Diabetes early warning model construction method and system based on machine learning

Abstract

The invention relates to the technical field of medical data analysis, in particular to a diabetes early warning model construction method and system based on machine learning, and the method comprises the steps: obtaining multi-source life data based on a plurality of information channels which comprise but are not limited to electronic medical records, physical examination reports and lifestyle investigation; generating multi-source medical data according to the multi-source life data; performing feature selection extraction on the multi-source medical data to obtain diabetes basic features, and constructing diabetes composite features based on the health basic features; training an initial machine learning model according to the diabetes mellitus composite features to obtain a diabetes mellitus analysis model, and obtaining a diabetes mellitus prediction result based on the diabetes mellitus analysis model; and generating a patient risk level based on the diabetes prediction result, and generating a personalized prevention scheme according to the patient risk level in combination with basic information of the patient. The method has the effect of improving the accuracy and reliability of diabetes early warning.

Description

Technical Field

[0001] The present invention relates to the technical field of medical data analysis, and in particular to a method and system for constructing a diabetes early warning model based on machine learning. Background Art

[0002] As a global chronic disease, the incidence of diabetes is gradually increasing, and it has become an important factor affecting human health and quality of life. Traditional diabetes diagnosis methods mainly rely on biochemical indicators such as blood sugar testing, but this method can only detect abnormalities when the disease has developed to a certain extent, making it difficult to achieve early warning and intervention.

[0003] Existing diabetes warning technologies include data analysis based on electronic medical records, interpretation of physical examination reports, and lifestyle questionnaires. Although the above methods provide support for early warning of diabetes to a certain extent, there are still certain problems, such as single and incomplete data sources, and excessive reliance on a single data source, such as electronic medical records, physical examination reports or questionnaires, resulting in incomplete information acquisition. In addition, due to the limitations of data sources and algorithms, the accuracy of diabetes warnings in existing technologies is often not high, which can easily lead to misdiagnosis or missed diagnosis, and it is impossible to improve the effective early warning mechanism, causing patients to miss the best time for intervention. Therefore, there is room for improvement. Summary of the invention

[0004] In order to improve the accuracy and reliability of diabetes early warning, help doctors quantify risk levels, and provide a scientific basis for patients' early treatment and lifestyle adjustments, the present application provides a method and system for constructing a diabetes early warning model based on machine learning.

[0005] In the first aspect, the above-mentioned invention objective of the present application is achieved through the following technical solutions: A method for constructing a diabetes early warning model based on machine learning, the method comprising the steps of: Acquire multi-source life data based on multiple information channels, the multiple information channels including but not limited to electronic medical records, physical examination reports, and lifestyle surveys, and generate multi-source medical data based on the multi-source life data; Performing feature selection and extraction on the multi-source medical data to obtain basic features of diabetes, and constructing composite features of diabetes based on the basic health features; Training an initial machine learning model according to the composite features of diabetes to obtain a diabetes analysis model, and obtaining a diabetes prediction result based on the diabetes analysis model; A patient risk level is generated based on the diabetes prediction result, and a personalized prevention plan is generated according to the patient risk level and combined with the patient's basic information.

[0006] By adopting the above technical solution, in the process of diabetes prediction and early warning, multi-source life data are collected through information channels such as electronic medical records, physical examination reports and lifestyle questionnaires, and the multi-source life data are deeply integrated to form comprehensive multi-source medical data, which provides a richer and more accurate information basis for diabetes risk assessment. The integrated multi-source medical data is subjected to feature selection and extraction, which not only identifies the basic characteristics of diabetes, such as blood sugar level, insulin sensitivity and other basic diabetes characteristics, but also further constructs diabetes composite characteristics including lifestyle, genetic factors and other factors, which significantly improves the accuracy and comprehensiveness of diabetes risk assessment. Based on the constructed diabetes composite characteristics, a diabetes analysis model with higher prediction accuracy is trained, which can accurately predict the risk of an individual suffering from diabetes, providing a scientific basis for early intervention. The trained diabetes analysis model is used to output diabetes prediction results, and the patient risk level is generated by the diabetes prediction results. Combined with the patient risk level and the patient's basic information (such as age, gender, historical medical history, etc.), a personalized prevention plan is automatically generated, which is convenient for patients and doctors to take the most suitable preventive measures according to the specific situation, improves the effectiveness of diabetes prevention intervention, and provides a new and efficient method for early warning and prevention of diabetes.

[0007] In a preferred example, the present application may be further configured as follows: generating multi-source medical data according to the multi-source life data specifically includes: Performing data preprocessing on the multi-source life data, wherein the data preprocessing includes data integration processing and data cleaning processing to obtain processed multi-source life data; The processed multi-source life data is subjected to data conversion processing to form multi-source medical data.

