Disease propagation simulation system based on large language model

Abstract

The invention discloses a disease transmission simulation architecture based on a large language model, and the architecture deeply integrates a mathematical model, agent modeling and the large language model, and achieves the efficient and accurate simulation of disease transmission. The architecture comprises a bottom-layer technical support module and an upper-layer disease transmission dynamics logic module, the bottom-layer module adopts a distributed computing framework, an asycio asynchronous programming library and an aiohttp network communication technology, and high-concurrency computing of a million-level agent is ensured; the upper layer module supports user-defined disease characteristic parameters and propagation modes, and defines agent state transition rules based on a classical mathematical model. A policy issuing-execution-adjustment closed loop module is designed, and a scientific basis is provided for a policy maker. According to the method, the accuracy and practicability of epidemic situation propagation simulation are remarkably improved, and powerful technical support is provided for precise epidemic situation prevention and control and scientific decision making.

Description

Technical Field

[0001] The present invention relates to the field of big data, and in particular to a disease propagation simulation system based on a large language model. Background Art

[0002] In the field of public health, disease transmission simulation technology is of great significance for predicting epidemic trends, evaluating the effectiveness of prevention and control measures, and formulating scientific and reasonable prevention and control strategies. Currently, disease transmission simulation technology is mainly based on mathematical models and agent-based models (ABM models).

[0003] By quantifying the dynamics of disease transmission, mathematical models can describe the state transitions of a population during disease transmission, thereby predicting the development of an epidemic. The classic SIR (Susceptible-Infectious-Recovered) model and the extended SEIR (Susceptible-Exposed-Infected-Recovered) model are commonly used mathematical models that simulate the spread of a disease through differential equations. However, mathematical models also have certain limitations. For example, they are highly sensitive to parameters, which can lead to significant deviations in prediction results, especially in cases of insufficient data or high uncertainty. They also have difficulty capturing complex nonlinear behaviors, such as super-spreading events or community transmission chains. Furthermore, they fail to fully consider the impact of sociocultural factors on disease transmission.

[0004] Agent-based models (ABMs) simulate individual behaviors and their interactions, enabling a more realistic reflection of disease spread within a population. ABMs treat each individual in a population as an independent agent, assigning them specific behavioral rules and interaction patterns, thereby capturing the dynamics of disease spread at a microscopic level. Furthermore, ABMs, combined with geographic information systems (GIS) and real-time data, enable spatial visualization of epidemics, supporting the development of regionalized prevention and control strategies. However, ABMs also have limitations. These include high computational complexity, particularly when simulating large populations and complex social networks, which significantly increases the demand for computing resources. They are highly data-dependent, requiring detailed individual behavioral and contact network data, which are often difficult to obtain or inaccurate in practice. Parameter calibration and validation are complex, potentially leading to biased model results. Agents often lack behavioral autonomy, with their behavioral rules often pre-set and unable to dynamically adapt to environmental changes or the complexity of individual decision-making. Furthermore, theoretical analysis and comprehensive assessment of macroeconomic factors are difficult, particularly when considering global factors such as policy changes or the economic environment.

[0005] With the rapid development of big data, artificial intelligence, and computing power, the application scope of mathematical models and ABM models has further expanded. The introduction of machine learning algorithms enables models to process massive amounts of data more efficiently, improving the accuracy of predictions; the application of high-performance computing technologies has enabled real-time simulation of complex models. However, while technological advances have opened up new possibilities for disease transmission simulation, the question of how to effectively integrate macroscopic mathematical frameworks with microscopic simulations of individual behavior remains a pressing issue.

[0006] In order to solve the above problems, the applicant proposes a disease transmission simulation system based on a large language model. Summary of the Invention

[0007] The purpose of the present invention is to provide a disease transmission simulation system based on a large language model to solve the problems in the prior art.

[0008] To achieve the above objectives, the present invention provides the following technical solutions: a disease transmission simulation system based on a large language model, comprising:

[0009] The underlying technical support module, based on the distributed computing framework, the asyncio asynchronous programming library, and the aiohttp network communication technology, enables high-concurrency simulation of millions of intelligent agents, ensuring the accuracy of model results.

