Tracking activity patterns in those recently infected by COVID-19

About the Project

The aim of this project is to discover methods of pre-symptomatic COVID-19 detection using data collected from wearables. Using a recent study, “Pre-symptomatic detection of COVID-19 from smartwatch data” (Mishra et al), as the baseline for research and the source of data for analysis, the Evidation team will use the described outlier detection algorithms to determine the likelihood of COVID-19 infection in an individual given their heart rate, step count, and sleep data.

Short term goals for the project include:

• Recreating visualizations from Mishra et al
• Researching Mishra et al’s HROS metric and exploring alternative metrics
• Using anomaly detection algorithms described in Mishra et al and perhaps tweaking and improving upon these established algorithms

Ultimately. the project is not only looking to replicate the Mishra et al study but also to improve upon it and try alternative methods (like changing time frames used in computing metrics, exploring different anomaly detection algorithms, etc.).

A long term goal of this project is to explore regularized cross-validated linear mixed models and causal modeling and inference.