A seq2seq model to forecast the COVID-19 cases, deaths and reproductive R numbers in US counties (preprint)

The global pandemic of coronavirus disease 2019 (COVID-19) has killed almost two million people worldwide and over 400 thousand in the United States (US). As the pandemic evolves, informed policy-making and strategic resource allocation relies on accurate forecasts. To predict the spread of the virus within US counties, we curated an array of county-level demographic and COVID-19-relevant health risk factors. In combination with the county-level case and death numbers curated by John Hopkins university, we developed a forecasting model using deep learning (DL). We implemented an autoencoder-based Seq2Seq model with gated recurrent units (GRUs) in the deep recurrent layers. We trained the model to predict future incident cases, deaths and the reproductive number, R. For most counties, it makes accurate predictions of new incident cases, deaths and R values, up to 30 days in the future. Our framework can also be used to predict other targets that are useful indices for policymaking, for example hospitalization or the occupancy of intensive care units. Our DL framework is publicly available on GitHub and can be adapted for other indices of the COVID-19 spread. We hope that our forecasts and model can help local governments in the continued fight against COVID-19.

A seq2seq model to forecast the COVID-19 cases, deaths and reproductive R numbers in US counties (preprint)

The global pandemic of coronavirus disease 2019 (COVID-19) has killed almost two million people worldwide and over 400 thousand in the United States (US). As the pandemic evolves, informed policy-making and strategic resource allocation relies on accurate forecasts. To predict the spread of the virus within US counties, we curated an array of county-level demographic and COVID-19-relevant health risk factors. In combination with the county-level case and death numbers curated by John Hopkins university, we developed a forecasting model using deep learning (DL). We implemented an autoencoder-based Seq2Seq model with gated recurrent units (GRUs) in the deep recurrent layers. We trained the model to predict future incident cases, deaths and the reproductive number, R. For most counties, it makes accurate predictions of new incident cases, deaths and R values, up to 30 days in the future. Our framework can also be used to predict other targets that are useful indices for policymaking, for example hospitalization or the occupancy of intensive care units. Our DL framework is publicly available on GitHub and can be adapted for other indices of the COVID-19 spread. We hope that our forecasts and model can help local governments in the continued fight against COVID-19.

An agent-based model of spread of a pandemic with validation using COVID-19 data from New York State (preprint)

We introduce a simple agent based model where each agent carries an effective viral load that captures the instantaneous state of infection of the agent and simulate the spread of a pandemic and subsequently validate it by using publicly available COVID-19 data. Our simulation tracks the temporal evolution of a virtual city or community of agents in terms of contracting infection, recovering asymptomatically, or getting hospitalized. The virtual community is divided into family groups with 2-6 individuals in each group. Agents interact with other agents in virtual public places like at grocery stores, on public transportation and in offices. We initially seed the virtual community with a very small number of infected individuals and then monitor the disease spread and hospitalization over a period of fifty days, which is a typical time-frame for the initial spread of a pandemic. An uninfected or asymptomatic agent is randomly selected from a random family group in each simulation step for visiting a random public space. An uninfected agent contracts infection if the public place is occupied by other infected agents. We have calibrated our simulation rounds according to the size of the population of the virtual community for simulating realistic exposure of agents to a contagion. Our simulation results are consistent with the publicly available hospitalization and ICU patient data from different communities of varying sizes in New York state. Our model can predict the trend in epidemic spread and hospitalization from a set of simple parameters and could be potentially useful in exploring strategies to keep a community safe.