[0010] The upper-level disease transmission dynamics logic module is used to simulate the dynamic process of disease transmission, provide disease feature definition and transmission model design, support users to customize disease feature parameters and transmission models, and define population state transition rules based on classic mathematical models (such as SIR model and SEIR model).

[0011] Optionally, an agent behavior modeling module is further included, and the agent behavior modeling module includes:

[0012] Multi-dimensional behavioral modules are loaded for each intelligent agent, including emotion generation, Internet communication, face-to-face communication, economic behavior, etc., to simulate complex interactions in the real society, including the impact of psychological changes, information dissemination and economic activities on disease transmission.

[0013] Optionally, a medical behavior and policy evaluation module is also included, and the medical behavior and policy evaluation module includes:

[0014] The agent's medical behavior decision-making module allows the agent to choose between self-healing, medication, or medical treatment based on their health status and financial conditions;

[0015] The policy evaluation closed-loop module is designed to form a closed loop of policy issuance, implementation, and adjustment, supporting the dynamic evaluation and optimization of prevention and control policies and providing scientific and actionable optimization suggestions for policymakers.

[0016] Optionally, a multi-scale scene modeling and visualization module is further included, and the multi-scale scene modeling and visualization module includes:

[0017] Based on the folium map library, free access to domestic and foreign map interfaces is achieved;

[0018] Combined with the pygame engine for scene classification modeling and refined visualization, including residential, office, school, hospital and other scenes;

[0019] The coordinate extraction function automatically obtains building information based on the center coordinates and radius range provided by the user, realizing scene modeling and visualization of different scales.

[0020] Beneficial effects: Improve the accuracy and practicality of simulation:

[0021] By introducing a large language model and combining the mathematical model of disease simulation with agent modeling (ABM model), the present invention successfully solves the problem of simplifying agent behavior in the prior art.

[0022] Loading multi-dimensional behavioral modules for the intelligent agent, including emotion generation, Internet communication, face-to-face communication, economic behavior, etc., enables the simulation to more realistically reflect the complex interactions in society, including the impact of psychological changes, information dissemination and economic activities on disease transmission.

[0023] Based on the distributed computing framework, the asyncio asynchronous programming library, and aiohttp network communication technology, high-concurrency task distribution and cloud computing support are achieved, ensuring the efficient simulation of millions of intelligent agents and significantly improving the accuracy and practicality of epidemic spread simulations.

[0024] Enhanced simulation flexibility and scalability:

[0025] The present invention supports user-defined disease characteristic parameters (such as basic reproduction number, incubation period, etc.) and transmission modes (such as contact transmission, droplet transmission, and airborne transmission), and defines population state transition rules based on classical mathematical models (such as SIR model and SEIR model).

[0026] Users can expand or customize new models as needed to meet simulation requirements in different scenarios. At the same time, the system has expansion potential and can be applied to social governance research in multiple fields such as natural disaster response and economic crisis management.

[0027] Providing scientific basis and policy optimization:

[0028] The closed-loop design of policy issuance, implementation and adjustment provides a scientific basis for the dynamic evaluation and optimization of prevention and control strategies.

[0029] Policymakers can make optimization suggestions based on simulation results, making policy making more precise and effective, which will help improve the effectiveness and efficiency of epidemic prevention and control.

[0030] Meeting multi-scale simulation needs:

[0031] The present invention realizes free access to domestic and foreign map interfaces based on the folium map library, combines the coordinate extraction function with the pygame engine, and supports classification modeling and refined visualization of scenes such as residences, offices, schools, and hospitals.

[0032] By combining multi-scale scenario modeling with real-time data, the present invention can meet the simulation needs at both macro and micro scales, providing strong technical support for precise epidemic prevention and control and scientific decision-making.

[0033] Improve computing efficiency and reduce time overhead:

[0034] By utilizing the distributed computing framework, the asyncio asynchronous programming library, and the aiohttp network communication technology, the present invention realizes high-concurrency task distribution and cloud computing support.

