Post-transplantation interaction system and post-transplantation interaction method

Abstract

The invention provides a post-transplantation interaction system and a post-transplantation interaction method. A mobile interaction terminal device obtains daily health data and basic information of a patient. And the server receives the data uploaded by the mobile interaction terminal equipment, inputs the data into the post-transplantation prediction model, calculates a rejection risk value and an infection risk value, and returns results to the mobile interaction terminal equipment. Wherein the transplantation post-operation prediction model is constructed based on a large amount of clinical data and a machine learning algorithm, and the rejection reaction and infection risk can be predicted according to daily health data and basic information of a patient. The system can collect and process data in real time, and can perform personalized risk assessment according to individual differences of patients. Through the system, a patient can perform health monitoring and risk assessment in a home environment without going to a hospital or a professional institution.

Description

Technical Field

[0001] This application relates to the field of biomedical technologies, and more particularly, to a post-transplant interaction system and a post-transplant interaction method. Background Art

[0002] Most existing post-transplant prediction models are based on detailed clinical data, including blood tests, urine tests, and imaging tests, etc. These data usually require professional instruments and laboratory environments to obtain. In a home environment, due to limited conditions, it is difficult for patients to perform these professional tests, making it difficult to accurately identify rejection and infection risks. Summary of the Invention

[0003] The purpose of the embodiments of this application is to provide a post-transplant interaction system and a post-transplant interaction method to solve the problem that in a home environment, due to limited conditions, patients are unable to perform professional tests, making it difficult to accurately identify rejection and infection risks.

[0004] A post-transplant interaction system provided by the embodiments of this application includes: a mobile interaction terminal device and a server;

[0005] The mobile interaction terminal device is used to obtain the patient's daily health data and basic information, and upload them to the server; among them, the basic information includes age, gender, medical history, transplant surgery information, post-transplant survival rate, and post-transplant time;

[0006] The server is used to input the patient's daily health data and basic information into a post-transplant prediction model to obtain a rejection risk value and an infection risk value; and return the rejection risk value and the infection risk value to the mobile interaction terminal device;

[0007] The mobile interaction terminal device is also used to display the rejection risk value and the infection risk value.

[0008] In the above technical solution, the mobile interaction terminal device acquires the patient's daily health data and basic information and uploads them to the server. Among them, the daily health data includes indicators that patients can monitor by themselves, such as body temperature, heart rate, blood pressure, urine volume, urine color, etc. Basic information includes age, gender, medical history, transplantation surgery information (such as surgery date, type of transplanted organ, etc.), survival rate after transplantation (which may be based on historical data or doctor's evaluation), and time after transplantation (the time from the surgery date to the present). The server receives the data uploaded by the mobile interaction terminal device, inputs it into the prediction model after transplantation, calculates the risk values of rejection and infection, and returns the results to the mobile interaction terminal device. Among them, the prediction model after transplantation is constructed based on a large amount of clinical data and machine learning algorithms, and can predict the risks of rejection and infection according to the patient's daily health data and basic information. The system can collect and process data in real time, and can perform personalized risk assessment according to the individual differences of patients (such as age, gender, medical history, etc.). Through this system, patients can conduct health monitoring and risk assessment in a home environment without having to go to a hospital or a professional institution.

[0009] In some alternative embodiments, the mobile interaction terminal device includes: a smart wearable device and a mobile interaction device; the daily health data includes body temperature, heart rate, blood pressure, exercise data, diet information, health information, and medication information;

[0010] The smart wearable device is used to acquire body temperature, heart rate, blood pressure, and exercise data;

[0011] The mobile interaction device is used to configure different questions to interact with the patient and acquire the patient's diet information, health information, and medication information.

[0012] In the above technical solution, according to the specific situation and needs of the patient, corresponding questions are configured to obtain targeted information. By combining the smart wearable device and the mobile interaction device, the system can collect comprehensive daily health data of the patient, including:

[0013] Body temperature: Reflects the body temperature change of the patient, which helps to identify abnormal conditions such as infection.

[0014] Heart rate: Reflects the health status of the heart, and abnormal heart rate may indicate heart problems or rejection.

[0015] Blood pressure: Reflects the health status of blood vessels, and abnormal blood pressure may indicate problems such as hypertension, hypotension, or rejection.

[0016] Exercise data: Reflects the patient's exercise habits and physical activity level, which helps to evaluate the patient's physical condition and recovery.

[0017] Dietary information: Reflects the patient's eating habits and nutritional intake, and helps to evaluate the patient's nutritional status and potential health risks.

[0018] Health information: Includes the patient's self-perception, symptom description, etc., and helps to identify potential health problems.

[0019] Medication information: Reflects the patient's medication use and helps to evaluate the efficacy and side effects of the medications.

[0020] In this embodiment, through the combination of the intelligent wearable device and the mobile interaction device, the post-transplantation interaction system can achieve comprehensive, real-time, and personalized monitoring of the patient's daily health data, providing strong support for the patient to better manage their health status in the home environment.

