Machine learning based chronic disease prediction system

Abstract

The application discloses a chronic disease prediction system based on machine learning, comprising a data acquisition module, a feature extraction module, a chronic disease prediction module and a hospitalization prediction module; the data acquisition module is used for collecting chronic disease related data and environmental parameters of each chronic disease prediction area, obtaining a per capita chronic disease distribution map in a first time period, preprocessing the per capita chronic disease distribution map, and generating a first chronic disease prediction image; the feature extraction module is used for obtaining a first chronic disease element and a second chronic disease element, and obtaining corresponding moving features and spatial features; the chronic disease prediction module is used for constructing a chronic disease element moving virtual track according to the moving features, and constructing a chronic disease prediction model for predicting the evolution of chronic diseases based on the chronic disease element moving virtual track; and the hospitalization prediction module is used for predicting hospitalization data corresponding to the evolution of chronic diseases, and providing a scientific basis for chronic disease early warning and planning.

Description

Technical Field

[0001] This invention relates to the field of machine learning technology, and in particular to a machine learning-based chronic disease prediction system. Background Technology

[0002] In modern society, the emergence and development of chronic diseases have become a focus of global public health attention. Chronic diseases, such as heart disease, diabetes, and asthma, were once considered problems of wealthy countries. With their persistent and progressive nature, chronic diseases continuously erode human health. High treatment costs, long-term disease management pressures, and numerous related complications place a significant economic burden on families and society. Especially driven by globalization and modernization, unhealthy lifestyles and the aging global population have exacerbated the threat of chronic diseases to human health. Chronic diseases also have profound impacts on individuals, potentially leading to loss of bodily function, decreased quality of life, and even the risk of unemployment and family breakdown. These serious consequences have a significant impact on individuals' psychological and social lives. Against this backdrop, the prevention and control of chronic diseases is particularly important and urgent.

[0003] Currently, Chinese invention patent CN115188470B discloses a multi-chronic disease prediction system based on a multi-task Cox learning model. This system includes a model building module for constructing a multi-task Cox learning model, which incorporates a regularization term to achieve parameter sharing across multiple chronic disease prediction tasks; a training and optimization module for acquiring a joint dataset of multiple chronic disease variables to train the multi-task Cox learning model using a multi-task learning method and optimizing it using an approximate gradient descent method; and a risk assessment module for defining risk indicators, acquiring the subjects' biometrics, and obtaining risk indicator values ​​based on the optimized multi-task Cox learning model to assess the subjects' disease risk. However, this related technology lacks a comprehensive representation of the movement and spatial patterns of chronic disease evolution, fails to predict corresponding hospitalization data based on these patterns, lacks dynamism and continuity, and does not calculate chronic disease risk based on comprehensive data, potentially leading to a loss of medical resources for chronic disease treatment. Summary of the Invention

[0004] The technical problem solved by this invention is that related technologies do not comprehensively characterize the movement and spatial evolution patterns of chronic diseases, do not predict corresponding hospitalization data based on the movement and spatial evolution patterns of chronic diseases, lack dynamism and continuity, and do not calculate chronic disease risk based on comprehensive data, which can easily lead to the loss of medical resources for chronic disease treatment.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a chronic disease prediction system based on machine learning, including a data acquisition module, a feature extraction module, a chronic disease prediction module, and a hospitalization prediction module; The data acquisition module is used to collect historical chronic disease data, corresponding hospitalization history data, environmental parameters, and historical evolution of chronic diseases from medical institutions, community surveys, and laboratory storage in various chronic disease prediction areas. The chronic disease prediction area is divided into several chronic disease prediction sub-regions, and the chronic disease prediction sub-regions are numbered. The per capita chronic disease distribution map of the chronic disease prediction sub-regions in the first time period is obtained. The per capita chronic disease distribution map is preprocessed to generate the first chronic disease prediction image. The feature extraction module is used to extract the color values ​​of the first chronic disease prediction image, select the part corresponding to the largest color value, take the part corresponding to the largest color value as the first chronic disease element, take the part corresponding to the smallest color value as the second chronic disease element, analyze the first chronic disease element and the second chronic disease element of the same chronic disease prediction sub-region within the first time period to obtain the movement features of the first chronic disease element and the second chronic disease element, and analyze the first chronic disease element and the second chronic disease element of different chronic disease prediction sub-regions within the first time period to obtain the spatial features of the first chronic disease element and the second chronic disease element. The chronic disease prediction module is used to construct virtual trajectories of chronic disease elements based on mobility characteristics, construct a chronic disease prediction model based on the virtual trajectories of chronic disease elements to predict the evolution of chronic diseases, and test the chronic disease prediction model. The hospitalization prediction module is used to construct a hospitalization prediction model. The hospitalization prediction model takes the historical evolution of chronic diseases as input and the historical hospitalization data as output. The hospitalization prediction model is trained based on a regression function to predict the hospitalization data corresponding to the evolution of chronic diseases.