[0035] This ensures efficient simulation of millions of intelligent agents with low time overhead, improves computing efficiency, and makes large-scale and complex disease transmission simulation possible. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a schematic diagram of an embodiment of the present invention; DETAILED DESCRIPTION

[0037] The following describes preferred embodiments of the present invention with reference to the accompanying drawings to make its technical content clearer and easier to understand. The present invention can be embodied in many different forms, and the scope of protection of the present invention is not limited to the embodiments mentioned herein.

[0038] This invention provides a disease transmission simulation system based on a large language model. By combining a mathematical model for disease simulation with agent-based modeling (ABM), and introducing a large language model, this system achieves high-precision simulation of the disease transmission process. The following is a detailed description of the technical solution of the invention:

[0039] 1. System Architecture Overview

[0040] The system architecture of the present invention is divided into a bottom-level technical support module and an upper-level disease transmission dynamics logic module, which work together to achieve efficient and accurate disease transmission simulation.

[0041] Bottom-level technical support module:

[0042] Responsibilities: Provide high-concurrency computing support for upper-level modules, ensure efficient operation and reduce time overhead in large-scale simulation scenarios, and be responsible for data storage and front-end visualization.

[0043] Technical implementation:

[0044] High-concurrency computing: Based on the distributed computing framework, the asyncio asynchronous programming library, and aiohttp network communication technology, computing tasks are distributed to cloud server clusters, fully utilizing computing resources to achieve concurrent simulation of millions of intelligent agents.

[0045] Front-end visualization: The folium map library is introduced to access domestic and foreign map interfaces (such as Google Maps and Amap) to achieve spatial visualization of the epidemic; a coordinate extraction function is designed to automatically obtain building information (name, latitude and longitude coordinates, etc.); based on building function classification scenarios (residential, office, school, entertainment consumption, road traffic, hospital, etc.), the pygame engine is used for refined modeling, and the tiled map tool is used for map drawing, providing users with intuitive and interactive simulation results display.

[0046] Data storage: MongoDB database is used to achieve fast data access and efficient management, supporting the storage and fast query of large-scale simulation data.

[0047] Upper-level disease transmission dynamics logic module:

[0048] Core function: Simulating the dynamic process of disease transmission and providing highly customized disease feature definition and transmission model design are the core logic layer of this invention.

[0049] 2. Upper-level disease transmission dynamics logic module

[0050] This module is responsible for defining the core logic of disease propagation, including disease characteristics, propagation patterns, and state transitions of intelligent agents.

[0051] Disease characteristics definition and transmission model design:

[0052] Users can customize disease characteristic parameters, such as the basic reproduction number R0, incubation period, infection period, recovery period, etc., to simulate the transmission characteristics of different diseases.

[0053] It supports multiple transmission modes, such as contact transmission, droplet transmission, and airborne transmission. Users can freely combine and define them to simulate different transmission routes.

[0054] Crowd state transfer:

[0055] Based on classic mathematical models, such as the SIR model and SEIR model, the state change rules of the intelligent agent are defined to describe the individual's transition process between susceptible, infected, recovered and other states.

[0056] Support users to expand or customize new state transition models to meet simulation requirements in different scenarios and improve the flexibility and scalability of the system.

[0057] 3. Agent Behavior Modeling

[0058] In order to simulate the spread of the disease more realistically, the present invention loads a multi-dimensional behavior module into the intelligent agent to simulate the complex behavior of individuals in the epidemic.

[0059] Multi-dimensional behavior module:

[0060] Emotion generation module: simulates the psychological changes of intelligent agents during the epidemic, such as panic, anxiety, and optimism. These emotions will affect individual behavioral decisions. For example, panic may cause individuals to be more inclined to take protective measures or escape.

[0061] Internet Communication Module: This module simulates intelligent agents obtaining information and spreading public opinion through social media, reflecting the impact of the Internet on the spread of the epidemic. For example, the spread of public opinion may affect individuals' risk perception and behavioral choices.

