Modeling Epidemic Networks Based On Autonomy-Oriented Computing

Epidemics generate great threats to the lives and properties in society, and the study of theepidemic spread would help government make decisions to inhibit the rapid spread of thedisease. Therefore, to use the historical epidemiological surveillance data to infer thecharacteristics of epidemic spread and the route of transmission network structure has avery important and practical significance. This paper is a study of epidemic spread inepidemic transmission dynamic model，network model，network inferring and dataadaptability. The specific research work is as follows:1) Analyze the research methods about epidemic disease spread in medical field aswell as network modeling of information spread. Based on the comparisonbetween these two with network modeling of epidemic disease spread, reasonablenetwork modeling of epidemic disease spread is proposed. This paper introducesthe SEIR model to describe the epidemiological dynamic behavior; meanwhile,applies the traditional epidemiological biological parameters to characterizedisease outbreak characteristics; groups people according to biological attributes,social attributes, location, and other attributes. Each group is modeled as a node inthe network, and the probability of interaction between each group is modeled aslink weight. By this way, the epidemic network structure is established.2) Due to uncertainties such as data missing and non-uniform data granularity existfor the epidemical surveillance data, the information transmission networklearning methods are difficult to solve the epidemic spread network inferenceproblem. In view of this, D-AOC(diffusion-AOC) system frame is built on thebasis of AOC system. And Particle Swarm Optimization algorithm whichcombines with Monte Carlo simulation method is applied, so that the epidemicbiological parameters and the network structure can be inferred.Design a rigorousexperimental procedure to discuss models in their reasonability, parametricinference accuracy, epidemic trend forecast, and epidemic outbreak risk estimates.3) Further expand the epidemic network model by introducing governmentdecision-making control and public feedback mechanism. Aiming at public degreeof disease info, the model analyzes the effect of epidemic disease spread and proposes a dynamic network model of epidemic disease spread with public sharingstrategy, which is GPFEM (Government decision-making control and People’sFeedback Epidemic Model), with which ideal simulation results can be achieved.Experimental results show that the D-AOC-based method has stronger data robustness. Itcan better simulate in the former pandemic process, infer in epidemic networks structureand biological parameters more accurately, and estimate the disease outburst trend morepractically. And according to the missing data, system parameters can be more accuratelyestimated, and missing data can be filled up. Practical application of D-AOC system in theexperiment can estimate the regional epidemic infection risk within future three months,and can show it on the map vividly and dynamically. Based on D-AOC, GPFEM extendsepidemic network model. Epidemic networks are modeling based on factors such aseconomy, population, distance etc. and according to the government’s implementation ofcontrol measures, GPFEM system can better simulate the outbreak trend of the wholeprocess with real data.