensity is similar. Since the cluster contains sample data generated by a user group when performing the same type of exercise, its motion data is also similar. This application calculates the mean motion intensity, the mean acceleration peak standard deviation, and the mean of the main frequency component of the acceleration frequency domain change waveform within the cluster, thereby reflecting the relationship between motion intensity and motion data in each intensity cluster.

[0114] Because multiple clusters were defined above, multiple mathematical expressions for exercise intensity can be constructed. In this application, the mathematical expression for exercise intensity is a linear equation in two variables. By solving these two equations simultaneously, the values ​​of the first weight and the second weight can be solved. In this application, any two mathematical expressions for exercise intensity are combined into a system of equations to calculate one or more solutions for the first weight and the second weight.

[0115] (7) The mathematical expressions of any two exercise intensities are combined into an equation system, and the first weight w1 and the second weight w2 are solved. When any data subunit has multiple different first weights w1 or multiple different second weights w2, the multiple different first weights w1 or multiple different second weights w2 are averaged to obtain the updated first weight w1 and the updated second weight w2;

[0116] Finally, if there is only one set of solutions, the first weight and the second weight can be directly obtained. If there are multiple sets of solutions, their average values ​​can be calculated separately.

[0117] (8) Construct the motion type-weight sequence relationship table based on the first weight w1 and the second weight w2 of each data subunit.

[0118] Through the above process, we can obtain the weight sequence corresponding to the exercise data of different user groups when performing different exercises. By bringing it into the exercise data of the current user, we can obtain a more accurate exercise intensity that conforms to the labeling rules.

[0119] When calculating the patient's exercise intensity during the current time period, if the intensity exceeds a certain value, the patient is considered to have engaged in high-intensity exercise. The duration of high-intensity exercise can reflect the patient's physical recovery. Therefore, the duration of exercise during the target time period is added to the cumulative duration of this exercise. This yields the key information: cumulative exercise duration.

[0120] S230, determining target reference vital sign data of the patient within the target time period based on the patient's current exercise intensity, the patient's identity information, and a pre-built vital sign reference database, wherein the vital sign reference database includes reference vital sign data of multiple user groups performing exercises of various intensities;

[0121] After obtaining the current exercise intensity calculated from the exercise data, the application also refers to the physical sign data of normal users and the current patient's physical sign data to determine whether the patient's physical sign data is abnormal. If the physical sign is abnormal, whether it is accompanied by abnormal symptoms is the key information to be extracted.

[0122] In this application, a pre-built vital sign reference database provides reference vital sign data of normal users when performing similar sports. The method for constructing the vital sign reference database is as follows:

[0123] like Figure 3 As shown, the construction of the physical sign reference database is based on the data subunits mentioned above.

[0124] (1) Clustering multiple sample data within each data subunit based on heart rate and blood pressure to obtain multiple physical sign clusters. This application clusters data based on heart rate and blood pressure, which can be considered as clustering two-dimensional data, and therefore uses the DSCAN density clustering algorithm. The degree of aggregation of data points and the number of clusters are controlled by setting the maximum distance within the cluster.

[0125] (2) removing sample data with abnormal exercise intensity from the physical sign cluster to obtain a screened physical sign cluster;

[0126] This application uses a cyclic elimination method to eliminate abnormal sample data, specifically including:

[0127] Calculate the mean and variance of exercise intensity within the sign cluster;

[0128] If the variance within the physical sign cluster is greater than the preset variance threshold, it means that the exercise intensity data within the cluster is not clustered enough. At this time, calculate the deviation between each data in the cluster and the mean, remove the sample data with the largest deviation, and return to calculating the mean and variance of the exercise intensity within the physical sign cluster until the variance is less than or equal to the variance threshold.

[0129] (3) calculating the mean value A1 of exercise intensity, the standard deviation σ1 of exercise intensity, the mean value A2 of heart rate, the standard deviation σ2 of heart rate, the mean value A3 of blood pressure, and the standard deviation σ3 of blood pressure of the plurality of sample data in the screened physical sign cluster;

[0130] After the previously eliminated physical sign clusters, heart rate and blood pressure distribution characteristics are similar, and exercise intensity values ​​are also relatively clustered. Exercise intensity within a physical sign cluster can be considered to have a corresponding relationship with heart rate and blood pressure. This means that if a normal user engages in a certain exercise intensity within the cluster, the corresponding heart rate and blood pressure will likely fall within the corresponding heart rate and blood pressure range within the cluster.