[0062] Face-to-face communication module: simulates the contact behavior of intelligent agents in physical spaces, such as interactions in homes, schools, workplaces, etc., which is an important way for disease transmission.

[0063] Economic behavior module: simulates the economic activities of intelligent entities, such as consumption and work, to reflect the impact of the epidemic on the social economy. For example, a reduction in economic activities may lead to a decrease in individual income, which in turn affects their medical behavior.

[0064] 4. Medical behavior and policy evaluation

[0065] The present invention also simulates the medical behavior of intelligent agents and designs a policy evaluation closed-loop module to evaluate the effectiveness of prevention and control policies.

[0066] Medical behavior module:

[0067] Decision logic: The intelligent agent chooses to self-heal, take medicine or seek medical treatment based on its own health status and economic conditions. Economic conditions will affect the individual's accessibility and choice of medical services.

[0068] Refined behavior: The agent's choice of pharmacy or hospital is based on economic ability and symptom severity. For example, individuals with better economic ability may be more inclined to choose hospital treatment, while individuals with poorer economic ability may choose self-healing or medication.

[0069] Policy evaluation closed-loop module:

[0070] Policy issuance module: simulates the government issuing prevention and control policies, such as lockdown measures, social distance requirements, vaccination policies, etc.

[0071] Policy execution module: simulates the agent's response behavior to the policy, such as compliance or violation of the policy. The effect of policy execution will affect the speed and scope of disease spread.

[0072] Policy adjustment module: Evaluate policy effects based on simulation results, put forward optimization suggestions, and form a closed loop of policy issuance-implementation-adjustment, providing scientific basis for policy makers and making policy making more accurate and effective.

[0073] Example 1: City-level epidemic dynamic simulation and policy optimization system

[0074] Application scenario: A new respiratory infectious disease breaks out in a large city, and the impact of different prevention and control policies (such as lockdowns, vaccinations, and social distancing) on ​​the spread of the epidemic needs to be evaluated.

[0075] Implementation steps:

[0076] Data input and scenario modeling:

[0077] Connect to the AutoNavi Map API through the folium map library and import city GIS data (population distribution, transportation network, and medical institution coordinates).

[0078] Use the pygame engine to build a multi-scale urban scene model, including key nodes such as residential areas, commercial centers, and subway stations.

[0079] Agent initialization:

[0080] Generate millions of intelligent agents and assign attributes such as age, occupation, and economic ability;

[0081] Load multi-dimensional behavioral modules: emotional module (panic index affects medical decisions), Internet communication module (simulates social media communication), and economic module (unemployment rate affects consumption behavior).

[0082] Disease transmission parameter settings:

[0083] Custom disease parameters: R0 = 3.2, incubation period 5 days, airborne transmission accounts for 60%;

[0084] The state transition rules are defined based on the SEIR model, and the asymptomatic infected person sub-state is introduced.

[0085] Concurrent computing and simulation running:

[0086] The asyncio asynchronous framework is used to distribute computing tasks to cloud server clusters, simulating the cross-regional flow and contact infection of intelligent agents in real time.

[0087] Policy evaluation and dynamic optimization:

[0088] Issued the "regional lockdown + mandatory mask order" policy;

[0089] Monitor agent compliance (collect violation data in real time via aiohttp);

[0090] The simulation results show that the infection rate in the lockdown area dropped by 40%, but the economic activity decreased by 25%;

[0091] The policy adjustment module recommends "phased unblocking + economic subsidies", and a second simulation verifies the feasibility.

[0092] Technical effects:

[0093] Supporting millions of intelligent agents to complete 30-day epidemic simulation within 72 hours;

[0094] The emotion module increases the rate of seeking medical treatment by 15% and reduces the rate of severe illness;

[0095] After the policy closed-loop module is optimized, the plan balances health and economic losses.

[0096] Example 2: Hospital Cross Infection Prevention and Control Simulation System

[0097] Application scenario: A nosocomial infection broke out in a tertiary hospital, and the triage process and isolation strategy needed to be optimized.

[0098] Implementation steps:

[0099] Microscopic scene modeling:

[0100] Use the tiled tool to draw a hospital floor plan, dividing the outpatient area, inpatient department, ICU, etc.