[0021] In some alternative embodiments, the server is further configured to:

[0022] Obtain clinical data from a public database, screen samples that meet the clinical conditions, and obtain initial sample data;

[0023] Preprocess and filter the initial sample data to obtain sample data with data categories the same as those of the daily health data and basic information;

[0024] Establish a univariate Cox proportional hazards model, and identify key features related to the risk of rejection or infection from the data categories of the daily health data and basic information;

[0025] Calculate the concordance index for each initial modeling feature, and select the feature with the highest concordance index as the starting feature;

[0026] Construct a multivariate Cox regression model, gradually add features in an iterative manner, and calculate the concordance index to maximize the model performance;

[0027] When the concordance index of the multivariate Cox regression model no longer increases, stop modeling to form a primary Cox proportional hazards model;

[0028] Optimize the weight coefficients of the key features of the primary Cox proportional hazards model to obtain a post-transplantation prediction model;

[0029] Configure the data that the mobile interaction terminal device needs to upload, and configure the questions set in the mobile interaction device according to the key features of the post-transplantation prediction model.

[0030] In the above technical solution, the server screens out the sample data that meets the requirements from the public database according to specific clinical conditions to form the initial sample data. The initial sample data is cleaned, filtered, and formatted to ensure the data quality and make it consistent with the data categories of daily health data and basic information. The univariate Cox proportional hazards model is used to identify the key features related to the risk of rejection or infection from the daily health data and basic information. Among them, the concordance index (such as C-index) is an important indicator to measure the prediction performance of the model. The server calculates the concordance index of each initial modeling feature and selects the feature with the highest concordance index as the starting feature. Based on the starting feature, other features are gradually added in an iterative manner, and the concordance index after each addition is calculated to maximize the performance of the model. When the concordance index of the multivariate Cox regression model no longer increases, the modeling is stopped to form the primary Cox proportional hazards model. The weight coefficients of the key features of the primary Cox proportional hazards model are optimized, for example, by methods such as Lasso regression, Ridge regression, and cross-validation. According to the key features of the postoperative transplantation prediction model, the data types and frequencies that the mobile interaction terminal device needs to upload are configured. This helps to ensure that the collected data is of practical significance for model prediction. According to the key features of the model, the questions set in the mobile interaction device are configured to obtain non-physiological data related to the prediction result. This helps to provide more comprehensive patient health information and improve the prediction accuracy of the model. Therefore, the system can more accurately identify the key features related to the risk of rejection and infection, construct a high-performance prediction model, and perform personalized configuration according to the key features of the model to provide more comprehensive and accurate health monitoring and risk assessment services for patients.

[0031] In some alternative embodiments, preprocessing of the initial sample data includes:

[0032] Identifying abnormal data in the initial sample data and performing replacement;

[0033] Performing normalization processing on the initial sample data.

[0034] In the above technical solution, abnormal data (also known as outliers or extreme values) refers to those values that are significantly different from most data points. These values may be caused by measurement errors, data entry errors, or other atypical factors. The existence of abnormal data may have a negative impact on the training of the model, resulting in a decline in model performance or inaccurate prediction results. For example, the 3σ principle (i.e., when the distance of a data point from the mean exceeds 3 times the standard deviation, it is regarded as abnormal) or the IQR (interquartile range) method can be used to identify outliers. Replace the abnormal data with a certain statistic (such as the mean, median, or mode), or use interpolation methods (such as linear interpolation, nearest neighbor interpolation, etc.) to estimate missing or abnormal values. Normalization processing is to convert data with different dimensions to the same scale, so that each feature has a comparable weight during the model training process. This helps to accelerate the convergence speed of the model and improve the prediction performance of the model.

[0035] In some alternative embodiments, identifying abnormal data in the initial sample data and performing replacement includes:

[0036] If the data in the initial sample data exceeds the corresponding reasonable range interval, then this data is considered abnormal data, and this abnormal data is replaced with the corresponding typical value.

[0037] In the above technical solution, for each feature (or variable), determine a reasonable value range according to existing knowledge, experience, or data distribution characteristics. This range should be able to cover most normal data points while excluding those outliers that are significantly deviated from the normal range. Traverse each data point in the initial sample data and check whether it exceeds the reasonable range interval of the corresponding feature. If a data point exceeds the reasonable range interval of its corresponding feature, then this data point is considered abnormal data. For the identified abnormal data, a typical value can be selected to replace it. This typical value can be the mean, median, mode, or other representative values of this feature. The purpose of replacing abnormal data is to make the data closer to the real situation and at the same time reduce the impact of outliers on model training.

[0038] In some alternative embodiments, inputting the patient's daily health data and basic information into the post-transplantation prediction model to obtain the rejection risk value and the infection risk value includes:

[0039] In the case where one or more data in the daily health data and basic information are missing, after filling the missing data with different typical values, input them into the post-transplantation prediction model for prediction respectively to obtain multiple prediction results;

[0040] Execute different processing flows according to multiple prediction results:

[0041] If, among some of the multiple prediction results, the risk value of rejection reaction and / or the risk value of infection are greater than the first threshold, a first warning message is sent to the mobile interactive terminal device, and the first warning message is used to prompt the patient to complete the data and then re - perform the prediction;

[0042] If, among all of the multiple prediction results, the risk value of rejection reaction and / or the risk value of infection are greater than the first threshold, a second warning message is sent to the mobile interactive terminal device, and the second warning message is used to prompt the patient to go to the hospital for professional testing as soon as possible;

[0043] If none of the multiple prediction results show that the risk value of rejection reaction and / or the risk value of infection are greater than the first threshold, no warning message is sent to the mobile interactive terminal device.

