COVIDcast: An ecosystem for COVID-19 tracking and forecasting
summary
An open COVIDcast Indicator Database ingests partner data from Facebook surveys, Change Healthcare claims, Quidel antigen tests, Google searches and JHU/USAFacts, produces ~70 real‑time indicators, and serves them daily via an HTTP API with R and Python clients. A historical database records revision tracking so backfilled claims change signals substantially; surveys collect about 50,000 responses per day (20M+ total). Interactive maps and forecasting/nowcasting tools exist; hospitalization signals from claims are unique and usable for modeling.
AI-written summary · about
chapters (9)
- 0:00Start
- 0:46The Delphi Group
- 6:04Data Floor
- 9:10Describe the Api
- 10:33R and Python Packages
- 11:10Python Github
- 12:03Examples
- 25:59Lessons Learned
- 31:24Taking Risks
related talks
22mCOVID-19: Data Science & Expertise
27mBeyond a Machine Learning Model
45mGoogle Health: Connecting Mobile Patients
28mThe COVID Scenario Pipeline: High Stakes Data Science
31mBayesian Risk Analysis For Large Multi-Modal Data
28mNLP Solutions for infodemic management
29mHow data science shortened the COVID-19 pandemic by 2 months
27mTech Talk