[0006] In a preferred embodiment of the machine learning-based chronic disease prediction system described in this invention, the chronic disease prediction region is divided into several chronic disease prediction sub-regions, and the chronic disease prediction sub-regions are numbered P. i ; Obtain the per capita chronic disease distribution map of the chronic disease prediction sub-region within the first time period. Perform segmentation, enhancement, and filtering processing on the per capita chronic disease distribution map to generate a first chronic disease prediction image. Number the first chronic disease prediction image as P. ij。

[0007] As a preferred embodiment of the machine learning-based chronic disease prediction system described in this invention, the specific process of obtaining chronic disease elements includes: A first chronic disease prediction image is obtained, the color values ​​of the first chronic disease prediction image are extracted, the color values ​​are sorted in descending order, the largest and smallest color values ​​are selected, the color range value of the chronic disease prediction sub-region is obtained, the part corresponding to the smallest color value is obtained, the part corresponding to the smallest color value is selected as the second chronic disease element, the part corresponding to the largest color value is obtained, and the part corresponding to the largest color value is selected as the first chronic disease element. The coordinates of the first chronic disease element and the second chronic disease element in the same chronic disease prediction sub-region within the first time period are obtained. The coordinates of the first chronic disease element in the same chronic disease prediction sub-region are compared to obtain the movement characteristics of the first chronic disease element. The coordinates of the second chronic disease element in the same chronic disease prediction sub-region are compared to obtain the movement characteristics of the second chronic disease element. The coordinates of the first chronic disease element and the second chronic disease element in different chronic disease prediction sub-regions within the first time period are obtained. The coordinates of the first chronic disease element in the different chronic disease prediction sub-regions are compared to obtain the spatial features of the first chronic disease element. The coordinates of the second chronic disease element in the different chronic disease prediction sub-regions are compared to obtain the spatial features of the second chronic disease element. The movement characteristics include movement speed and movement direction; The spatial features include the types of chronic disease evolution and the number of chronic disease intersection points.

[0008] As a preferred embodiment of the machine learning-based chronic disease prediction system of the present invention, the method for constructing the virtual trajectory of chronic disease elements specifically includes: Starting from the coordinates of the first chronic disease element in the same initial chronic disease prediction sub-region, the time of the same first chronic disease element in the same chronic disease prediction sub-region is recorded as T1, and the time of the neighboring first chronic disease element in the corresponding chronic disease prediction sub-region is recorded as T2. Chronic disease interval = T2 - T 1， The chronic disease interval time is recorded as the first interval, and the first distance is obtained by subtracting the color value corresponding to T1 from the color value corresponding to T2. Divide the first distance by the first interval to obtain the moving speed; A virtual trajectory for the movement of chronic disease elements is constructed based on the movement direction and speed.

[0009] As a preferred embodiment of the machine learning-based chronic disease prediction system of the present invention, the environmental parameters, historical data of chronic diseases, and virtual trajectories and spatial features of chronic disease elements in the chronic disease prediction sub-region are obtained respectively. The environmental parameters include air quality, average daily temperature, and average daily humidity. The environmental parameters, historical data of chronic diseases, virtual trajectories of chronic disease elements, and spatial characteristics are used as independent variables in the multiple regression analysis. The weights of environmental parameters, historical data of chronic diseases, and the virtual trajectories and spatial characteristics of the movement of chronic disease elements are calculated based on multiple regression analysis, and a chronic disease prediction model is constructed. The output of the chronic disease prediction model is the evolution of chronic diseases.