[0044] Among them, the selection of different typical values is from samples with different labels. For example, 4 different typical values are selected. These 4 typical values are: one typical value is selected from samples with a higher risk value of rejection reaction; one typical value is selected from samples with a certain risk value of rejection reaction; one typical value is selected from samples with a higher risk value of infection; one typical value is selected from samples with a lower risk value of infection.

[0045] In the above - mentioned technical solution, when there are missing items in the patient's daily health data and basic information, these missing items are first identified. To fill in these missing data, 4 different typical values are selected from the existing samples. The basis for the selection of these typical values is:

[0046] One typical value is selected from samples with a higher risk value of rejection reaction, representing a high - risk situation.

[0047] One typical value is selected from samples with a lower risk value of rejection reaction, representing a low - risk situation.

[0048] One typical value is selected from samples with a higher risk value of infection, also representing a high - risk situation.

[0049] One typical value is selected from samples with a lower risk value of infection, representing a low - risk situation.

[0050] These 4 typical values are used to fill in the missing data, generating 4 complete data sets.

[0051] Each filled - in complete data set is respectively input into the post - transplantation prediction model to predict the risk value of rejection reaction and the risk value of infection.

[0052] Each data set will generate a set of prediction results, including the risk value of rejection reaction and the risk value of infection.

[0053] Execute different processing flows based on multiple prediction results:

[0054] If the risk value of rejection reaction and / or the risk value of infection are greater than the first threshold in some of the prediction results, this may mean that the data incompleteness has a certain impact on the prediction results, or the patient indeed has certain health risks. At this time, send a first warning prompt message to the mobile interactive terminal device, prompting the patient to complete the data and re - perform the prediction, and give suggestions on which data points are missing and may lead to abnormal prediction results.

[0055] If all the prediction results show that the risk value of rejection reaction and / or the risk value of infection are greater than the first threshold, this may mean that the patient indeed has a relatively high health risk. At this time, send a second warning prompt message to the mobile interactive terminal device, strongly suggesting that the patient go to the hospital for professional testing as soon as possible, and give some suggestions for emergency treatment.

[0056] If all the prediction results show that the risk value of rejection reaction and the risk value of infection are lower than the first threshold, then the patient's health condition is currently within the safe range. At this time, do not send any warning prompt message to the mobile interactive terminal device, but can continue to monitor the patient's health data.

[0057] In some alternative embodiments, the mobile interactive terminal device is also used for:

[0058] Show different prompt messages according to the risk value of rejection reaction and the risk value of infection:

[0059] If the risk value of rejection reaction or the risk value of infection is greater than the first threshold, then show a prompt message for prompting the patient to go to the hospital for professional testing as soon as possible;

[0060] If the risk value of rejection reaction or the risk value of infection is less than or equal to the first threshold and greater than the second threshold, then compare the patient's daily health data with the corresponding standard health data range, and show the data in the patient's daily health data that exceeds the standard health data range; where the first threshold is greater than the second threshold.

[0061] In the above - mentioned technical solution, when the risk value of rejection reaction or the risk value of infection is greater than the first threshold, the mobile interactive terminal device shows an emergency prompt message, suggesting that the patient go to the hospital for professional testing as soon as possible.

[0062] When the rejection risk value or the infection risk value is less than or equal to the first threshold but greater than the second threshold (i.e., at a medium risk level), the mobile interactive terminal device compares the patient's daily health data with the corresponding standard health data range. This step aims to help the patient identify which health indicators may be problematic or deviate from the normal range. Through the comparison, the device will display the data points in the patient's daily health data that exceed the standard health data range. For example, if the patient's blood pressure value is higher than the normal range, the device will specifically mark it and may give some suggestions or reminders, such as "Your blood pressure is high. Please pay attention to controlling your diet and exercising moderately."

[0063] When both the rejection risk value and the infection risk value are less than or equal to the second threshold (i.e., at a low risk or no risk level), the mobile interactive terminal device may not display any special prompt information, or only display a confirmation message, such as "Your health condition is currently good. Please continue to maintain it."

[0064] A post-transplant interaction method provided by an embodiment of the present application includes:

[0065] Obtain the patient's daily health data and basic information, and upload them to the server; wherein, the basic information includes age, gender, medical history, transplant surgery information, post-transplant survival rate, and post-transplant time;

[0066] Input the patient's daily health data and basic information into a post-transplant prediction model to obtain a rejection risk value and an infection risk value;

[0067] Display the rejection risk value and the infection risk value.

[0068] In some optional implementation manners, before obtaining the patient's daily health data and basic information, it further includes:

[0069] Obtain clinical data from a public database, screen samples that meet the clinical conditions to obtain initial sample data;

[0070] Preprocess and filter the initial sample data to obtain sample data with the same data categories as the daily health data and basic information;

[0071] Establish a univariate Cox proportional hazards model, and identify key features related to the rejection risk or infection risk from the data categories of the daily health data and basic information;

[0072] Calculate the concordance index of each initial modeling feature, and select the feature with the highest concordance index as the starting feature;

[0073] Construct a multivariate Cox regression model, gradually add features in an iterative manner, and calculate the concordance index to maximize the model performance;

[0074] When the concordance index of the multi-factor Cox regression model no longer increases, stop modeling to form a primary Cox proportional hazards model;

[0075] Optimize the weight coefficients of the key features of the primary Cox proportional hazards model to obtain a prediction model after transplantation.