[0008] By adopting the above technical solution, a comprehensive data set containing various health information of patients is formed by integrating data from different channels, such as electronic medical records, physical examination reports, lifestyle surveys, etc., which can solve the problems of data format, unit, naming inconsistency and other problems that may exist between different data sources, so that the data can be seamlessly connected in subsequent analysis, and the integrated comprehensive data set is cleaned to remove duplicate, erroneous and incomplete data, thereby improving the accuracy and reliability of the data, and thus improving the accuracy of subsequent diabetes risk analysis. The processed multi-source life data is converted into a data format suitable for subsequent analysis, such as converting text data into numerical data, converting unstructured data into structured data, extracting medical data that has an important impact on diabetes risk assessment from multi-source life data, and forming multi-source medical data. The multi-source medical data contains data in multiple dimensions such as the patient's medical history, physiological parameters, biochemical indicators, lifestyle, etc., which provides reliable data for subsequent diabetes risk assessment, construction of prediction models and formulation of personalized prevention plans.

[0009] In a preferred example, the present application may be further configured as follows: before acquiring multi-source life data based on multiple information channels, wherein the multiple information channels include but are not limited to electronic medical records, physical examination reports, and lifestyle surveys, and generating multi-source medical data based on the multi-source life data, the method for constructing a diabetes early warning model based on machine learning also includes: During the data collection process, the data access logs recorded in multiple information channels using blockchain technology were used to obtain the initial patient record data; Encrypting and encoding the initial patient record data, and adding random noise to the encoded initial patient data to blur the individual data of the patient; The initial patient data after adding random noise is aggregated to form multi-source life data.

[0010] By adopting the above technical solution, in the process of collecting multi-source data of patients, the privacy of patients' personal data is involved. In the process of data collection, the differential factor algorithm is used to process the collected data to improve the privacy of patients' personal data. Specifically, blockchain technology is used to record data access logs in multiple information channels to obtain initial patient record data. The obtained initial patient record data is first encrypted and encoded for subsequent randomization and aggregation steps. Random noise is added to the encrypted initial patient record data to blur the patient's individual data, thereby effectively protecting the patient's personal sensitive data in the data collection process. The initial patient data after adding random noise is then aggregated to eliminate the influence of some noise and form the patient's multi-source life data, thereby ensuring the utility of the collected patient data while minimizing the risk of privacy leakage.

[0011] In a preferred example, the present application can be further configured as follows: the feature selection and extraction of the multi-source medical data to obtain the basic features of diabetes, and the construction of the composite features of diabetes based on the basic health features, specifically including: Acquire a data source type based on the multi-source medical data, and determine a suitable feature algorithm according to the data source type; Performing feature extraction on the multi-source medical data according to a suitable feature algorithm to extract basic features of diabetes and time series features; The basic features of diabetes and time series features are formed into a feature set, and the features in the feature set are combined and transformed to form a composite feature of diabetes.

[0012] By adopting the above technical solution, in the process of feature analysis of the collected multi-source medical data, the multi-source medical data will have different data types. According to the multi-source medical data, the data source type is obtained, such as text data, digital data, medical image data and other different data source types. According to the data source type, the appropriate feature algorithm is determined to efficiently and accurately extract features from each type of data source. According to the appropriate feature algorithm, the multi-source medical data is feature extracted to extract the basic features of diabetes and time series features. The basic features of diabetes refer to the basic features of diabetes such as blood sugar level and insulin sensitivity. The time series features refer to the time series characteristics of the patient's medical data, such as blood sugar monitoring data. The time series characteristics will be included in the data, and the extracted basic features of diabetes and time series features will be integrated together. The time series features will be temporally associated with the basic features of diabetes to form a feature set, so as to analyze the changing trends and patterns of medical data about diabetes, and to combine and transform the features in the feature set. For example, the blood sugar level feature will be combined with the insulin level feature to form a composite feature of insulin sensitivity index. For example, qualitative features such as family medical history and lifestyle will be quantified and combined with quantitative features to form a composite feature of diabetes. The composite feature of diabetes can more comprehensively reflect the patient's physiological state, thereby facilitating a more accurate risk assessment of diabetes.

[0013] In a preferred example, the present application may be further configured as follows: the initial machine learning model is trained according to the diabetes composite features to obtain a diabetes analysis model, specifically including: Performing feature dimensionality reduction processing on the diabetes composite features, screening out diabetes-related composite features, and integrating them to form a training set, a validation set, and a test set; Determining data characteristics based on the training set, determining a machine learning algorithm based on the data characteristics, and training a preset initial machine learning model based on the machine learning algorithm and the training set; Inputting the test set into the trained machine learning model to determine the optimal model parameters; Adjust the trained machine learning model based on the optimal model parameters, input the verification set into the machine learning model after parameter adjustment, and output the model verification result; According to the model verification result, it is judged whether the initial machine model meets the diabetes risk assessment effect. If so, a diabetes analysis model is obtained.