[0010] As a preferred embodiment of the machine learning-based chronic disease prediction system of the present invention, the environmental parameters of the chronic disease prediction sub-region, historical data of chronic diseases, and the average chronic disease rate per capita in the first time period are used as a sample set, and the sample set is divided into a training set and a test set according to a certain ratio. A chronic disease prediction model is trained to predict the evolution of chronic diseases. The chronic disease prediction model is trained on the actual evolution of chronic diseases. The chronic disease prediction model is tested until the accuracy reaches a first threshold.

[0011] As a preferred embodiment of the machine learning-based chronic disease prediction system described in this invention, the method for constructing a hospitalization prediction model specifically includes: Data on the historical evolution of chronic diseases and corresponding hospitalization histories were obtained from medical institutions, community surveys, and laboratories in areas where chronic diseases are predicted. The historical evolution of chronic diseases is used as the input to the regression function of the hospitalization prediction model, and the corresponding historical hospitalization data is used as the output of the regression function to obtain the corresponding error coefficient. A hospitalization prediction model is trained to predict hospitalization data. The hospitalization prediction model is trained using actual historical hospitalization data as the training target. The hospitalization prediction model is tested until the error coefficient is less than a preset error coefficient threshold.

[0012] As a preferred embodiment of the machine learning-based chronic disease prediction system of the present invention, the historical data of chronic diseases includes first indicator data and second indicator data. The first indicator data includes patient age, body mass index, family history, and blood lipid levels; The second indicator data includes electronic medical records, drug data, laboratory indicator data, and imaging data.

[0013] As a preferred embodiment of the machine learning-based chronic disease prediction system of the present invention, the chronic disease evolution status includes no chronic disease, types of chronic diseases, number of chronic diseases and corresponding evolution degree, wherein the evolution degree includes first-level evolution degree, second-level evolution degree and third-level evolution degree.

[0014] As a preferred embodiment of the machine learning-based chronic disease prediction system described in this invention, the hospitalization data includes hospitalization duration, number of re-hospitalizations, and ward level.

[0015] The beneficial effects of this invention are as follows: By dividing each chronic disease prediction area into chronic disease prediction sub-regions and numbering them, the distribution of chronic diseases in different regions can be analyzed and compared more precisely. A per capita chronic disease distribution map of each chronic disease prediction sub-region within the first time period can be obtained, providing spatial information on chronic disease distribution. The per capita chronic disease distribution map is preprocessed to generate a first chronic disease prediction image, providing visualization and basic data for subsequent analysis. Analyzing the movement characteristics and spatial distribution characteristics of chronic disease elements within the same chronic disease prediction sub-region and between different chronic disease prediction sub-regions helps to understand the evolutionary dynamics and regional differences of chronic disease events. The weights of environmental parameters, virtual movement trajectories, and spatial characteristics are calculated to construct a chronic disease prediction model. The output of the chronic disease prediction model is used as the input of a hospitalization prediction model to obtain corresponding hospitalization data for chronic disease evolution. This can bring convenience and predictability to hospitalization arrangements for chronic disease patients, providing a scientific basis for chronic disease early warning and planning. The trained chronic disease prediction model and hospitalization prediction model are tested to evaluate their prediction accuracy and reliability, providing reliable chronic disease prediction services for practical applications. Attached Figure Description

[0016] Figure 1 This is a basic flowchart of a machine learning-based chronic disease prediction system provided in one embodiment of the present invention. Detailed Implementation

[0017] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0018] Example, refer to Figure 1 As an embodiment of the present invention, a machine learning-based chronic disease prediction system is provided, including a data acquisition module, a feature extraction module, a chronic disease prediction module, and a hospitalization prediction module; The data acquisition module is used to collect historical chronic disease data, corresponding hospitalization history data, environmental parameters, and historical evolution of chronic diseases from medical institutions, community surveys, and laboratory storage in various chronic disease prediction areas. The chronic disease prediction area is divided into several chronic disease prediction sub-regions, and the chronic disease prediction sub-regions are numbered. The per capita chronic disease distribution map of the chronic disease prediction sub-regions in the first time period is obtained. The per capita chronic disease distribution map is preprocessed to generate the first chronic disease prediction image. The feature extraction module is used to extract the color values ​​of the first chronic disease prediction image, select the part corresponding to the largest color value, take the part corresponding to the largest color value as the first chronic disease element, take the part corresponding to the smallest color value as the second chronic disease element, analyze the first chronic disease element and the second chronic disease element of the same chronic disease prediction sub-region within the first time period to obtain the movement features of the first chronic disease element and the second chronic disease element, and analyze the first chronic disease element and the second chronic disease element of different chronic disease prediction sub-regions within the first time period to obtain the spatial features of the first chronic disease element and the second chronic disease element. The chronic disease prediction module is used to construct virtual trajectories of chronic disease elements based on mobility characteristics, construct a chronic disease prediction model based on the virtual trajectories of chronic disease elements to predict the evolution of chronic diseases, and test the chronic disease prediction model. The hospitalization prediction module is used to construct a hospitalization prediction model. The hospitalization prediction model takes the historical evolution of chronic diseases as input and the historical hospitalization data as output. The hospitalization prediction model is trained based on a regression function to predict the hospitalization data corresponding to the evolution of chronic diseases.

