Cardiovascular Data Pipeline, API, and Analysis with Neural Network

This project has taken over my brain lately. So fun! The point of this project is to get a understand of my day-to-day heart health in the presence of exercise by only using devices and data that we passively produce. Here is the workflow in bullet points:

• Wrote integrations with Gmail, Apple Health, Strava, and Withings in order to scrape my own personal health data as well as environmental data like the weather.

• Collected all the health data into a database set up on a raspberry pi. The collection workflows run on an automated schedule.

• Cleaned up the data for analysis.

• Propped up a Convolutional Neural Network that approximates the solution to an ordinary differential equation whose variables predict metrics of heart health - responsiveness to new demand, responsiveness towards baseline after demand is met, max HR, resting HR, VO2 max, and impact of any regressors such as temperature, humidity, sleep quality, recovery index, etc.

This project touches a little bit of all the skills I have picked up the past few years: AI/ML, data pipelines, API (building and access), automation, etc. I am definitely expanding this more and hopefully generalizing the app so others can join!