[0131] (4) constructing the exercise intensity range (A1-n×σ1, A1+n×σ1) of the physical sign cluster based on the exercise intensity mean A1 and the exercise intensity standard deviation σ1; constructing the heart rate range (A2-n×σ2, A2+n×σ2) of the physical sign cluster based on the heart rate mean A2 and the heart rate standard deviation σ2; and constructing the blood pressure range (A3-n×σ3, A3+n×σ3) of the physical sign cluster based on the blood pressure mean A3 and the blood pressure standard deviation σ3, where n is a range adjustment parameter;

[0132] (5) Construct a reference database of physical signs based on the heart rate range and blood pressure range of multiple exercise intensity ranges.

[0133] Finally, a range based on the mean and standard deviation was constructed to construct a reference database of physical signs.

[0134] In this application, the DSCAN clustering algorithm is used to obtain clusters and then obtain the motion intensity range. The constructed motion intensity range may have blind spots, that is, the area cannot be covered. In this case, the blind spots can be covered by adjusting the range adjustment parameter n.

[0135] After building the vital sign reference database, the corresponding vital sign reference data (range) can be found by collecting the patient's user information, exercise type and exercise intensity.

[0136] S240, calculating the similarity between the measured physical sign data and the reference physical sign data; when the similarity is less than a preset similarity threshold, executing a question-and-answer process with the patient, wherein the question-and-answer process is used to extract symptom entities;

[0137] In this application, the similarity can reflect whether the patient's measured physical sign data is consistent with the corresponding reference physical sign data. If it is consistent, it means that the recovery is good. If there is a large discrepancy, it means that the recovery is poor and even abnormal symptoms may occur.

[0138] The similarity calculation process is as follows:

[0139] S241, normalize the measured heart rate and blood pressure at multiple time points respectively, and calculate the normalized measured heart rate H at multiple time points t The mean value H and standard deviation σ H , and calculate the normalized measured blood pressure B at multiple time points t The mean value B and standard deviation σ B ;

[0140] This application uses Min-Max Normalization to eliminate the data differences between heart rate and blood pressure, which facilitates subsequent calculations and processing.

[0141] Then calculate the measured heart rate H t The mean value H and standard deviation σ H , measured blood pressure B t The mean value B and standard deviation σ B , to reflect the distribution of the patient's heart rate and blood pressure during the target time period.

[0142] S242, construct the heart rate distribution range (Hm×σ H ,H+m×σ H ) and blood pressure distribution range (Bm×σ B ,B+m×σ B ), where m is the range adjustment parameter;

[0143] S243, calculating similarity based on the heart rate reference range and the blood pressure reference range in the reference vital sign data, wherein the similarity includes a heart rate similarity D1 and a blood pressure similarity D2, wherein the mathematical expressions of the heart rate similarity D1 and the blood pressure similarity D2 are:

[0144]

[0145] Where, L H is the length of the heart rate reference range, l H L is the length of the overlapping interval between the heart rate reference range and the heart rate distribution range, B is the length of the blood pressure reference range, l B It is the length of the overlapping interval between the blood pressure reference range and the blood pressure distribution range.

[0146] This application compares two ranges to obtain similarity. H Indicates the length of the heart rate distribution range, Indicates the ratio of the length of the heart rate distribution range to the length of the heart rate reference range. Indicates the ratio of the length of the overlap between the heart rate reference range and the heart rate distribution range to the length of the heart rate reference range. Both parameters reflect the similarity of the ranges.

[0147] In this application, m < n, so that the reference range is relatively large, while the heart rate distribution range within the target time period is relatively small. The advantage of such a setting is that more reference data can be retained as much as possible, and the main heart rate data within the target time period can be extracted, increasing the error tolerance rate of abnormal judgment and avoiding an overly sensitive abnormal judgment mechanism.

[0148] After obtaining the similarity, if the similarity is too small, it means that the physical signs of the current patient during exercise do not conform to the big data law, and some symptoms may occur, such as feeling a strong or rapid heartbeat, shortness of breath or difficulty breathing, dizziness or mild headache, chest discomfort or pain, blurred vision or temporary vision loss, flushing of the face or tinnitus, etc.