[0019] This invention divides and numbers various chronic disease prediction areas into chronic disease prediction sub-regions, enabling more refined analysis and comparison of chronic disease distribution in different regions. It obtains a per capita chronic disease distribution map for each sub-region within a first time period, providing spatial information on chronic disease distribution. Preprocessing the per capita chronic disease distribution map generates a first chronic disease prediction image, providing visualization and foundational data for subsequent analysis. Analyzing the movement and spatial distribution characteristics of chronic disease elements within the same and different sub-regions helps understand the evolutionary dynamics and regional differences of chronic disease events. Calculating the weights of environmental parameters, virtual movement trajectories, and spatial characteristics allows for the construction of a chronic disease prediction model. The output of this model is used as input to a hospitalization prediction model, yielding corresponding hospitalization data based on chronic disease evolution. This provides convenience and predictability for hospitalization arrangements for chronic disease patients, offering a scientific basis for chronic disease early warning and planning. Testing the trained chronic disease prediction model and hospitalization prediction model evaluates their prediction accuracy and reliability, providing credible chronic disease prediction services for practical applications.

[0020] The chronic disease prediction region is divided into several chronic disease prediction sub-regions, and these sub-regions are numbered P. i ; Obtain the per capita chronic disease distribution map of the chronic disease prediction sub-region within the first time period. Perform segmentation, enhancement, and filtering processing on the per capita chronic disease distribution map to generate a first chronic disease prediction image. Number the first chronic disease prediction image as P. ij。

[0021] In this embodiment, by dividing the chronic disease prediction area into multiple chronic disease prediction sub-regions and numbering them, the chronic disease situation in different regions can be analyzed and compared more precisely. The specific process of obtaining chronic disease elements includes: A first chronic disease prediction image is obtained, the color values ​​of the first chronic disease prediction image are extracted, the color values ​​are sorted in descending order, the largest and smallest color values ​​are selected, the color range value of the chronic disease prediction sub-region is obtained, the part corresponding to the smallest color value is obtained, the part corresponding to the smallest color value is selected as the second chronic disease element, the part corresponding to the largest color value is obtained, and the part corresponding to the largest color value is selected as the first chronic disease element. The coordinates of the first chronic disease element and the second chronic disease element in the same chronic disease prediction sub-region within the first time period are obtained. The coordinates of the first chronic disease element in the same chronic disease prediction sub-region are compared to obtain the movement characteristics of the first chronic disease element. The coordinates of the second chronic disease element in the same chronic disease prediction sub-region are compared to obtain the movement characteristics of the second chronic disease element. The coordinates of the first chronic disease element and the second chronic disease element in different chronic disease prediction sub-regions within the first time period are obtained. The coordinates of the first chronic disease element in the different chronic disease prediction sub-regions are compared to obtain the spatial features of the first chronic disease element. The coordinates of the second chronic disease element in the different chronic disease prediction sub-regions are compared to obtain the spatial features of the second chronic disease element. The movement characteristics include movement speed and movement direction; The spatial features include the types of chronic disease evolution and the number of chronic disease intersection points.

[0022] In this embodiment, the number of chronic disease intersection points determines the number of chronic disease types in the evolution of chronic diseases. The chronic disease elements are divided into first chronic disease elements and second chronic disease elements. Analyzing the movement characteristics and spatial distribution characteristics of the first and second chronic disease elements within the same chronic disease prediction sub-region and between different chronic disease prediction sub-regions helps to understand the evolution dynamics and regional differences of chronic diseases.