[0014] By adopting the above technical solution, in the process of building a diabetes analysis model, the composite features of diabetes are subjected to feature dimensionality reduction processing, such as using principal component analysis and linear discriminant analysis, which can remove redundant and noise features in the composite features of diabetes, screen out composite features that are highly associated with diabetes, and integrate the screened composite features to form training sets, validation sets, and test sets to ensure the independence and representativeness of the data. According to the data characteristics of the training set, a suitable machine learning algorithm is selected, and the preset initial machine learning model is trained in combination with the training set, which is conducive to improving the interpretability and stability of the model. The test set is then used to continuously iteratively optimize the machine learning model and adjust the model parameters. Finally, the validation set is used to verify whether the model meets the effect requirements of diabetes risk assessment, thereby completing the construction of the diabetes analysis model to improve the early detection of diabetes.

[0015] In a preferred example, the present application may be further configured as follows: generating a patient risk level based on the diabetes prediction result, and generating a personalized prevention plan according to the patient risk level and in combination with the patient's basic information, specifically including: Inputting the collected multi-source medical data into the diabetes analysis model, outputting a binary classification prediction result, and calculating the probability of having diabetes based on the binary classification prediction result; The patient risk level is set according to the probability of suffering from diabetes, and a corresponding diabetes prevention plan is generated based on the patient risk level and in combination with the patient's basic information.

[0016] By adopting the above technical solution and utilizing the trained diabetes analysis model, multi-source medical data is input into the diabetes analysis model to output a binary prediction result, i.e., whether the patient has diabetes, so as to achieve precise prevention and improve the prevention effect. The probability of the patient having diabetes is calculated based on the binary prediction result output by the diabetes analysis model. According to the probability of the patient having diabetes, the patient's risk level is assessed to form a patient risk level, such as dividing the probability into three levels: low risk, medium risk and high risk. Each level has a corresponding prevention plan. According to the patient's risk level and combined with the patient's basic information, such as the patient's age, living habits, allergy history and other basic information, a corresponding diabetes prevention plan is generated, thereby achieving the effect of precise prevention of diabetes.

[0017] In a preferred example, the present application may be further configured as follows: after generating a patient risk level based on the diabetes prediction result, and generating a personalized prevention plan according to the patient risk level and in combination with the patient's basic information, the method for constructing a diabetes early warning model based on machine learning also includes: Acquire patient monitoring data after adopting personalized prevention plan in real time, and generate prevention plan adjustment instructions based on the patient monitoring data; In response to the prevention plan adjustment instruction, a new personalized prevention plan is generated.

[0018] By adopting the above-mentioned technical solution, by acquiring the patient's monitoring data in real time, such as blood sugar level and other data, even if it is found that the patient's health status changes after taking the prevention plan, a prevention plan adjustment instruction is generated according to the monitoring data, and the original personalized prevention plan is adjusted in response to the prevention plan adjustment instruction to form a new personalized prevention plan. This can ensure that the prevention plan matches the patient's current condition, thereby improving the accuracy and effectiveness of diabetes prevention.

[0019] In the second aspect, the above invention objective of the present application is achieved through the following technical solutions: A diabetes early warning model construction system based on machine learning, the diabetes early warning model construction system based on machine learning comprising: A patient data collection module, used to obtain multi-source life data based on multiple information channels, including but not limited to electronic medical records, physical examination reports, and lifestyle surveys, and generate multi-source medical data based on the multi-source life data; A data feature extraction module, used to perform feature selection and extraction on the multi-source medical data to obtain basic features of diabetes, and construct composite features of diabetes based on the basic health features; A model building module, used to train an initial machine learning model according to the composite characteristics of diabetes to obtain a diabetes analysis model, and obtain diabetes prediction results based on the diabetes analysis model; The prevention plan generation module is used to generate a patient risk level based on the diabetes prediction result, and generate a personalized prevention plan according to the patient risk level and combined with the patient's basic information.

[0020] By adopting the above technical solution, in the process of diabetes prediction and early warning, multi-source life data are collected through information channels such as electronic medical records, physical examination reports and lifestyle questionnaires, and the multi-source life data are deeply integrated to form comprehensive multi-source medical data, which provides a richer and more accurate information basis for diabetes risk assessment. The integrated multi-source medical data is subjected to feature selection and extraction, which not only identifies the basic characteristics of diabetes, such as blood sugar level, insulin sensitivity and other basic diabetes characteristics, but also further constructs diabetes composite characteristics including lifestyle, genetic factors and other factors, which significantly improves the accuracy and comprehensiveness of diabetes risk assessment. Based on the constructed diabetes composite characteristics, a diabetes analysis model with higher prediction accuracy is trained, which can accurately predict the risk of an individual suffering from diabetes, providing a scientific basis for early intervention. The trained diabetes analysis model is used to output diabetes prediction results, and the patient risk level is generated by the diabetes prediction results. Combined with the patient risk level and the patient's basic information (such as age, gender, historical medical history, etc.), a personalized prevention plan is automatically generated, which is convenient for patients and doctors to take the most suitable preventive measures according to the specific situation, improves the effectiveness of diabetes prevention intervention, and provides a new and efficient method for early warning and prevention of diabetes.

[0021] On the third aspect, the above-mentioned purpose of the present application is achieved through the following technical solutions: An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned method for constructing a diabetes early warning model based on machine learning when executing the computer program.