[0149] Specifically, when the heart rate similarity D1 is less than the preset similarity threshold or the blood pressure similarity D2 is less than the preset similarity threshold, the question-and-answer process with the patient is executed.

[0150] In order to confirm whether the patient has the above symptoms, this application constructs an inquiry template in advance, such as:

[0151] "Do you have the following symptoms: (1) feeling a strong or rapid heartbeat; (2) shortness of breath or difficulty breathing; (3) dizziness or mild headache; (4) chest discomfort or pain; (5) blurred vision or temporary vision loss; (6) flushing of the face or tinnitus..., if so, please reply with the corresponding serial number"

[0152] Then the pre-constructed question template is sent to the patient; when the feedback information from the patient is received, the question-and-answer process is completed; for example, if (2) and (5) are returned, the corresponding symptom entities "shortness of breath or difficulty breathing" and "blurred vision or temporary vision loss" can be obtained.

[0153] When the feedback information from the patient is not received, the pre-constructed question template is sent to the patient until the feedback information from the patient is received, or when the number of times the question template is sent exceeds the preset threshold, the question-and-answer process is completed. If the user does not reply all the time, after exceeding three times, it will no longer be sent, the communication will end, and only the physical sign data will be extracted.

[0154] S250, construct the key information for the patient's recovery period based on the patient's cumulative exercise duration, the current exercise intensity, the measured physical sign data of the target time period, and the symptom entities.

[0155] Finally, the cumulative exercise duration, current exercise intensity, measured physical sign data during the target time period, and symptom entities related to recovery are extracted as key information, and this key information is sent to the target recipient. This key information allows the recipient (such as a medical professional or monitoring personnel) to understand the patient's recovery status in real time and make timely recommendations or interventions to optimize the patient's recovery process.

[0156] The present invention provides an intelligent extraction method for patient recovery period reference information, which uses motion data collected by smart wearable devices to identify the user's exercise status and exercise intensity, and when the exercise intensity reaches a certain value, accumulates the patient's current exercise duration. At the same time, the user's vital sign data is collected in real time, and the reference vital sign data that matches the patient is determined from the vital sign reference database in combination with the current exercise intensity, and the similarity between the measured vital sign data and the reference vital sign data is calculated. If the similarity is too low, it can be inferred that the patient may be uncomfortable during exercise. At this time, a pre-built question template is used to execute the question-and-answer process to find the symptom entities that may appear in the patient during exercise. Finally, the accumulated exercise duration, symptom entity, measured vital sign data, and exercise intensity data are all extracted as key data to provide to the doctor, so that the doctor has relevant data to refer to and judge the patient's recovery.

[0157] like Figure 4 As shown, the present application also provides an intelligent extraction system for patient recovery period reference information, comprising:

[0158] an acquisition module, configured to acquire the patient's measured vital sign data and motion data during a target time period, wherein the measured vital sign data and motion data are collected by the patient's smart wearable device, and the target time period is a time period of a target duration before the current time point;

[0159] an intensity calculation module, configured to calculate the patient's current exercise intensity based on the exercise data of the target time period; and when the current exercise intensity is greater than a set value, add the target duration to the cumulative exercise duration of the current exercise;

[0160] a reference vital sign determination module, configured to determine target reference vital sign data of the patient within the target time period based on the patient's current exercise intensity, the patient's identity information, and a pre-built vital sign reference database, wherein the vital sign reference database includes reference vital sign data of multiple user groups performing exercises of various intensities;

[0161] a question-and-answer module, configured to calculate the similarity between the measured physical sign data and the reference physical sign data; and when the similarity is less than a preset similarity threshold, executing a question-and-answer process with the patient, wherein the question-and-answer process is used to extract symptom entities;

[0162] An information extraction module is used to construct key information of the patient's recovery period based on the patient's cumulative exercise time, the current exercise intensity, the measured physical sign data of the target time period, and the symptom entity.

[0163] The present invention provides an intelligent extraction system for patient recovery period reference information. The system uses motion data collected by smart wearable devices to identify the user's exercise status and exercise intensity, and when the exercise intensity reaches a certain value, it accumulates the patient's current exercise duration. At the same time, the user's vital sign data is collected in real time, and the reference vital sign data that matches the patient is determined from the vital sign reference database in combination with the current exercise intensity, and the similarity between the measured vital sign data and the reference vital sign data is calculated. If the similarity is too low, it can be inferred that the patient may be uncomfortable during exercise. At this time, a pre-built question template is used to execute the question-and-answer process to find the symptom entities that may appear in the patient during exercise. Finally, the accumulated exercise duration, symptom entity, measured vital sign data, and exercise intensity data are all extracted as key data to provide to the doctor, so that the doctor has relevant data to refer to and judge the patient's recovery.