[0023] The specific methods for constructing virtual trajectories of chronic disease elements include: Starting from the coordinates of the first chronic disease element in the same initial chronic disease prediction sub-region, the time of the same first chronic disease element in the same chronic disease prediction sub-region is recorded as T1, and the time of the neighboring first chronic disease element in the corresponding chronic disease prediction sub-region is recorded as T2. Chronic disease interval = T2 - T 1，The chronic disease interval time is recorded as the first interval, and the first distance is obtained by subtracting the color value corresponding to T1 from the color value corresponding to T2. Divide the first distance by the first interval to obtain the moving speed; A virtual trajectory for the movement of chronic disease elements is constructed based on the movement direction and speed.

[0024] In this embodiment, the virtual trajectories of the first chronic disease element and the second chronic disease element are different. The greater the difference in color value of the corresponding elements and the smaller the interval time, the faster the evolution of the chronic disease element. Constructing the virtual trajectories of the chronic disease element can enhance the predictive ability and accuracy of the chronic disease prediction model.

[0025] Environmental parameters, historical data of chronic diseases, and virtual trajectories and spatial characteristics of chronic disease elements are obtained for the chronic disease prediction sub-regions respectively. The environmental parameters include air quality, average daily temperature, and average daily humidity. The environmental parameters, historical data of chronic diseases, virtual trajectories of chronic disease elements, and spatial characteristics are used as independent variables in the multiple regression analysis. The weights of environmental parameters, historical data of chronic diseases, and the virtual trajectories and spatial characteristics of the movement of chronic disease elements are calculated based on multiple regression analysis, and a chronic disease prediction model is constructed. The output of the chronic disease prediction model is the evolution of chronic diseases.

[0026] In this embodiment, the weights of environmental parameters, historical data of chronic diseases, and the virtual trajectories and spatial characteristics of the movement of chronic disease elements are calculated by multivariate regression analysis using simulation software. A chronic disease prediction model is then constructed to predict the evolution of chronic diseases, providing a scientific basis for chronic disease early warning and planning.

[0027] The environmental parameters, historical data of chronic diseases, and per capita chronic disease rate in the first time period of the chronic disease prediction sub-region are used as the sample set, and the sample set is divided into a training set and a test set according to a certain ratio. A chronic disease prediction model is trained to predict the evolution of chronic diseases. The chronic disease prediction model is trained on the actual evolution of chronic diseases. The chronic disease prediction model is tested until the accuracy reaches a first threshold.

[0028] In this embodiment, the first threshold is 0.95, and the sample set is divided into a training set and a test set in a 7:3 ratio. Different air quality, average daily temperature and average daily humidity correspond to different evolution speeds, evolution directions and evolution types of chronic diseases. This can fully take into account the impact of environmental factors on chronic diseases, making the chronic disease prediction model more comprehensive and accurate in predicting the evolution of chronic diseases.

[0029] The specific methods for constructing hospitalization prediction models include: Data on the historical evolution of chronic diseases and corresponding hospitalization histories were obtained from medical institutions, community surveys, and laboratories in areas where chronic diseases are predicted. The historical evolution of chronic diseases is used as the input to the regression function of the hospitalization prediction model, and the corresponding historical hospitalization data is used as the output of the regression function to obtain the corresponding error coefficient. A hospitalization prediction model is trained to predict hospitalization data. The hospitalization prediction model is trained using actual historical hospitalization data as the training target. The hospitalization prediction model is tested until the error coefficient is less than a preset error coefficient threshold.

[0030] In this embodiment, the error coefficient threshold is 0.1. The chronic disease evolution output by the chronic disease prediction model is used as the input of the hospitalization prediction model to obtain the length of hospital stay, number of re-hospitalizations, and ward level corresponding to the chronic disease evolution.

[0031] The historical data on chronic diseases includes data on the first indicator and data on the second indicator. The first indicator data includes patient age, body mass index, family history, and blood lipid levels; The second indicator data includes electronic medical records, drug data, laboratory indicator data, and imaging data.

[0032] The evolution of chronic diseases includes the absence of chronic diseases, the types of chronic diseases, the number of chronic diseases, and the corresponding degree of evolution. The degree of evolution includes primary evolution, secondary evolution, and tertiary evolution.

[0033] In this embodiment, the severity of chronic diseases is ordered as follows: first-level evolution is less than second-level evolution, and second-level evolution is less than third-level evolution.

[0034] The hospitalization data includes length of stay, number of hospitalizations, and ward level.

