CN-122000075-A - Calculation simulation disease risk prediction analysis system based on big data

CN122000075ACN 122000075 ACN122000075 ACN 122000075ACN-122000075-A

Abstract

The invention discloses a calculation simulation disease risk prediction analysis system based on big data, which relates to the technical field of medical health information and comprises a data fusion processing module, a data analysis module and a data analysis module, wherein the data fusion processing module is used for receiving and processing heterogeneous health data from a plurality of data sources; and the patient dynamic modeling module is connected with the data fusion processing module and is used for constructing and initializing the individual state representation of the patient based on the processed data. The calculation simulation disease risk prediction analysis system based on big data breaks through the constraint of the traditional prediction model on the strict limitation of the data format and the Markov assumption, can integrate data in different forms such as time sequence monitoring, images, genome and text medical records, and the like, characterizes complex association under a unified graph structure frame, further simulates a disease nonlinear evolution process with long-term dependence and multi-factor interaction, and enhances the utilization depth of real world complex health information and the reality of disease dynamics simulation.

Inventors

PU JIE
HUANG LIYUAN
WANG XUELI
CHEN FAN
LI SHANSHAN

Assignees

淮安市第二人民医院

Dates

Publication Date: 20260508
Application Date: 20260128

Claims (10)

1. A big data based computational simulation disease risk prediction analysis system, comprising: The data fusion processing module is used for receiving and processing heterogeneous health data from a plurality of data sources; the patient dynamic modeling module is connected with the data fusion processing module and is used for constructing and initializing individual state representation of the patient based on the processed data; The disease dynamic simulation engine is connected with the patient dynamic modeling module, adopts a state transition model with a learnable rule and is used for carrying out multi-step forward deduction on the individual state representation of the patient and simulating the dynamic evolution process of the disease state; And the online learning and risk quantifying module is connected with the disease dynamic simulation engine and is used for carrying out incremental updating on model parameters of the disease dynamic simulation engine according to newly input real-time health data and calculating and outputting dynamic risk indexes based on deduction results.
2. The system for predicting and analyzing risk of computational simulation diseases based on big data according to claim 1, wherein the data fusion processing module specifically comprises: The first data processing unit is used for extracting characteristics of the time sequence monitoring data; A second data processing unit for performing feature extraction on the genomics data; the third data processing unit is used for extracting the characteristics of the medical image data; the fourth data processing unit is used for extracting characteristics of the text medical record data; And the data fusion unit is connected with the first data processing unit, the second data processing unit, the third data processing unit and the fourth data processing unit and is used for aligning and fusing the extracted various features.
3. The computational simulated disease risk prediction analysis system based on big data of claim 2, wherein the patient dynamic modeling module constructs the patient individual state representation by: abstracting various health entities of a patient into nodes in a graph structure, wherein the health entities comprise organ entities, biochemical index entities and gene locus entities; establishing edges between the nodes based on medical knowledge or data-driven dependencies to represent interactions or associations between entities; And mapping the features extracted and fused by the data fusion processing module to corresponding nodes in the graph structure to form a patient individual state graph carrying space-time and semantic information.
4. The computational simulation of disease risk prediction analysis system based on big data of claim 3, wherein the patient dynamic modeling module comprises a space-time diagram neural network encoder for: processing time sequence characteristics through a time convolution network, and updating the state of a corresponding node; updating the association strength among nodes by focusing on the information of the network aggregation node neighbors through the graph; the structured graph representation is output as the patient individual status representation.
5. The system for predicting and analyzing risk of a computational simulation disease based on big data as set forth in claim 4, wherein said disease dynamic simulation engine comprises a memory-enhanced cellular automaton simulation layer; The memory-enhanced cellular automaton simulation layer maps nodes or node clusters in the patient individual state diagram into cells; The state evolution rule of each cell is dynamically generated by a rule generating network; And when the rule generating network simulates each step, receiving the historical state sequence of the corresponding cell and the current state information of the neighbor cell thereof, and comprehensively generating a state transition instruction of the cell at the next moment.
6. The system for analysis of computational simulation of disease risk prediction based on big data of claim 5 wherein the rule generating network is a recurrent neural network with an attention mechanism; the memory unit of the cyclic neural network is used for storing and reading long-term history state information of the cells; The attention mechanism is used to calculate and focus neighbor cell information that has a greater impact on the current state transition.
7. The system for predicting and analyzing risk of computational simulation diseases based on big data as set forth in claim 6, wherein the online learning and risk quantification module is specifically configured to: Taking model parameters of the disease dynamic simulation engine as random variables to construct a Bayes posterior estimation framework; Triggering a parameter sampling process when new real-time health data of a patient are received, and updating posterior distribution of the model parameters; controlling the disease dynamic simulation engine to quickly deduce based on the updated parameter distribution, and generating a plurality of possible future state evolution tracks; Based on the multi-state evolution track, counting the occurrence frequency of the target disease state in a specific time window, and generating a dynamic risk probability curve.
8. The system for predicting and analyzing the risk of a computational simulation disease based on big data as set forth in claim 7, wherein the online learning and risk quantification module further comprises a risk visualization unit for displaying the dynamic risk probability curve in correspondence with a time axis and highlighting a period of time during which the risk exceeds a preset threshold.
9. The system for analyzing the prediction of the risk of a computational simulation disease based on big data according to claim 8, wherein the system further comprises an interpretability analysis module which is connected with the disease dynamic simulation engine and the online learning and risk quantification module; The interpretability analysis module is used for recording the attention weight distribution of the rule generation network and correlating back to the original health entity to identify the historical moment and key health index which have significant influence on the risk evolution in the simulation process.
10. A disease risk prediction analysis method employing the system of any one of claims 1 to 9, comprising the steps of: s1, receiving and fusing multi-source heterogeneous patient historical health data and real-time monitoring data; S2, constructing fused data into a patient individual state diagram by using a space-time diagram neural network encoder; s3, performing multi-step state deduction simulation on the individual state diagram of the patient by using a memory-enhanced cellular automaton simulation layer; s4, under the Bayesian framework, carrying out online incremental update on parameters of the simulation layer according to the newly received real-time data; S5, calculating and outputting dynamic disease risk prediction information in a future time period based on a simulation deduction result after updating parameters; and S6, providing a visual interface of risk prediction information and an interpretability analysis of key influence factors.