[0076] In some alternative embodiments, input the patient's daily health data and basic information into the prediction model after transplantation to obtain a rejection risk value and an infection risk value, including:

[0077] In the case where one or more of the daily health data and basic information are missing, fill in the missing data with different typical values, and then input them into the prediction model after transplantation for prediction respectively to obtain multiple prediction results;

[0078] Execute different processing procedures according to multiple prediction results:

[0079] If some of the multiple prediction results show that the rejection risk value and / or the infection risk value are greater than the first threshold, send a first warning prompt message, which is used to prompt the patient to complete the data and re-predict;

[0080] If all of the multiple prediction results show that the rejection risk value and / or the infection risk value are greater than the first threshold, send a second warning prompt message, which is used to prompt the patient to go to the hospital for professional testing as soon as possible;

[0081] If none of the multiple prediction results show that the rejection risk value and / or the infection risk value are greater than the first threshold, do not send a warning prompt message. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required to be used in the embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0083] Figure 1 A functional module diagram of an interactive system after transplantation provided by an embodiment of the present application;

[0084] Figure 2 A flowchart of the steps of an interactive method after transplantation provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0085] The following will describe the technical solutions in the embodiments of the present application in conjunction with the accompanying drawings in the embodiments of the present application.

[0086] Please refer to Figure 1 , Figure 1 , which is a functional module diagram of a post-transplant interaction system provided by an embodiment of the present application, including: a mobile interaction terminal device and a server;

[0087] The mobile interaction terminal device is used to obtain the patient's daily health data and basic information, and upload them to the server; among them, the basic information includes age, gender, medical history, transplant surgery information, post-transplant survival rate, and post-transplant time;

[0088] The server is used to input the patient's daily health data and basic information into the post-transplant prediction model to obtain the rejection risk value and the infection risk value; and, return the rejection risk value and the infection risk value to the mobile interaction terminal device;

[0089] The mobile interaction terminal device is also used to display the rejection risk value and the infection risk value.

[0090] In the embodiments of the present application, the mobile interaction terminal device obtains the patient's daily health data and basic information, and uploads them to the server. Among them, the daily health data includes indicators that patients can monitor by themselves, such as body temperature, heart rate, blood pressure, urine volume, urine color, etc. Basic information: includes age, gender, medical history, transplant surgery information (such as surgery date, type of transplanted organ, etc.), post-transplant survival rate (which may be based on historical data or doctor evaluation), and post-transplant time (the time from the surgery date to the present). The server receives the data uploaded by the mobile interaction terminal device, inputs it into the post-transplant prediction model, calculates the rejection risk value and the infection risk value, and returns the results to the mobile interaction terminal device. Among them, the post-transplant prediction model is constructed based on a large amount of clinical data and machine learning algorithms, and can predict the risks of rejection and infection according to the patient's daily health data and basic information. The system can collect and process data in real time, and can perform personalized risk assessment according to the individual differences of patients (such as age, gender, medical history, etc.). Through this system, patients can conduct health monitoring and risk assessment in a home environment without having to go to a hospital or a professional institution.

[0091] In some alternative embodiments, the mobile interaction terminal device includes: a smart wearable device and a mobile interaction device; the daily health data includes body temperature, heart rate, blood pressure, exercise data, diet information, health information, and medication information;

[0092] The smart wearable device is used to obtain body temperature, heart rate, blood pressure, and exercise data;

[0093] The mobile interaction device is used to configure different questions to interact with patients and obtain the patients' dietary information, health information, and medication information.

[0094] In the embodiments of the present application, according to the specific conditions and needs of the patients, corresponding questions are configured to obtain targeted information. By combining the smart wearable device and the mobile interaction device, the system can collect comprehensive daily health data of the patients, including:

[0095] Body temperature: Reflects the change in the patient's body temperature, which helps to identify abnormal conditions such as infections.

[0096] Heart rate: Reflects the health status of the heart. An abnormal heart rate may indicate heart problems or rejection reactions.

[0097] Blood pressure: Reflects the health status of the blood vessels. Abnormal blood pressure may indicate problems such as hypertension, hypotension, or rejection reactions.

[0098] Exercise data: Reflects the patient's exercise habits and physical activity level, which helps to evaluate the patient's physical condition and recovery.

[0099] Dietary information: Reflects the patient's eating habits and nutritional intake, which helps to evaluate the patient's nutritional status and potential health risks.

[0100] Health information: Includes the patient's self-perception, symptom description, etc., which helps to identify potential health problems.

[0101] Medication information: Reflects the patient's medication situation, which helps to evaluate the efficacy and side effects of the drugs.

[0102] Through the combination of the smart wearable device and the mobile interaction device in this embodiment, the post-transplant interaction system can achieve comprehensive, real-time, and personalized monitoring of the patients' daily health data, providing strong support for the patients to better manage their health status in the home environment.