[0035] In this embodiment, different chronic disease evolution states correspond to different hospitalization durations, number of hospitalizations, and ward levels. The more severe the chronic disease evolution state, the longer the hospitalization duration, the more hospitalizations, and the higher the ward level. The ward levels include Level 1 nursing wards, Level 2 nursing wards, and Level 3 nursing wards.

[0036] By dividing and numbering various chronic disease prediction areas into sub-regions, the distribution of chronic diseases in different regions can be analyzed and compared more precisely. A per capita chronic disease distribution map for each sub-region within the first time period can be obtained, providing spatial information on chronic disease distribution. Preprocessing the per capita chronic disease distribution map generates the first chronic disease prediction image, providing visualization and foundational data for subsequent analysis. Analyzing the movement and spatial distribution characteristics of chronic disease elements within the same sub-region and between different sub-regions helps to understand the evolutionary dynamics and regional differences of chronic disease events. The weights of environmental parameters, virtual movement trajectories, and spatial characteristics are calculated to construct a chronic disease prediction model. The output of this model is used as the input to a hospitalization prediction model, yielding corresponding hospitalization data based on chronic disease evolution. This provides convenience and predictability for hospitalization arrangements for chronic disease patients, offering a scientific basis for chronic disease early warning and planning. The trained chronic disease prediction model and hospitalization prediction model are tested to evaluate their prediction accuracy and reliability, providing credible chronic disease prediction services for practical applications.

[0037] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0038] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the technical solutions of the present invention.

Claims

1. A machine learning-based chronic disease prediction system, characterized in that, include: Data acquisition module, feature extraction module, chronic disease prediction module, and hospitalization prediction module; The data acquisition module is used to collect historical chronic disease data, corresponding hospitalization history data, environmental parameters, and historical evolution of chronic diseases from medical institutions, community surveys, and laboratory storage in various chronic disease prediction areas. The chronic disease prediction area is divided into several chronic disease prediction sub-regions, and the chronic disease prediction sub-regions are numbered. The per capita chronic disease distribution map of the chronic disease prediction sub-regions in the first time period is obtained. The per capita chronic disease distribution map is preprocessed to generate the first chronic disease prediction image. The feature extraction module is used to extract the color values ​​of the first chronic disease prediction image, select the part corresponding to the largest color value, take the part corresponding to the largest color value as the first chronic disease element, take the part corresponding to the smallest color value as the second chronic disease element, analyze the first chronic disease element and the second chronic disease element in the same chronic disease prediction sub-region within the first time period to obtain the movement features of the first chronic disease element and the second chronic disease element, and analyze the first chronic disease element and the second chronic disease element in different chronic disease prediction sub-regions within the first time period to obtain the spatial features of the first chronic disease element and the second chronic disease element. The chronic disease prediction module is used to construct virtual trajectories of chronic disease elements based on mobility characteristics, construct a chronic disease prediction model based on the virtual trajectories of chronic disease elements to predict the evolution of chronic diseases, and test the chronic disease prediction model. The hospitalization prediction module is used to construct a hospitalization prediction model. The hospitalization prediction model takes the historical evolution of chronic diseases as input and the historical hospitalization data as output. The hospitalization prediction model is trained based on a regression function to predict the hospitalization data corresponding to the evolution of chronic diseases.

2. The machine learning-based chronic disease prediction system as described in claim 1, characterized in that: The chronic disease prediction region is divided into several chronic disease prediction sub-regions, and these sub-regions are numbered P. i ; Obtain the per capita chronic disease distribution map of the chronic disease prediction sub-region within the first time period. Perform segmentation, enhancement, and filtering processing on the per capita chronic disease distribution map to generate a first chronic disease prediction image. Number the first chronic disease prediction image as P. ij。 3. The machine learning-based chronic disease prediction system as described in claim 2, characterized in that: The specific process of obtaining chronic disease elements includes: A first chronic disease prediction image is obtained, the color values ​​of the first chronic disease prediction image are extracted, the color values ​​are sorted in descending order, the largest and smallest color values ​​are selected, the color range value of the chronic disease prediction sub-region is obtained, the part corresponding to the smallest color value is obtained, the part corresponding to the smallest color value is selected as the second chronic disease element, the part corresponding to the largest color value is obtained, and the part corresponding to the largest color value is selected as the first chronic disease element. The coordinates of the first chronic disease element and the second chronic disease element in the same chronic disease prediction sub-region within the first time period are obtained. The coordinates of the first chronic disease element in the same chronic disease prediction sub-region are compared to obtain the movement characteristics of the first chronic disease element. The coordinates of the second chronic disease element in the same chronic disease prediction sub-region are compared to obtain the movement characteristics of the second chronic disease element. The coordinates of the first chronic disease element and the second chronic disease element in different chronic disease prediction sub-regions within the first time period are obtained. The coordinates of the first chronic disease element in the different chronic disease prediction sub-regions are compared to obtain the spatial features of the first chronic disease element. The coordinates of the second chronic disease element in the different chronic disease prediction sub-regions are compared to obtain the spatial features of the second chronic disease element. The movement characteristics include movement speed and movement direction; The spatial features include the types of chronic disease evolution and the number of chronic disease intersection points.