[0164] This embodiment also provides an electronic terminal, including: a processor and a memory;

[0165] The memory is used to store computer programs, and the processor is used to execute the computer programs stored in the memory, so that the terminal executes any one of the methods in this embodiment.

[0166] Regarding the computer-readable storage medium in this embodiment, those skilled in the art will appreciate that all or part of the steps in the aforementioned method embodiments can be implemented using hardware associated with the computer program. The aforementioned computer program can be stored in a computer-readable storage medium. When executed, the program performs the steps in the aforementioned method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0167] The electronic terminal provided in this embodiment includes a processor, a memory, a transceiver and a communication interface. The memory and the communication interface are connected to the processor and the transceiver and complete communication with each other. The memory is used to store computer programs, the communication interface is used for communication, and the processor and the transceiver are used to run computer programs so that the electronic terminal executes the various steps of the above method.

[0168] In this embodiment, the memory may include a random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0169] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, and discrete hardware components.

[0170] In the above embodiments, although the present invention has been described in conjunction with specific embodiments of the present invention, many replacements, modifications and variations of these embodiments will be apparent to those skilled in the art based on the foregoing description. The embodiments of the present invention are intended to cover all such replacements, modifications and variations that fall within the broad scope of the appended claims.

[0171] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the present invention. Anyone skilled in the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by one of ordinary skill in the art without departing from the spirit and technical principles disclosed herein are intended to be covered by the claims of the present invention.

Claims

1. An intelligent extraction method for patient recovery period reference information, characterized in that: Including steps: Obtaining the patient's identity information, the patient's measured vital sign data and motion data in a target time period, wherein the measured vital sign data and the motion data are collected by the patient's smart wearable device, and the target time period is a time period of a target duration before the current time point; obtaining the patient's motion data in the target time period, including: obtaining the current motion type and three-axis acceleration data collected by the smart wearable device, wherein the three-axis acceleration data includes the three-axis acceleration (X) at multiple time points t ,Y t ,Z t ); Filter the three-axis acceleration data to obtain three-axis acceleration filtered data, wherein the three-axis acceleration filtered data includes the three-axis acceleration (X) at multiple time points after filtering. t ',Y t ',Z t '); The three-axis acceleration (X t ',Y t ',Z t ') to combine and get the combined acceleration a t ; Based on the combined acceleration a at multiple time points t Plotting the acceleration time-domain variation waveform, extracting all peak points in the acceleration time-domain variation waveform, and calculating the standard deviation σ of all peak points; performing a fast Fourier transform on the acceleration time-domain variation waveform to obtain an acceleration frequency-domain variation waveform, and extracting the main frequency component f0 from the acceleration frequency-domain variation waveform; constructing motion data based on the standard deviation σ and the main frequency component f0; Calculating the patient's current exercise intensity based on the patient's identity information and the exercise data of the target time period; and when the current exercise intensity is greater than a set value, adding the target duration to the cumulative exercise duration of this exercise; Determining target reference vital sign data of the patient within the target time period based on the patient's current exercise intensity, the patient's identity information, and a pre-built vital sign reference database, wherein the vital sign reference database includes reference vital sign data of multiple user groups performing exercises of various intensities; Calculating the similarity between the measured vital sign data and the reference vital sign data; when the similarity is less than a preset similarity threshold, executing a question-and-answer process with the patient, wherein the question-and-answer process is used to extract symptom entities; the measured vital sign data includes measured heart rates and measured blood pressures at multiple time points, and calculating the similarity between the measured vital sign data and the reference vital sign data, including: normalizing the measured heart rates and measured blood pressures at multiple time points, and calculating the normalized measured heart rates H at multiple time points. t The mean value H and standard deviation σ H , and calculate the normalized measured blood pressure B at multiple time points t The mean value B and standard deviation σ B ; Construct the heart rate distribution range of the target time period (Hm×σ H ,H+m×σ H ) and blood pressure distribution range (Bm×σ B ,B+m×σ B ), where m is a range adjustment parameter; similarity is calculated based on the heart rate reference range and the blood pressure reference range in the reference vital sign data, the similarity including the heart rate similarity D1 and the blood pressure similarity D2, wherein the mathematical expressions of the heart rate similarity D1 and the blood pressure similarity D2 are: Where, L H is the length of the heart rate reference range, l H L is the length of the overlapping interval between the heart rate reference range and the heart rate distribution range, B is the length of the blood pressure reference range, l B is the length of the overlapping interval between the blood pressure reference range and the blood pressure distribution range; The key information of the patient's recovery period is constructed based on the patient's cumulative exercise time, the current exercise intensity, the measured physical sign data of the target time period, and the symptom entity.