[0103] In some alternative embodiments, the server is further used for:

[0104] Obtaining clinical data from a public database, screening samples that meet the clinical conditions, and obtaining initial sample data;

[0105] Preprocessing and filtering the initial sample data to obtain sample data with the same data categories as the daily health data and basic information;

[0106] Establishing a univariate Cox proportional hazards model to identify key features related to the risk of rejection or infection from the data categories of the daily health data and basic information;

[0107] Calculate the concordance index for each initial modeling feature, and select the feature with the highest concordance index as the starting feature;

[0108] Construct a multi-factor Cox regression model, gradually add features in an iterative manner, and calculate the concordance index to maximize the model performance;

[0109] When the concordance index of the multi-factor Cox regression model no longer increases, stop modeling to form a primary Cox proportional hazards model;

[0110] Optimize the weight coefficients of the key features of the primary Cox proportional hazards model to obtain a postoperative transplantation prediction model;

[0111] According to the key features of the postoperative transplantation prediction model, configure the data that the mobile interaction terminal device needs to upload, and configure the questions set in the mobile interaction device.

[0112] In the embodiments of the present application, the server screens out the sample data that meets the requirements from the public database according to specific clinical conditions to form the initial sample data. Clean, filter, and format the initial sample data to ensure the data quality and make it consistent with the data categories of daily health data and basic information. Use a single-factor Cox proportional hazards model to identify the key features related to the rejection risk or infection risk from the daily health data and basic information. Among them, the concordance index (such as C-index) is an important indicator to measure the model prediction performance. The server calculates the concordance index for each initial modeling feature and selects the feature with the highest concordance index as the starting feature. Based on the starting feature, gradually add other features in an iterative manner and calculate the concordance index after each addition to maximize the model performance. When the concordance index of the multi-factor Cox regression model no longer increases, stop modeling to form a primary Cox proportional hazards model. Optimize the weight coefficients of the key features of the primary Cox proportional hazards model, such as through methods like Lasso regression, Ridge regression, cross-validation, etc. According to the key features of the postoperative transplantation prediction model, configure the data types and frequencies that the mobile interaction terminal device needs to upload. This helps to ensure that the collected data is meaningful for model prediction. According to the key features of the model, configure the questions set in the mobile interaction device to obtain non-physiological data related to the prediction results. This helps to provide more comprehensive patient health information and improve the prediction accuracy of the model. Therefore, the system can more accurately identify the key features related to the rejection risk and infection risk, construct a high-performance prediction model, and perform personalized configuration according to the key features of the model to provide more comprehensive and accurate health monitoring and risk assessment services for patients.

[0113] In some alternative embodiments, preprocess the initial sample data, including:

[0114] Identify the abnormal data in the initial sample data and replace it;

[0115] Perform normalization processing on the initial sample data.

[0116] In the embodiments of the present application, abnormal data (also known as outliers or extreme values) refer to those values that are significantly different from most data points. These values may be caused by measurement errors, data entry errors, or other atypical factors. The existence of abnormal data may have a negative impact on the training of the model, resulting in a decline in model performance or inaccurate prediction results. For example, the 3σ principle (i.e., when the distance of a data point from the mean exceeds 3 times the standard deviation, it is regarded as abnormal) or the IQR (interquartile range) method can be used to identify outliers. Replace the abnormal data with a certain statistic (such as the mean, median, or mode), or use interpolation methods (such as linear interpolation, nearest neighbor interpolation, etc.) to estimate the missing or abnormal values. Normalization processing is to convert data with different dimensions to the same scale, so that each feature has a comparable weight in the model training process. This helps to accelerate the convergence speed of the model and improve the prediction performance of the model.

[0117] In some alternative embodiments, identifying the abnormal data in the initial sample data and replacing it includes:

[0118] If the data in the initial sample data exceeds the corresponding reasonable range interval, then consider this data as abnormal data and replace this abnormal data with the corresponding typical value.

[0119] In the embodiments of the present application, for each feature (or variable), determine a reasonable value range according to existing knowledge, experience, or data distribution characteristics. This range should be able to cover most normal data points while excluding those abnormal values that are significantly deviated from the normal range. Traverse each data point in the initial sample data and check whether it exceeds the reasonable range interval of the corresponding feature. If a data point exceeds the reasonable range interval of its corresponding feature, then consider this data point as abnormal data. For the identified abnormal data, a typical value can be selected to replace it. This typical value can be the mean, median, mode, or other representative values of this feature. The purpose of replacing abnormal data is to make the data closer to the real situation and reduce the impact of outliers on model training.

[0120] In some alternative embodiments, inputting the patient's daily health data and basic information into the post-transplant prediction model to obtain the rejection risk value and the infection risk value includes:

[0121] In the case where one or more data in the daily health data and basic information are missing, after filling the missing data with different typical values, input them into the post-transplant prediction model for prediction respectively to obtain multiple prediction results;

[0122] Execute different processing flows according to multiple prediction results:

[0123] If, for some of the multiple prediction results, the rejection reaction risk value and / or the infection risk value is greater than the first threshold, a first warning prompt message is sent to the mobile interactive terminal device, and the first warning prompt message is used to prompt the patient to complete the data and then re - perform the prediction;

[0124] If, for all of the multiple prediction results, the rejection reaction risk value and / or the infection risk value is greater than the first threshold, a second warning prompt message is sent to the mobile interactive terminal device, and the second warning prompt message is used to prompt the patient to go to the hospital for professional testing as soon as possible;

[0125] If none of the multiple prediction results shows that the rejection reaction risk value and / or the infection risk value is greater than the first threshold, no warning prompt message is sent to the mobile interactive terminal device.

