A study on stochastic epidemic diffusion forecasting model with observation delay and reaction delay
- 주제(키워드) Epidemic diffusion , delay , Gillespie algorithm , chemical reaction model , simulation
- 주제(DDC) 658
- 발행기관 아주대학교
- 지도교수 장병윤
- 발행년도 2022
- 학위수여년월 2022. 2
- 학위명 박사
- 학과 및 전공 일반대학원 경영학과
- 실제URI http://www.dcollection.net/handler/ajou/000000031505
- 본문언어 영어
- 저작권 아주대학교 논문은 저작권에 의해 보호받습니다.
초록/요약
In 2019, COVID-19 emerged worldwide, and it still continues to spread. In order to prevent the spread of disease, there have been many efforts, such as developing medicine and vaccine or studying forecasting epidemic diffusion. Especially, forecasting studies help government officials take proper action in time. However, there were few forecasting studies considering the nature of the disease and the data problem in the real-world. Thus, the stochastic SIQR(susceptible-infected-quarantine-removed) model is proposed. Unlike the traditional model, the SIQR model considers asymptomatic or pre-symptomatic patients who are able to transmit the disease. This research also combines the delays that occurred in the observation to consider the data problem and in the reaction to reflect the gradual trend change. Finally, a simulation of the complex model using the Gillespie algorithm is established. This research shows that the proposed SIQR model explains COVID-19 epidemic diffusion better than the traditional SEIR(susceptible-exposed-infected-released) in terms of forecasting errors such as MAPE, RMSE and MAD.
more목차
1 Introduction 1
1.1 Backgrounds 1
1.2 Research objective 7
2 Compartment model : SIR model 9
2.1 Deterministic SIR model 9
2.2 Stochastic SIR model 13
2.3 Variants of the SIR model 15
3 The Gillespie algorithm 17
4 Parameter estimation 20
4.1 Maximum likelihood estimation 20
4.2 Bayesian inference 23
4.3 Markov Chain Monte Carlo 27
5 Research model 30
5.1 SIQR model 30
5.2 Observation delay 32
5.3 Reaction delay 34
6 Simulation 39
7 Conclusion and future research 46
References 48
8 Appendix 51
8.1 Model prior 51
8.2 Traceplot 52
8.3 Model comparison (2020.11.01 – 2021.07.01) 54
8.4 Simulation (2020.11.01 – 2021.07.01) 56
8.5 Python code 58
