task-ts
task-ts copied to clipboard
Work related to time series prediction and forecasting of Coronavirus
CoronaWhy Time Series Modeling Task Force
A worldwide effort by volunteers to fight Coronavirus (COVID-19)
About CoronaWhy Time Series Modeling Team
The Time Series Modeling Task Team is a subgroup of the CoronaWhy team, it is formed by an interdisciplinary group of volunteers from across the world focused on understand how the many different geography-related factors may affect the spread of the virus.
The team is currently being led by Isaac (@isaacmg).
We are currently focusing on two streams:
Disease Modeling
This sub-task focuses on modeling the diesase spread. Specifically, here we want to forecast the number of cases, critical care beds used, and deaths at different level of geographic granularity. We also want to explore what features the model considers important to predicting these in order to provide insights.
Critical Care Forecasting (blocked due to lack of ICU data)
Our second sub-task focuses on utilizing critical care data to forecast patient length of stay, risk of decompensation, and overall mortality at each time step in their medical record charting events.
If you are interested feel free to fork and reach out to us on the CoronaWhy Slack channel.
Our forecasts are currently powered by Flow Forecast: a PyTorch deep learning for time series forecasting