[0126] Among them, different typical values are selected from samples with different labels. For example, 4 different typical values are selected. These 4 typical values are: select a typical value from samples with a relatively high rejection reaction risk value; select a typical value from samples with a certain rejection reaction risk value; select a typical value from samples with a relatively high infection risk value; select a typical value from samples with a relatively low infection risk value.

[0127] In the embodiments of the present application, when there are missing items in the patient's daily health data and basic information, these missing items are first identified. To fill these missing data, 4 different typical values are selected from the existing samples. The basis for selecting these typical values is:

[0128] Select a typical value from samples with a relatively high rejection reaction risk value to represent a high - risk situation.

[0129] Select a typical value from samples with a relatively low rejection reaction risk value to represent a low - risk situation.

[0130] Select a typical value from samples with a relatively high infection risk value, which also represents a high - risk situation.

[0131] Select a typical value from samples with a relatively low infection risk value to represent a low - risk situation.

[0132] Use these 4 typical values to fill the missing data and generate 4 complete data sets.

[0133] Input each filled complete data set into the post - transplantation prediction model respectively to predict the rejection reaction risk value and the infection risk value.

[0134] Each data set generates a set of prediction results, including the risk value of rejection and the risk value of infection.

[0135] Execute different processing flows based on multiple prediction results:

[0136] If in some of the prediction results, the risk value of rejection and / or the risk value of infection is greater than the first threshold, this may mean that the data incompleteness has had a certain impact on the prediction results, or the patient indeed has certain health risks. At this time, send a first warning prompt message to the mobile interaction terminal device, prompting the patient to complete the data and re - perform the prediction, and give suggestions on which missing data points may have caused the abnormal prediction results.

[0137] If all prediction results show that the risk value of rejection and / or the risk value of infection is greater than the first threshold, this may mean that the patient indeed has a relatively high health risk. At this time, send a second warning prompt message to the mobile interaction terminal device, strongly recommending that the patient go to the hospital for professional testing as soon as possible, and give some suggestions for emergency treatment.

[0138] If all prediction results show that the risk value of rejection and the risk value of infection are lower than the first threshold, then the patient's health condition is currently within the safe range. At this time, do not send any warning prompt message to the mobile interaction terminal device, but the patient's health data can be continuously monitored.

[0139] In some alternative embodiments, the mobile interaction terminal device is also used for:

[0140] Show different prompt messages according to the risk value of rejection and the risk value of infection:

[0141] If the risk value of rejection or the risk value of infection is greater than the first threshold, then show a prompt message for prompting the patient to go to the hospital for professional testing as soon as possible;

[0142] If the risk value of rejection or the risk value of infection is less than or equal to the first threshold and greater than the second threshold, then compare the patient's daily health data with the corresponding standard health data range, and show the data in the patient's daily health data that exceeds the standard health data range; where the first threshold is greater than the second threshold.

[0143] In the embodiments of the present application, when the risk value of rejection or the risk value of infection is greater than the first threshold, the mobile interaction terminal device shows an emergency prompt message, recommending that the patient go to the hospital for professional testing as soon as possible.

[0144] When the rejection risk value or the infection risk value is less than or equal to the first threshold but greater than the second threshold (i.e., at a medium risk level), the mobile interaction terminal device compares the patient's daily health data with the corresponding standard health data range. This step aims to help the patient identify which health indicators may be problematic or deviate from the normal range. Through the comparison, the device will display the data points in the patient's daily health data that exceed the standard health data range. For example, if the patient's blood pressure value is higher than the normal range, the device will specifically mark it and may give some suggestions or reminders, such as "Your blood pressure is on the high side. Please pay attention to controlling your diet and exercising moderately."

[0145] When both the rejection risk value and the infection risk value are less than or equal to the second threshold (i.e., at a low risk or no risk level), the mobile interaction terminal device may not display any special prompt information, or only display a confirmation message, such as "Your current health condition is good. Please continue to maintain it."

[0146] In addition, the post-transplant interaction system also has a science popularization push and medication reminder function:

[0147] The science popularization push function aims to provide the patient with educational content and knowledge related to post-transplant health management. By regularly pushing science popularization articles, videos or graphic materials, it helps the patient understand the precautions after transplantation, common health problems and their treatment methods. The science popularization content should be based on the latest medical research and clinical practice to ensure the accuracy and scientific nature of the information. The content can cover aspects such as the prevention and management of post-transplant rejection, reduction of infection risk, precautions for drug use, diet and lifestyle adjustment, etc. According to factors such as the patient's health condition, risk value changes, and historical reading records, relevant science popularization content is intelligently recommended. The patient is allowed to set interest preferences to receive science popularization information that better meets personal needs.