2. The intelligent extraction method of patient recovery period reference information according to claim 1, characterized in that: Calculating the patient's current exercise intensity based on the patient's identity information and the exercise data of the target time period includes: Determining a target weight sequence based on the current exercise type, the patient's identity information, and a pre-constructed exercise type-weight sequence relationship table, wherein the exercise type-weight sequence relationship table includes weight sequences for various user groups when performing various exercise types; The motion data of the target time period is weighted based on the target weight sequence to obtain a current motion intensity S, wherein the mathematical expression of the current motion intensity S is: S=w1×σ+w2×f0 Where w1 is the first weight and w2 is the second weight.

3. The intelligent extraction method of patient recovery period reference information according to claim 2, characterized in that: The process of constructing the movement type-weight sequence relationship table includes: Acquire a plurality of sample data, wherein the sample data includes user information of a normal user, exercise type, measured physical sign sample data and exercise sample data sampled when the normal user exercises within a target duration; Labeling the exercise intensity of multiple sample data i based on the measured physical sign sample data, and normalizing the measured physical sign sample data to obtain multiple sample data i containing labels, wherein the measured physical sign sample data includes heart rate and blood pressure; Dividing a plurality of sample data containing labels into a plurality of data units based on the user information; Classifying the data unit based on the motion type to obtain a plurality of data subunits; Clustering multiple sample data within the data subunit based on motion intensity to obtain multiple intensity clusters; Calculate the mean value of motion intensity, the mean value of the standard deviation of the acceleration peak value, and the mean value of the main frequency component of the acceleration frequency domain variation waveform in each intensity cluster; and construct a mathematical expression for motion intensity based on the mean value of motion intensity, the mean value of the standard deviation of the acceleration peak value, and the mean value of the main frequency component of the acceleration frequency domain variation waveform: S i =w1×σ i +w2×f 0i Where S i is the mean motion intensity of the ith intensity cluster, σ i is the standard deviation of the acceleration peak of the i-th intensity cluster, f 0i is the mean value of the main frequency component of the acceleration frequency domain variation waveform of the i-th intensity cluster; The mathematical expressions of any two exercise intensities are combined into an equation system, and the first weight w1 and the second weight w2 are solved. When any data subunit has multiple different first weights w1 or multiple different second weights w2, the multiple different first weights w1 or multiple different second weights w2 are averaged to obtain the updated first weight w1 and the updated second weight w2; The motion type-weight sequence relationship table is constructed based on the first weight w1 and the second weight w2 of each data subunit.

4. The intelligent extraction method of patient recovery period reference information according to claim 3, characterized in that: The construction process of the physical sign reference database includes: Clustering multiple sample data in each data subunit based on heart rate and blood pressure to obtain multiple physical sign clusters; Eliminating sample data with abnormal exercise intensity from the physical sign cluster to obtain a screened physical sign cluster; Calculating the exercise intensity mean A1, exercise intensity standard deviation σ1, heart rate mean A2, heart rate standard deviation σ2, blood pressure mean A3 and blood pressure standard deviation σ3 of the plurality of sample data in the screened physical sign cluster; Based on the exercise intensity mean A1 and the exercise intensity standard deviation σ1, the exercise intensity range of the sign cluster is constructed (A1-n×σ1, A1+n×σ1); based on the heart rate mean A2 and the heart rate standard deviation σ2, the heart rate range of the sign cluster is constructed (A2-n×σ2, A2+n×σ2); and based on the blood pressure mean A3 and the blood pressure standard deviation σ3, the blood pressure range of the sign cluster is constructed (A3-n×σ3, A3+n×σ3), where n is a range adjustment parameter; A physical sign reference database is constructed based on heart rate ranges and blood pressure ranges in multiple exercise intensity ranges.