Description

Calculation simulation disease risk prediction analysis system based on big data Technical Field The invention relates to the technical field of medical health information, in particular to a calculation simulation disease risk prediction analysis system based on big data. Background Along with the rapid development of medical health informatization, big data and artificial intelligence technology are increasingly widely applied in the fields of disease prediction and health management. The disease risk prediction analysis system can provide quantitative evaluation of disease development trend for individuals or groups through integration and modeling of massive medical data, and assist clinical decision and health intervention. Currently, disease risk prediction relies mainly on statistical models, machine learning methods, and dynamic modeling techniques based on markov chains. For example, publication number "CN113724868a" describes "a continuous time markov chain-based chronic disease prediction method" that predicts the trend of chronic disease and the average life of a patient by constructing a health state transition rate matrix and estimating the transition rate based on bayesian theory. The invention can describe the continuous evolution process of the disease state to a certain extent, and improves the time precision of prediction. However, the method still has the following limitations that firstly, the modeling process depends on specific types of medical data (such as state transition times and stay time), the requirements on the structuring and integrity of the data are high, the method is difficult to directly adapt to multi-source, heterogeneous and high-dimensional large-scale health data, secondly, the model is based on a Markov chain, the state transition is assumed to be related to the current state only, long-term dependence and multi-factor interaction effects in disease development are difficult to capture, and furthermore, the method does not fully integrate real-time monitoring data and external environment factors, and dynamic and personalized risk early warning and intervention suggestion cannot be realized. In addition, most of the existing disease prediction systems are still in the stage of static modeling and post analysis, and the lack of support and dynamic updating mechanisms for real-time data flows leads to prediction results which lag behind the actual development process of the disease, thus limiting the practical application value of the disease prediction systems in clinical real-time decision making and health management. Therefore, it is necessary to develop a disease risk prediction analysis system that can integrate multi-source big data, support dynamic calculation simulation, and have high interpretability and real-time early warning capability, so as to improve accuracy, timeliness and practicability of disease prediction. Disclosure of Invention The invention aims to provide a calculation simulation disease risk prediction analysis system based on big data so as to solve the problems in the background technology. In order to solve the technical problems, the invention provides the following technical scheme that the calculation simulation disease risk prediction analysis system based on big data comprises: The data fusion processing module is used for receiving and processing heterogeneous health data from a plurality of data sources; the patient dynamic modeling module is connected with the data fusion processing module and is used for constructing and initializing individual state representation of the patient based on the processed data; The disease dynamic simulation engine is connected with the patient dynamic modeling module, adopts a state transition model with a learnable rule and is used for carrying out multi-step forward deduction on the individual state representation of the patient and simulating the dynamic evolution process of the disease state; And the online learning and risk quantifying module is connected with the disease dynamic simulation engine and is used for carrying out incremental updating on model parameters of the disease dynamic simulation engine according to newly input real-time health data and calculating and outputting dynamic risk indexes based on deduction results. Further, the data fusion processing module specifically includes: The first data processing unit is used for extracting characteristics of the time sequence monitoring data; A second data processing unit for performing feature extraction on the genomics data; the third data processing unit is used for extracting the characteristics of the medical image data; the fourth data processing unit is used for extracting characteristics of the text medical record data; And the data fusion unit is connected with the first data processing unit, the second data processing unit, the third data processing unit and the fourth data processing unit and is used for aligning and fusi