[0022] Fourthly, the above-mentioned purpose of the present application is achieved through the following technical solutions: A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for constructing a diabetes early warning model based on machine learning.

[0023] In summary, the present application includes at least one of the following beneficial technical effects: 1. Multi-source life data is collected through electronic medical records, physical examination reports, lifestyle questionnaires and other information channels. The multi-source life data is deeply integrated to form comprehensive multi-source medical data, which provides a richer and more accurate information basis for diabetes risk assessment. The integrated multi-source medical data is subjected to feature selection and extraction, which not only identifies the basic characteristics of diabetes, such as blood sugar level, insulin sensitivity and other basic diabetes characteristics, but also further constructs diabetes composite characteristics including lifestyle, genetic factors and other factors, which significantly improves the accuracy and comprehensiveness of diabetes risk assessment. Based on the constructed diabetes composite characteristics, a diabetes analysis model with higher prediction accuracy is trained, which can accurately predict the risk of an individual suffering from diabetes, providing a scientific basis for early intervention. The trained diabetes analysis model is used to output diabetes prediction results, and the patient risk level is generated by the diabetes prediction results. Combined with the patient risk level and the patient's basic information (such as age, gender, historical medical history, etc.), a personalized prevention plan is automatically generated, which is convenient for patients and doctors to take the most suitable preventive measures according to the specific situation, improves the effectiveness of diabetes prevention intervention, and provides a new and efficient method for early warning and prevention of diabetes. 2. By integrating data from different channels, such as electronic medical records, physical examination reports, and lifestyle surveys, a comprehensive data set containing various aspects of patients' health information is formed, which can solve the problems of data format, unit, and naming inconsistency between different data sources, so that the data can be seamlessly connected in subsequent analysis. The integrated comprehensive data set is cleaned to remove duplicate, erroneous, and incomplete data, improve the accuracy and reliability of the data, and thus improve the accuracy of subsequent diabetes risk analysis. The processed multi-source life data is converted into a data format suitable for subsequent analysis, such as converting text data into numerical data and unstructured data into structured data. Medical data that has an important impact on diabetes risk assessment is extracted from multi-source life data to form multi-source medical data. Multi-source medical data contains data in multiple dimensions such as the patient's medical history, physiological parameters, biochemical indicators, and lifestyle, providing reliable data for subsequent diabetes risk assessment, the construction of prediction models, and the formulation of personalized prevention plans; 3. In the process of collecting multi-source data of patients, the privacy of patients' personal data is involved. In the process of data collection, the differential factor algorithm is applied to process the collected data to improve the privacy of patients' personal data. Specifically, blockchain technology is used to record data access logs in multiple information channels to obtain initial patient record data. The initial patient record data is first encrypted for subsequent randomization and aggregation steps. Random noise is added to the encrypted initial patient record data to blur the patient's individual data, thereby effectively protecting the patient's personal sensitive data during the data collection process. The initial patient data after adding random noise is then aggregated to eliminate the influence of some noise and form the patient's multi-source life data, thereby ensuring the utility of the collected patient data while minimizing the risk of privacy leakage; 4. Perform feature dimensionality reduction processing on the composite features of diabetes, such as using principal component analysis and linear discriminant analysis, to remove redundant and noise features in the composite features of diabetes, screen out composite features that are highly correlated with diabetes, and integrate the screened composite features to form training sets, validation sets, and test sets to ensure the independence and representativeness of the data. Select a suitable machine learning algorithm based on the data characteristics of the training set, and train the preset initial machine learning model in combination with the training set, which is conducive to improving the interpretability and stability of the model. Then use the test set to continuously iteratively optimize the machine learning model and adjust the model parameters. Finally, use the validation set to verify whether the model meets the effect requirements of diabetes risk assessment, thereby completing the construction of the diabetes analysis model to improve the early detection of diabetes. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 is a flow chart of a method for constructing a diabetes early warning model based on machine learning in one embodiment of the present application; Figure 2 is a flowchart for implementing step S10 in the method for constructing a diabetes early warning model based on machine learning in one embodiment of the present application; Figure 3 is another implementation flow chart of the method for constructing a diabetes early warning model based on machine learning in one embodiment of the present application; Figure 4 is a flowchart for implementing step S20 in the method for constructing a diabetes early warning model based on machine learning in one embodiment of the present application; Figure 5 is a flowchart for implementing step S30 in the method for constructing a diabetes early warning model based on machine learning in one embodiment of the present application; Figure 6 is a flowchart for implementing step S40 in the method for constructing a diabetes early warning model based on machine learning in one embodiment of the present application; Figure 7is another implementation flow chart of the method for constructing a diabetes early warning model based on machine learning in one embodiment of the present application; Figure 8 It is a principle block diagram of a diabetes early warning model construction system based on machine learning in one embodiment of the present application; Fig. 9 It is a schematic diagram of an electronic device in an embodiment of the present application. DETAILED DESCRIPTION

[0025] The present application is further described in detail below in conjunction with the accompanying drawings.