[0148] The medication reminder function aims to ensure that the patient takes the medications required after transplantation on time and in the correct dosage. By regularly sending reminder messages, it helps the patient establish good medication habits and improve the therapeutic effect of the medications. The system should allow the patient to input or import their personal medication plan, including information such as the drug name, dosage, medication time and frequency. According to the patient's medication plan, the system sends reminder messages to the mobile interaction terminal device at regular intervals. The reminder messages can include key information such as the drug name, medication time and dosage. According to the patient's preferences and habits, the system can provide various reminder methods such as voice reminder, vibration reminder or text reminder.

[0149] Both the science popularization push and medication reminder functions can further introduce intelligent recommendation algorithms to provide more accurate and personalized services based on data such as the patient's health status, interest preferences, and historical behaviors. Establish a user feedback mechanism to collect patients' feedback on science popularization content and medication reminders. Continuously optimize the push content and reminder strategies according to user feedback to improve the practicality of the system and the user experience.

[0150] In some alternative embodiments, the post-transplant interaction system further includes a medical staff terminal. The medical staff terminal enables doctors or nurses to interact with patients through this terminal and solve the patients' personalized problems. At the same time, the medical staff terminal also allows doctors or nurses to set questions on the mobile interaction device to collect the patients' personalized data, which can also be used for modeling the post-transplant prediction model, thereby enhancing the interactivity and personalized service capabilities of the system.

[0151] Please refer to Figure 2 , Figure 2 which is a flowchart of the steps of a post-transplant interaction method provided by an embodiment of the present application, including:

[0152] Step S1: Obtain the patient's daily health data and basic information and upload them to the server; where the basic information includes age, gender, medical history, transplant surgery information, post-transplant survival rate, and post-transplant time;

[0153] Step S2: Input the patient's daily health data and basic information into the post-transplant prediction model to obtain the rejection risk value and the infection risk value;

[0154] Step S3: Display the rejection risk value and the infection risk value.

[0155] In some alternative embodiments, before obtaining the patient's daily health data and basic information, it further includes:

[0156] Obtain clinical data from a public database, screen samples that meet the clinical conditions to obtain initial sample data;

[0157] Preprocess and filter the initial sample data to obtain sample data with the same data categories as the daily health data and basic information;

[0158] Establish a univariate Cox proportional hazards model, and identify key features related to the rejection risk or infection risk from the data categories of the daily health data and basic information;

[0159] Calculate the concordance index of each initial modeling feature, and select the feature with the highest concordance index as the starting feature;

[0160] Construct a multi-factor Cox regression model, gradually add features in an iterative manner, and calculate the concordance index to maximize the model performance;

[0161] When the concordance index of the multi-factor Cox regression model no longer increases, stop modeling to form a primary Cox proportional hazards model;

[0162] Optimize the weight coefficients of the key features of the primary Cox proportional hazards model to obtain a prediction model after transplantation.

[0163] In some alternative embodiments, input the patient's daily health data and basic information into the prediction model after transplantation to obtain the rejection risk value and the infection risk value, including:

[0164] In the case where one or more data in the daily health data and basic information are missing, fill the missing data with different typical values, and then input them into the prediction model after transplantation for prediction respectively to obtain multiple prediction results;

[0165] Execute different processing flows according to multiple prediction results:

[0166] If, for some of the multiple prediction results, the rejection risk value and / or the infection risk value is greater than the first threshold, send a first warning prompt message, which is used to prompt the patient to complete the data and re-predict;

[0167] If, for all of the multiple prediction results, the rejection risk value and / or the infection risk value is greater than the first threshold, send a second warning prompt message, which is used to prompt the patient to go to the hospital for professional testing as soon as possible;

[0168] If none of the multiple prediction results show that the rejection risk value and / or the infection risk value is greater than the first threshold, do not send a warning prompt message.

[0169] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. Also, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point, the displayed or discussed couplings or direct couplings or communication connections to each other can be through some communication interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical or other forms.

[0170] In addition, the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0171] Furthermore, in each embodiment of the present application, the various functional modules may be integrated together to form an independent part, or each module may exist alone, or two or more modules may be integrated to form an independent part.

[0172] In this text, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations.

[0173] The above description is only for the embodiments of the present application and is not intended to limit the protection scope of the present application. For those skilled in the art, the present application can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A post-transplantation interactive system, characterized in that: include: Mobile interactive terminal devices and servers; The mobile interactive terminal device is used to obtain the patient's daily health data and basic information and upload them to the server; wherein the basic information includes age, gender, medical history, transplantation operation information, post-transplantation survival rate and post-transplantation time; The server is used to input the patient's daily health data and basic information into a post-transplantation prediction model to obtain a rejection risk value and an infection risk value; and return the rejection risk value and the infection risk value to the mobile interactive terminal device; The mobile interactive terminal device is also used to display the rejection reaction risk value and the infection risk value.

2. The system according to claim 1, characterized in that The mobile interactive terminal device includes: a smart wearable device and a mobile interactive device; the daily health data includes body temperature, heart rate, blood pressure, exercise data, diet information, health information and medication information; The smart wearable device is used to obtain body temperature, heart rate, blood pressure and exercise data; The mobile interactive device is used to configure different questions to interact with the patient and obtain the patient's dietary information, health information and medication information.