4. The chronic disease prediction system based on machine learning as described in claim 3, characterized in that: The specific methods for constructing virtual trajectories of chronic disease elements include: Starting from the coordinates of the first chronic disease element in the same initial chronic disease prediction sub-region, the time of the same first chronic disease element in the same chronic disease prediction sub-region is recorded as T1, and the time of the neighboring first chronic disease element in the corresponding chronic disease prediction sub-region is recorded as T2. Chronic disease interval = T2 - T 1， The chronic disease interval time is recorded as the first interval, and the first distance is obtained by subtracting the color value corresponding to T1 from the color value corresponding to T2. Divide the first distance by the first interval to obtain the moving speed; A virtual trajectory for the movement of chronic disease elements is constructed based on the movement direction and speed.

5. The machine learning-based chronic disease prediction system as described in claim 4, characterized in that: Environmental parameters, historical data of chronic diseases, and virtual trajectories and spatial characteristics of chronic disease elements are obtained for the chronic disease prediction sub-regions respectively. The environmental parameters include air quality, average daily temperature, and average daily humidity. The environmental parameters, historical data of chronic diseases, virtual trajectories of chronic disease elements, and spatial characteristics are used as independent variables in the multiple regression analysis. The weights of environmental parameters, historical data of chronic diseases, and the virtual trajectories and spatial characteristics of the movement of chronic disease elements are calculated based on multiple regression analysis, and a chronic disease prediction model is constructed. The output of the chronic disease prediction model is the evolution of chronic diseases.

6. The machine learning-based chronic disease prediction system as described in claim 5, characterized in that: The environmental parameters, historical data of chronic diseases, and per capita chronic disease rate in the first time period of the chronic disease prediction sub-region are used as the sample set, and the sample set is divided into a training set and a test set according to a certain ratio. A chronic disease prediction model is trained to predict the evolution of chronic diseases. The chronic disease prediction model is trained on the actual evolution of chronic diseases. The chronic disease prediction model is tested until the accuracy reaches a first threshold.

7. The machine learning-based chronic disease prediction system as described in claim 6, characterized in that: The specific methods for constructing hospitalization prediction models include: Data on the historical evolution of chronic diseases and corresponding hospitalization histories were obtained from medical institutions, community surveys, and laboratories in areas where chronic diseases are predicted. The historical evolution of chronic diseases is used as the input to the regression function of the hospitalization prediction model, and the corresponding historical hospitalization data is used as the output of the regression function to obtain the corresponding error coefficient. A hospitalization prediction model is trained to predict hospitalization data. The hospitalization prediction model is trained using actual historical hospitalization data as the training target. The hospitalization prediction model is tested until the error coefficient is less than a preset error coefficient threshold.

8. The machine learning-based chronic disease prediction system as described in claim 7, characterized in that: The historical data on chronic diseases includes data on the first indicator and data on the second indicator. The first indicator data includes patient age, body mass index, family history, and blood lipid levels; The second indicator data includes electronic medical records, drug data, laboratory indicator data, and imaging data.

9. The machine learning-based chronic disease prediction system as described in claim 8, characterized in that: The evolution of chronic diseases includes the absence of chronic diseases, the types of chronic diseases, the number of chronic diseases, and the corresponding degree of evolution. The degree of evolution includes primary evolution, secondary evolution, and tertiary evolution.

10. The machine learning-based chronic disease prediction system as described in claim 9, characterized in that: The hospitalization data includes length of stay, number of hospitalizations, and ward level.

Images