COVID-19: Data Science & Expertise
summary
Covers two concrete COVID projects: a county-level outbreak predictor built from ~3,000 counties using census, mobility and later airport/hospital features and updated daily, whose top 50 predictions include 48 of the top 50 counties confirmed within ten days; and a chest x‑ray classifier that reports one mistake out of 150 but fails on inspection because data leakage (embedded clusters show labels and markers, not pathology). Notes AutoML platforms like DataRobot, Microsoft and AWS lower barriers but require domain expertise.
AI-written summary · about
chapters (15)
- 0:00Intro
- 2:23Machine learning models
- 4:25Reality is more complex than one variable
- 5:28Lowering the barrier to Al - AutoML
- 6:36Expertise in Real Estate
- 8:25Data Science and COVID-19
- 8:49COVID-19 in the United States
- 9:32Daily Predictions to identify Counties to Have Covid-19
- 9:52Daily Predictions for Covid-19
- 11:09Evaluating the Model
- 12:10Explaining the Model
- 15:05Next Steps -- Second Generation
- 15:57What data was used?
- 16:34COVID-19 Chest X-rays
- 18:51COVID Embeddings
related talks
29mHow data science shortened the COVID-19 pandemic by 2 months
27mBeyond a Machine Learning Model
47mKeynote: Using Data Science for Social Good: Examples, Opportunities, and Challenges
33mData Science in Healthcare: Beyond the Hype
33mCOVIDcast: An ecosystem for COVID-19 tracking and forecasting
37mAI/ML Strategies & Techniques for a Post Pandemic World
57mHow Machine Learning and AI Can Support the Fight Against COVID-19
28mThe COVID Scenario Pipeline: High Stakes Data Science