3. The system according to claim 2, characterized in that The server is also used to: Obtain clinical data from public databases, screen samples that meet clinical conditions, and obtain initial sample data; Preprocess and filter the initial sample data to obtain sample data with the same data category as the daily health data and basic information; A univariate Cox proportional hazards model was established to identify key features associated with the risk of rejection or infection from data categories of daily health data and basic information; Calculate the consistency index of each initial modeling feature and select the feature with the highest consistency index as the starting feature; Construct a multi-factor Cox regression model, add features step by step in an iterative manner, and calculate the consistency index to maximize the model performance; When the consistency index of the multivariate Cox regression model no longer increased, modeling was stopped and a primary Cox proportional hazard model was formed; The weight coefficients of the key features of the primary Cox proportional hazards model were optimized to obtain a post-transplantation prediction model; According to the key features of the post-transplantation prediction model, the data that needs to be uploaded by the mobile interactive terminal device is configured, and the questions set in the mobile interactive device are configured.

4. The system according to claim 1, characterized in that Preprocess the initial sample data, including: Identify abnormal data in the initial sample data and replace them; Normalize the initial sample data.

5. The system according to claim 4, characterized in that The identifying and replacing abnormal data in the initial sample data includes: If the data in the initial sample data exceeds the corresponding reasonable range, the data is considered to be abnormal data and is replaced by the corresponding typical value.

6. The system according to claim 1, wherein: The patient's daily health data and basic information are input into the post-transplantation prediction model to obtain the rejection risk value and infection risk value, including: In the case where one or more data in the daily health data and basic information are missing, the missing data are filled with different typical values ​​and then respectively input into the post-transplantation prediction model for prediction to obtain multiple prediction results; Execute different processing flows based on multiple prediction results: If the rejection risk value and / or infection risk value of some of the multiple prediction results is greater than the first threshold, a first warning prompt information is sent to the mobile interaction terminal device, and the first warning prompt information is used to prompt the patient to complete the data and re-predict; If all the prediction results among the multiple prediction results show that the rejection risk value and / or the infection risk value is greater than the first threshold, a second alarm prompt information is sent to the mobile interaction terminal device, and the second alarm prompt information is used to prompt the patient to go to the hospital for professional testing as soon as possible; If none of the multiple prediction results shows that the rejection risk value and / or the infection risk value is greater than the first threshold, no warning prompt information is sent to the mobile interaction terminal device.

7. The system according to claim 1, characterized in that The mobile interactive terminal device is also used for: Different prompt information is displayed according to the rejection risk value and the infection risk value: If the rejection risk value or the infection risk value is greater than the first threshold, displaying a prompt message for prompting the patient to go to the hospital for professional testing as soon as possible; If the rejection reaction risk value or the infection risk value is less than or equal to the first threshold and greater than the second threshold, the patient's daily health data is compared with the corresponding standard health data range, and the data in the patient's daily health data that exceeds the standard health data range is displayed; wherein the first threshold is greater than the second threshold.

8. A post-transplantation interaction method, characterized in that: include: Obtain the patient's daily health data and basic information and upload them to the server; wherein the basic information includes age, gender, medical history, transplantation operation information, post-transplantation survival rate and post-transplantation time; The patient's daily health data and basic information are input into the post-transplant prediction model to obtain the rejection risk value and infection risk value; The rejection risk value and the infection risk value are displayed.

9. The method according to claim 8, characterized in that Before obtaining the patient's daily health data and basic information, it also includes: Obtain clinical data from public databases, screen samples that meet clinical conditions, and obtain initial sample data; Preprocess and filter the initial sample data to obtain sample data with the same data category as the daily health data and basic information; A univariate Cox proportional hazards model was established to identify key features associated with the risk of rejection or infection from data categories of daily health data and basic information; Calculate the consistency index of each initial modeling feature and select the feature with the highest consistency index as the starting feature; Construct a multi-factor Cox regression model, add features step by step in an iterative manner, and calculate the consistency index to maximize the model performance; When the consistency index of the multivariate Cox regression model no longer increased, modeling was stopped and a primary Cox proportional hazard model was formed; The weight coefficients of the key features of the primary Cox proportional hazards model were optimized to obtain a prediction model for transplantation.

10. The method according to claim 8, characterized in that The patient's daily health data and basic information are input into the post-transplantation prediction model to obtain the rejection risk value and infection risk value, including: In the case where one or more data in the daily health data and basic information are missing, the missing data are filled with different typical values ​​and then respectively input into the post-transplantation prediction model for prediction to obtain multiple prediction results; Execute different processing flows based on multiple prediction results: If the rejection risk value and / or infection risk value of some of the multiple prediction results is greater than the first threshold, a first warning prompt information is sent, wherein the first warning prompt information is used to prompt the patient to complete the data and re-predict; If all the prediction results in the multiple prediction results show that the rejection risk value and / or the infection risk value is greater than the first threshold, a second warning prompt information is sent, and the second warning prompt information is used to prompt the patient to go to the hospital for professional testing as soon as possible; If none of the multiple prediction results shows that the rejection risk value and / or the infection risk value is greater than the first threshold, no alarm prompt information is sent.

Images