[0026] In one embodiment, if Figure 1 As shown, the present application discloses a method for constructing a diabetes early warning model based on machine learning, which specifically includes the following steps: S10: Acquire multi-source life data based on multiple information channels, where the multiple information channels include but are not limited to electronic medical records, physical examination reports, and lifestyle surveys, and generate multi-source medical data based on the multi-source life data.

[0027] Specifically, we use information channels such as electronic medical records, physical examination reports and lifestyle questionnaires to collect multi-source life data, which are deeply integrated to form comprehensive multi-source medical data, providing a richer and more accurate information basis for diabetes risk assessment.

[0028] S20: performing feature selection and extraction on the multi-source medical data to obtain basic features of diabetes, and constructing composite features of diabetes based on the basic health features.

[0029] Specifically, the integrated multi-source medical data is subjected to feature selection and extraction, which not only identifies the basic characteristics of diabetes, such as blood sugar levels, insulin sensitivity and other basic characteristics of diabetes, but also further constructs complex characteristics of diabetes including lifestyle, genetic factors, etc., significantly improving the accuracy and comprehensiveness of diabetes risk assessment.

[0030] S30: Training an initial machine learning model according to the composite features of diabetes to obtain a diabetes analysis model, and obtaining a diabetes prediction result based on the diabetes analysis model.

[0031] Specifically, based on the constructed diabetes composite features, a diabetes analysis model with higher prediction accuracy is trained, which can accurately predict the risk of an individual developing diabetes, providing a scientific basis for early intervention. The trained diabetes analysis model is used to output diabetes prediction results.

[0032] S40: Generate a patient risk level based on the diabetes prediction result, and generate a personalized prevention plan based on the patient risk level and combined with the patient's basic information.

[0033] Specifically, by generating a patient risk level based on the diabetes prediction results, and combining the patient risk level and the patient's basic information (such as age, gender, medical history, etc.), a personalized prevention plan is automatically generated, allowing patients and doctors to take the most appropriate preventive measures based on specific circumstances, thereby improving the effectiveness of diabetes prevention interventions.

[0034] In this embodiment, in the process of diabetes prediction and early warning, multi-source life data are collected through information channels such as electronic medical records, physical examination reports and lifestyle questionnaires, and the multi-source life data are deeply integrated to form comprehensive multi-source medical data, which provides a richer and more accurate information basis for diabetes risk assessment. The integrated multi-source medical data is subjected to feature selection and extraction, not only the basic characteristics of diabetes, such as blood sugar level, insulin sensitivity and other basic characteristics of diabetes, but also the composite characteristics of diabetes including lifestyle, genetic factors and the like are further constructed, which significantly improves the accuracy and comprehensiveness of diabetes risk assessment. Based on the constructed composite characteristics of diabetes, a diabetes analysis model with higher prediction accuracy is trained, which can accurately predict the risk of an individual suffering from diabetes, providing a scientific basis for early intervention. The trained diabetes analysis model is used to output diabetes prediction results, and the patient risk level is generated by the diabetes prediction results. Combined with the patient risk level and the patient's basic information (such as age, gender, historical medical history, etc.), a personalized prevention plan is automatically generated, which is convenient for patients and doctors to take the most suitable preventive measures according to the specific situation, improves the effectiveness of diabetes prevention intervention, and provides a new and efficient method for early warning and prevention of diabetes.

[0035] In one embodiment, if Figure 2 As shown, in step S10, multi-source medical data is generated according to multi-source life data, which specifically includes: S11: performing data preprocessing on the multi-source life data, wherein the data preprocessing includes data integration processing and data cleaning processing to obtain processed multi-source life data.

[0036] Specifically, by integrating data from different channels, such as electronic medical records, physical examination reports, lifestyle surveys, etc., a comprehensive data set containing various health information of patients is formed, which can solve the problems of data format, unit, naming inconsistency and so on that may exist between different data sources, so that the data can be seamlessly connected in subsequent analysis. The integrated comprehensive data set is cleaned to remove duplicate, erroneous, and incomplete data, improve the accuracy and reliability of the data, and thus improve the accuracy of subsequent diabetes risk analysis.

[0037] S12: Performing data conversion processing on the processed multi-source life data to form multi-source medical data.

[0038] Specifically, the processed multi-source life data are converted into a data format suitable for subsequent analysis, such as converting text data into numerical data, converting unstructured data into structured data, and extracting medical data that has an important impact on diabetes risk assessment from the multi-source life data to form multi-source medical data. The multi-source medical data contains data in multiple dimensions such as the patient's medical history, physiological parameters, biochemical indicators, and lifestyle, providing reliable data for subsequent diabetes risk assessment, construction of prediction models, and formulation of personalized prevention plans.

[0039] In one embodiment, if Figure 3 As shown, before collecting multi-source medical data, the method for constructing a diabetes early warning model based on machine learning also includes: S101: During the data collection process, the data access logs recorded in multiple information channels are recorded using blockchain technology to obtain initial patient record data.