5. The intelligent extraction method of patient recovery period reference information according to claim 1, characterized in that: Conduct a Q&A process with the patient, including: Sending a pre-built question template to the patient, wherein the question template is built based on possible symptom entities; When feedback information is received from the patient, the question-and-answer process is completed; when no feedback information is received from the patient, the pre-built question template is sent to the patient until feedback information is received from the patient, or the number of times the question template is sent exceeds a preset threshold, the question-and-answer process is completed.

6. The intelligent extraction method of patient recovery period reference information according to claim 1, characterized in that: When the similarity is less than a preset similarity threshold, executing a question-and-answer process with the patient, including: When the heart rate similarity D1 is less than a preset similarity threshold or the blood pressure similarity D2 is less than a preset similarity threshold, a question-and-answer process with the patient is executed.

7. The intelligent extraction method of patient recovery period reference information according to claim 1, characterized in that: Also includes: The patient's key recovery period information is sent to a target object.

8. An intelligent extraction system for patient recovery period reference information, characterized in that: include: The acquisition module is used to acquire the patient's measured vital sign data and motion data in the target time period, wherein the measured vital sign data and the motion data are collected by the patient's smart wearable device, and the target time period is a time period of a target duration before the current time point; acquiring the patient's motion data in the target time period includes: acquiring the current motion type and three-axis acceleration data collected by the smart wearable device, wherein the three-axis acceleration data includes the three-axis acceleration (X) at multiple time points t ,Y t ,Z t ); Filter the three-axis acceleration data to obtain three-axis acceleration filtered data, wherein the three-axis acceleration filtered data includes the three-axis acceleration (X) at multiple time points after filtering. t ',Y t ',Z t '); The three-axis acceleration (X t ',Y t ',Z t ') to combine and get the combined acceleration a t ; Based on the combined acceleration a at multiple time points t Plotting the acceleration time-domain variation waveform, extracting all peak points in the acceleration time-domain variation waveform, and calculating the standard deviation σ of all peak points; performing a fast Fourier transform on the acceleration time-domain variation waveform to obtain an acceleration frequency-domain variation waveform, and extracting the main frequency component f0 from the acceleration frequency-domain variation waveform; constructing motion data based on the standard deviation σ and the main frequency component f0; an intensity calculation module, configured to calculate the patient's current exercise intensity based on the exercise data of the target time period; and when the current exercise intensity is greater than a set value, add the target duration to the cumulative exercise duration of the current exercise; a reference vital sign determination module, configured to determine target reference vital sign data of the patient within the target time period based on the patient's current exercise intensity, the patient's identity information, and a pre-built vital sign reference database, wherein the vital sign reference database includes reference vital sign data of multiple user groups performing exercises of various intensities; The question-answering module is configured to calculate the similarity between the measured vital sign data and the reference vital sign data; when the similarity is less than a preset similarity threshold, execute a question-answering process with the patient, wherein the question-answering process is used to extract symptom entities; the measured vital sign data includes measured heart rates and measured blood pressures at multiple time points, and calculating the similarity between the measured vital sign data and the reference vital sign data includes: normalizing the measured heart rates and measured blood pressures at multiple time points, and calculating the normalized measured heart rates H at multiple time points. t The mean value H and standard deviation σ H , and calculate the normalized measured blood pressure B at multiple time points t The mean value B and standard deviation σ B ; Construct the heart rate distribution range of the target time period (Hm×σ H ,H+m×σ H ) and blood pressure distribution range (Bm×σ B ,B+m×σ B ), where m is a range adjustment parameter; similarity is calculated based on the heart rate reference range and the blood pressure reference range in the reference vital sign data, the similarity including the heart rate similarity D1 and the blood pressure similarity D2, wherein the mathematical expressions of the heart rate similarity D1 and the blood pressure similarity D2 are: Where, L H is the length of the heart rate reference range, l H L is the length of the overlapping interval between the heart rate reference range and the heart rate distribution range, B is the length of the blood pressure reference range, l B is the length of the overlapping interval between the blood pressure reference range and the blood pressure distribution range; An information extraction module is used to construct key information of the patient's recovery period based on the patient's cumulative exercise time, the current exercise intensity, the measured physical sign data of the target time period, and the symptom entity.

Images