[0101] Based on MongoDB to store medical staff schedules and patient medical records.

[0102] Refined agent modeling:

[0103] Define four types of agents: patients (mild / severe), doctors, nurses, and cleaning staff;

[0104] Loading behavioral modules: logic for doctors wearing protective equipment (economic module affects material allocation), patient movement path decision-making (emotional module drives anxiety aggregation).

[0105] Propagation mode settings:

[0106] Definition: contact transmission (0.8% infection probability through door handle contact) + droplet transmission (12% infection rate within 1 meter);

[0107] Expand the SEIR model and add the "hospital infection" marker state.

[0108] Real-time simulation and visualization:

[0109] Dynamically display patient flow hotspots (such as registration window clusters) through the pygame engine;

[0110] Achieve second-level response based on distributed computing and simulate different triage strategies (such as pre-examination triage buffer).

[0111] Policy effectiveness verification:

[0112] Issued the policy of "strengthening ICU disinfection frequency + patient zoning management";

[0113] The simulation shows that the cross-infection rate will drop from 8.7% to 2.1%, but the workload of medical staff will increase by 20%;

[0114] The adjustment module recommends "AI triage and diversion + automatic disinfection robot deployment", and the secondary simulation verification efficiency is increased by 35%.

[0115] Technical effects:

[0116] The micro-model accurately locates the registration window as a super-spreading node;

[0117] Dynamic policy adjustments reduced hospital infection control costs by 50%;

[0118] Multidimensional behavioral modeling revealed a positive correlation between medical staff fatigue and protection failure.

[0119] Innovative description of the embodiment:

[0120] Both implementations use the "large language model + ABM" to achieve autonomous decision-making by intelligent agents (such as automatically generating optimization recommendations for patient treatment paths), and combine multi-scale modeling to cover macro policies and micro scenarios, breaking through the limitations of traditional models in behavioral complexity, real-time performance, and cross-dimensional analysis.

[0121] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, from all points of view, the embodiments should be regarded as illustrative and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all changes that fall within the meaning and range of equivalents of the claims be included in the present invention. Any reference signs in the claims should not be construed as limiting the claim to which they relate.

[0122] In addition, it should be understood that although this specification is described in terms of implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

Claims

1. A disease transmission simulation system based on a large language model, characterized by: include: The underlying technical support module, based on the distributed computing framework, the asyncio asynchronous programming library, and the aiohttp network communication technology, enables high-concurrency simulation of millions of intelligent agents, ensuring the accuracy of model results. The upper-level disease transmission dynamics logic module is used to simulate the dynamic process of disease transmission, provide disease feature definition and transmission model design, support users to customize disease feature parameters and transmission models, and define population state transition rules based on classic mathematical models (such as SIR model and SEIR model).

2. The system according to claim 1, wherein: It also includes an agent behavior modeling module, the agent behavior modeling module including: Multi-dimensional behavioral modules are loaded for each intelligent agent, including emotion generation, Internet communication, face-to-face communication, economic behavior, etc., to simulate complex interactions in the real society, including the impact of psychological changes, information dissemination and economic activities on disease transmission.

3. The system according to claim 1, wherein: It also includes a medical behavior and policy evaluation module, which includes: The agent's medical behavior decision-making module allows the agent to choose between self-healing, medication, or medical treatment based on their health status and financial conditions; The policy evaluation closed-loop module is designed to form a closed loop of policy issuance, implementation, and adjustment, supporting the dynamic evaluation and optimization of prevention and control policies and providing scientific and actionable optimization suggestions for policymakers.

4. The system according to claim 1, wherein: It also includes a multi-scale scene modeling and visualization module, which includes: Based on the folium map library, free access to domestic and foreign map interfaces is achieved; Combined with the pygame engine for scene classification modeling and refined visualization, including residential, office, school, hospital and other scenes; The coordinate extraction function automatically obtains building information based on the center coordinates and radius range provided by the user, realizing scene modeling and visualization of different scales.

Images