[0040] Specifically, in the process of collecting multi-source data of patients, the privacy of patients' personal data is involved. During the data collection process, the differential factor algorithm is used to process the collected data to improve the privacy of patients' personal data. Specifically, blockchain technology is used to record data access logs in multiple information channels to obtain initial patient record data.

[0041] S102: Encrypt and encode the initial patient record data, and add random noise to the encoded initial patient data to blur the individual data of the patient.

[0042] Specifically, the initial patient record data obtained is first encrypted and encoded so that in the subsequent randomization and aggregation steps, random noise is added to the encrypted initial patient record data to blur the patient's individual data, thereby effectively protecting the patient's personal sensitive data during the data collection process.

[0043] S103: Aggregate the initial patient data after adding random noise to form multi-source life data.

[0044] Specifically, the initial patient data after adding random noise is aggregated to eliminate the influence of some noise and form the patient's multi-source life data, thereby ensuring the utility of the collected patient data while minimizing the risk of privacy leakage.

[0045] In one embodiment, if Figure 4 As shown, in step S20, feature selection and extraction are performed on multi-source medical data to obtain basic features of diabetes, and complex features of diabetes are constructed based on basic health features, specifically including: S21: Acquire a data source type based on the multi-source medical data, and determine a suitable feature algorithm according to the data source type.

[0046] Specifically, in the process of feature analysis of the collected multi-source medical data, the multi-source medical data will have different data types. The data source type is obtained according to the multi-source medical data, such as text data, digital data, medical image data and other different data source types. The appropriate feature algorithm determined according to the data source type can efficiently and accurately extract features from each type of data source.

[0047] S22: Extracting features from the multi-source medical data according to a suitable feature algorithm to extract basic features of diabetes and time series features.

[0048] Specifically, feature extraction is performed on multi-source medical data according to a suitable feature algorithm to extract basic diabetes features and time series features. Basic diabetes features refer to basic diabetes features such as blood glucose level and insulin sensitivity. Time series features refer to the time series characteristics of the patient's medical data. For example, blood glucose monitoring data will contain time series characteristics.

[0049] S23: The basic diabetes features and time series features are combined to form a feature set, and the features in the feature set are combined and transformed to form a composite diabetes feature.

[0050] Specifically, the extracted basic features of diabetes and time series features are integrated together, and the time series features are temporally associated with the basic features of diabetes to form a feature set, so as to analyze the changing trends and patterns of medical data about diabetes, and to combine and transform the features in the feature set. For example, the blood sugar level feature is combined with the insulin level feature to form a composite feature of the insulin sensitivity index. Another example is that qualitative features such as family medical history and lifestyle are quantified and combined with quantitative features to form a composite feature of diabetes. The composite feature of diabetes can more comprehensively reflect the patient's physiological state, thereby facilitating a more accurate risk assessment of diabetes.

[0051] In one embodiment, if Figure 5 As shown, in step S30, the initial machine learning model is trained according to the composite features of diabetes to obtain a diabetes analysis model, which specifically includes: S31: performing feature dimensionality reduction processing on the diabetes composite features, screening out diabetes-related composite features, and integrating them to form a training set, a validation set, and a test set.

[0052] Specifically, in the process of constructing the diabetes analysis model, the composite features of diabetes are subjected to feature dimensionality reduction processing, such as using principal component analysis and linear discriminant analysis, which can remove redundant and noise features in the composite features of diabetes, screen out composite features that are highly correlated with diabetes, and integrate the screened composite features into training sets, validation sets, and test sets to ensure the independence and representativeness of the data.

[0053] S32: Determine data characteristics based on the training set, determine a machine learning algorithm based on the data characteristics, and train a preset initial machine learning model based on the machine learning algorithm and the training set.

[0054] S33: Input the test set into the trained machine learning model to determine the optimal model parameters.

[0055] S34: Adjust the trained machine learning model based on the optimal model parameters, input the verification set into the machine learning model after parameter adjustment, and output the model verification result.

[0056] S35: judging whether the initial machine model meets the diabetes risk assessment effect according to the model verification result, and if so, obtaining a diabetes analysis model.

[0057] Specifically, selecting a suitable machine learning algorithm based on the data characteristics of the training set and training the preset initial machine learning model in combination with the training set is conducive to improving the interpretability and stability of the model. The test set is then used to continuously iteratively optimize the machine learning model and adjust the model parameters. Finally, the validation set is used to verify whether the model meets the effect requirements of diabetes risk assessment, thereby completing the construction of the diabetes analysis model to improve the early detection of diabetes.

[0058] In one embodiment, if Figure 6 As shown, in step S40, the patient risk level is generated based on the diabetes prediction result, and a personalized prevention plan is generated according to the patient risk level and combined with the patient's basic information, specifically including: S41: Inputting the collected multi-source medical data into the diabetes analysis model, outputting a binary classification prediction result, and calculating the probability of having diabetes based on the binary classification prediction result.

