A Data Assimilation Study Towards The Dynamics Of Infectious Disease

In recent years,a full-blown outbreak of Coronavirus Disease 2019(COVID-19)has had a major negative impact on economies,livelihoods and travel around the world.Until an absolute solution is found,non-pharmaceutical interventions such as maintaining social distance,using disinfectants,and wearing masks and gloves can help control the outbreak.Accurately predicting the trend of the epidemic under different prevention measures can help to improve the efficiency of prevention and minimize the consumption of prevention costs.This thesis optimizes the traditional Susceptible-Infected-Removed(SIR)structure by combining realistic factors such as population mobility and vaccination,and an improved Susceptible-Vaccinated-Infected-Recovered-Deceased(SVIRD)model is constructed.The model takes into account the effects of population movements between origin and destination,vaccination and re-positive populations.The model trajectory is modified using real-time observations through data assimilation methods.The details and results of the study are as follows:(1)The SVIRD model,a rational model of infectious disease,is constructed.The SVIRD model is constructed by adding vaccinators to the SIR model and splitting the removers into recovered and sickened individuals.In addition to this,a feedback mechanism is added to describe the possibility of recovered individuals becoming susceptible again.The SVIRD model is therefore more consistent than the SIR model with the dynamics of the transmission process of a new corona virus outbreak.(2)The Ensemble Kalman filter(EnKF)theory is introduced and the SVIRD-EnKF research framework is proposed.To address the problem that the warehouse model tends to ignore uncertainties such as environment and social distance,the EnKF algorithm is used,and then combined with the actual daily epidemic data to continuously update the prediction error covariance and analysis error covariance,so as to correct the prediction results of the infectious disease model.(3)The parameters and variables associated with the simulated experimental scenarios are set to simulate the outbreak trends under the seven epidemic prevention measures.The final results show that reducing the number of contacts per person per day,limiting the size of the population entering the destination per day,opening travel to areas with smaller outbreaks and increasing vaccine protection rates are the most effective in achieving short-term outbreak control,with a maximum reduction of84.79%,27.07%,38.77% and 16.12% in the mean number of existing confirmed cases over 150 days.(4)The fifth round of the Hong Kong epidemic is used as the study context to analyse the possible impact of the Hong Kong epidemic on Shenzhen if population movement resumes under different epidemic prevention measures.The results show that there is only a difference of 10,000 existing confirmed cases between the scenarios of Hong Kong(origin)and Shenzhen(destination)with border control lifted,when the number of contacts per person per day in Shenzhen is set at 3 and the vaccination rate is set at 5% and 9.5% respectively.In addition,controlling the number of contacts of infected persons is more effective in reducing the number of infections than increasing the vaccination rate after the resumption of productive life.In summary,the SVIRD-EnKF research framework constructed in this thesis is used to simulate the epidemic trends in destinations under different levels and scenarios of epidemic control measures,and to analyse the epidemic control measures that can effectively control the epidemic in the short term,which can provide some theoretical basis for the formulation of relevant policies.