[论文解读] SEIR and Regression Model based COVID-19 outbreak predictions in India

Gaurav Pandey, Poonam Chaudhary+2 more|arXiv (Cornell University)|Apr 1, 2020

COVID-19 epidemiological studies参考文献 11被引用 118

一句话总结

论文比较 SEIR 和回归模型以预测印度的短期 COVID-19 病例(2020 年 1 月 30 日–2020 年 3 月 30 日)，并预测未来两周将有 5,000–6,000 例，使用 RMSLE 进行评估并估算 R0。

ABSTRACT

COVID-19 pandemic has become a major threat to the country. Till date, well tested medication or antidote is not available to cure this disease. According to WHO reports, COVID-19 is a severe acute respiratory syndrome which is transmitted through respiratory droplets and contact routes. Analysis of this disease requires major attention by the Government to take necessary steps in reducing the effect of this global pandemic. In this study, outbreak of this disease has been analysed for India till 30th March 2020 and predictions have been made for the number of cases for the next 2 weeks. SEIR model and Regression model have been used for predictions based on the data collected from John Hopkins University repository in the time period of 30th January 2020 to 30th March 2020. The performance of the models was evaluated using RMSLE and achieved 1.52 for SEIR model and 1.75 for the regression model. The RMSLE error rate between SEIR model and Regression model was found to be 2.01. Also, the value of R0 which is the spread of the disease was calculated to be 2.02. Expected cases may rise between 5000-6000 in the next two weeks of time. This study will help the Government and doctors in preparing their plans for the next two weeks. Based on the predictions for short-term interval, these models can be tuned for forecasting in long-term intervals.

研究动机与目标

• 分析截至 2020 年 3 月 30 日的印度 COVID-19 疫情。
• 使用 SEIR 和回归模型为未来两周制定短期预测。
• 评估并比较两种模型的预测性能。
• 估计研究期间印度的基本再生数 R0。

提出的方法

• 使用 SEIR 和回归模型进行的预测，数据校准自 2020-01-30 至 2020-03-30。
• 数据来自 Johns Hopkins University 的数据仓库。
• 使用 RMSLE 评估模型性能（SEIR = 1.52；回归 = 1.75）。
• 比较模型，SEIR 与回归预测之间的 RMSLE 差异为 2.01。
• 估计疾病传播的基本再生数 R0（R0 = 2.02）。
• 提供短期预测，预计未来两周将有 5,000–6,000 例。

实验结果

研究问题

• RQ1 在研究窗口内，SEIR 和回归模型是否能准确预测印度的短期 COVID-19 病例数？
• RQ2 SEIR 与回归模型在印度 COVID-19 数据的预测性能（RMSLE）方面如何比较？
• RQ3 研究期间印度的估计基本再生数（R0）是多少？
• RQ4 基于这些模型，未来两周的预测病例区间是多少？

主要发现

• SEIR 模型的 RMSLE 为 1.52。
• 回归模型的 RMSLE 为 1.75。
• SEIR 与回归模型之间的 RMSLE 差异为 2.01。
• 疫情的估计 R0 为 2.02。
• 短期预测在未来两周内预计新增 5,000–6,000 例。