[0059] Specifically, the multi-source medical data is input into the diabetes analysis model using the trained diabetes analysis model, and a binary prediction result is output, that is, whether the patient has diabetes, so as to achieve accurate prevention and improve the prevention effect. The probability of the patient having diabetes is calculated based on the binary prediction result output by the diabetes analysis model.

[0060] S42: setting a patient risk level according to the probability of having diabetes, and generating a corresponding diabetes prevention plan based on the patient risk level and in combination with the patient's basic information.

[0061] Specifically, based on the probability of a patient suffering from diabetes, the patient's risk level is assessed to form a patient risk level, such as dividing the probability into three levels: low risk, medium risk and high risk. Each level has a corresponding prevention plan. Based on the patient's risk level and combined with the patient's basic information, such as the patient's age, living habits, allergy history and other basic information, a corresponding diabetes prevention plan is generated, thereby achieving the effect of precise prevention of diabetes.

[0062] In one embodiment, if Figure 7 As shown, the method for constructing a diabetes early warning model based on machine learning also includes: S50: Acquire patient monitoring data after adopting the personalized prevention plan in real time, and generate a prevention plan adjustment instruction based on the patient monitoring data.

[0063] S60: Responding to the prevention plan adjustment instruction, generating a new personalized prevention plan.

[0064] Specifically, the patient's monitoring data, such as blood sugar levels, is obtained in real time. Even if the patient's health condition is found to have changed after taking a prevention plan, a prevention plan adjustment instruction is generated based on the monitoring data. The original personalized prevention plan is adjusted in response to the prevention plan adjustment instruction to form a new personalized prevention plan. This can ensure that the prevention plan matches the patient's current condition, thereby improving the accuracy and effectiveness of diabetes prevention.

[0065] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0066] In one embodiment, a diabetes early warning model construction system based on machine learning is provided, and the diabetes early warning model construction system based on machine learning corresponds one to one with the diabetes early warning model construction method based on machine learning in the above embodiment. Figure 8 As shown, the diabetes early warning model construction system based on machine learning includes a patient data collection module, a data feature extraction module, a model construction module and a prevention plan generation module. The detailed description of each functional module is as follows: A patient data collection module, used to obtain multi-source life data based on multiple information channels, including but not limited to electronic medical records, physical examination reports, and lifestyle surveys, and generate multi-source medical data based on the multi-source life data; A data feature extraction module, used to perform feature selection and extraction on the multi-source medical data to obtain basic features of diabetes, and construct composite features of diabetes based on the basic health features; A model building module, used to train an initial machine learning model according to the composite characteristics of diabetes to obtain a diabetes analysis model, and obtain diabetes prediction results based on the diabetes analysis model; The prevention plan generation module is used to generate a patient risk level based on the diabetes prediction result, and generate a personalized prevention plan according to the patient risk level and combined with the patient's basic information.

[0067] Preferably, the patient data collection module includes: For the specific limitations of the diabetes early warning model construction system based on machine learning, please refer to the limitations of the diabetes early warning model construction method based on machine learning above, which will not be repeated here. Each module in the above-mentioned diabetes early warning model construction system based on machine learning can be implemented in whole or in part through software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the electronic device in the form of hardware, or can be stored in the memory of the electronic device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0068] In one embodiment, an electronic device is provided. The electronic device may be a server, and its internal structure diagram may be as follows: Fig. 9 As shown. The electronic device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the electronic device is used to store monitoring data and diabetes analysis models of multi-source patients. The network interface of the electronic device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for constructing a diabetes early warning model based on machine learning is implemented.

[0069] In one embodiment, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program: Acquire multi-source life data based on multiple information channels, the multiple information channels including but not limited to electronic medical records, physical examination reports, and lifestyle surveys, and generate multi-source medical data based on the multi-source life data; Performing feature selection and extraction on the multi-source medical data to obtain basic features of diabetes, and constructing composite features of diabetes based on the basic health features; Training an initial machine learning model according to the composite features of diabetes to obtain a diabetes analysis model, and obtaining a diabetes prediction result based on the diabetes analysis model; A patient risk level is generated based on the diabetes prediction result, and a personalized prevention plan is generated according to the patient risk level and combined with the patient's basic information.

[0070] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented: Acquire multi-source life data based on multiple information channels, the multiple information channels including but not limited to electronic medical records, physical examination reports, and lifestyle surveys, and generate multi-source medical data based on the multi-source life data; Performing feature selection and extraction on the multi-source medical data to obtain basic features of diabetes, and constructing composite features of diabetes based on the basic health features; Training an initial machine learning model according to the composite features of diabetes to obtain a diabetes analysis model, and obtaining a diabetes prediction result based on the diabetes analysis model; A patient risk level is generated based on the diabetes prediction result, and a personalized prevention plan is generated according to the patient risk level and combined with the patient's basic information.

[0071] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0072] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0073] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for constructing a diabetes early warning model based on machine learning, characterized in that: The method for constructing a diabetes early warning model based on machine learning comprises the following steps: Acquire multi-source life data based on multiple information channels, the multiple information channels including but not limited to electronic medical records, physical examination reports, and lifestyle surveys, and generate multi-source medical data based on the multi-source life data; Performing feature selection and extraction on the multi-source medical data to obtain basic features of diabetes, and constructing composite features of diabetes based on the basic health features; Training an initial machine learning model according to the composite features of diabetes to obtain a diabetes analysis model, and obtaining a diabetes prediction result based on the diabetes analysis model; A patient risk level is generated based on the diabetes prediction result, and a personalized prevention plan is generated according to the patient risk level and combined with the patient's basic information.

2. The method for constructing a diabetes early warning model based on machine learning according to claim 1, characterized in that: The generating multi-source medical data according to the multi-source life data specifically includes: Performing data preprocessing on the multi-source life data, wherein the data preprocessing includes data integration processing and data cleaning processing to obtain processed multi-source life data; The processed multi-source life data is subjected to data conversion processing to form multi-source medical data.

3. The method for constructing a diabetes early warning model based on machine learning according to claim 1, characterized in that: Before acquiring multi-source life data based on multiple information channels, wherein the multiple information channels include but are not limited to electronic medical records, physical examination reports, and lifestyle surveys, and generating multi-source medical data based on the multi-source life data, the method for constructing a diabetes early warning model based on machine learning further includes: During the data collection process, the data access logs recorded in multiple information channels using blockchain technology were used to obtain initial patient record data; Encrypting and encoding the initial patient record data, and adding random noise to the encoded initial patient data to blur the individual data of the patient; The initial patient data after adding random noise is aggregated to form multi-source life data.

4. The method for constructing a diabetes early warning model based on machine learning according to claim 1, characterized in that: The feature selection and extraction of the multi-source medical data to obtain the basic features of diabetes, and the construction of the composite features of diabetes based on the basic health features specifically include: Acquire a data source type based on the multi-source medical data, and determine a suitable feature algorithm according to the data source type; Performing feature extraction on the multi-source medical data according to a suitable feature algorithm to extract basic features of diabetes and time series features; The basic features of diabetes and time series features are formed into a feature set, and the features in the feature set are combined and transformed to form a composite feature of diabetes.

5. The method for constructing a diabetes early warning model based on machine learning according to claim 1, characterized in that: The initial machine learning model is trained according to the composite features of diabetes to obtain a diabetes analysis model, specifically comprising: Performing feature dimensionality reduction processing on the diabetes composite features, screening out diabetes-related composite features, and integrating them to form a training set, a validation set, and a test set; Determining data characteristics based on the training set, determining a machine learning algorithm based on the data characteristics, and training a preset initial machine learning model based on the machine learning algorithm and the training set; Inputting the test set into the trained machine learning model to determine the optimal model parameters; Adjust the trained machine learning model based on the optimal model parameters, input the verification set into the machine learning model after parameter adjustment, and output the model verification result; According to the model verification result, it is judged whether the initial machine model meets the diabetes risk assessment effect. If so, a diabetes analysis model is obtained.

6. The method for constructing a diabetes early warning model based on machine learning according to claim 1, characterized in that: The step of generating a patient risk level based on the diabetes prediction result and generating a personalized prevention plan according to the patient risk level and in combination with the patient's basic information specifically includes: Inputting the collected multi-source medical data into the diabetes analysis model, outputting a binary classification prediction result, and calculating the probability of having diabetes based on the binary classification prediction result; The patient risk level is set according to the probability of suffering from diabetes, and a corresponding diabetes prevention plan is generated based on the patient risk level and in combination with the patient's basic information.

7. The method for constructing a diabetes early warning model based on machine learning according to claim 1, characterized in that: After generating the patient risk level based on the diabetes prediction result and generating a personalized prevention plan according to the patient risk level and in combination with the patient's basic information, the method for constructing a diabetes early warning model based on machine learning further includes: Acquire patient monitoring data after adopting personalized prevention plan in real time, and generate prevention plan adjustment instructions based on the patient monitoring data; In response to the prevention plan adjustment instruction, a new personalized prevention plan is generated.

8. A diabetes early warning model construction system based on machine learning, characterized in that: The diabetes early warning model construction system based on machine learning includes: A patient data collection module, used to obtain multi-source life data based on multiple information channels, including but not limited to electronic medical records, physical examination reports, and lifestyle surveys, and generate multi-source medical data based on the multi-source life data; A data feature extraction module, used to perform feature selection and extraction on the multi-source medical data to obtain basic features of diabetes, and construct composite features of diabetes based on the basic health features; A model building module, used to train an initial machine learning model according to the composite characteristics of diabetes to obtain a diabetes analysis model, and obtain diabetes prediction results based on the diabetes analysis model; The prevention plan generation module is used to generate a patient risk level based on the diabetes prediction result, and generate a personalized prevention plan according to the patient risk level and combined with the patient's basic information.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method for constructing a diabetes early warning model based on machine learning as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of a method for constructing a diabetes early warning model based on machine learning as described in any one of claims 1 to 7 are implemented.

Images