Project Synopsis

What is the problem being addressed?

This project acts as a proof of concept for medical providers interested in the capabilities available in everyday wearables.

Partnering with LT (Lean Techniques) and Boon Logic, our task was to develop a solution that could capture ECG data and use Boon Logic's anomaly detection to identify anomalous ECG data of the end-user wearing the device. 

What is the solution? 

Design and implement a backend (Java) API and a WatchOS (Swift) application that collects ECG data from an end user's Apple Watch, stream/store this data within a MySQL database (MariaDB on a server), and stream it to Boon Logic’s Machine Learning API (Boon Amber) where a sensor computes the ECG data and returns any detected anomalies. 

After having collected the detected anomalies from the Boon Amber API, we then stream the anomalous data to a custom web application/UI where the data can be viewed.

 

Software/Tools being used:

Swift/Apple Watch Application: 

  • WatchOS (Swift) Application
    • WatchKit Framework
    • HealthKit API
    • HTTP/POST
      • Used to send ECG data from WatchOS application to our server-side database.

Frontend Web Application:

  • Web Portal constructed using AngularJS and TypeScript
    • Functions communicate with the Backend API via HTTP Requests 
    • Fontawesome icons
    • Created an anomaly table on user homepage to display anomaly data detected by Boon Logic

Backend API:

  • Spring Boot (Java) API Application
    • Maven Dependencies
      • OKHTTP3
        • For constructing our REST API to send our collected data via JSON requests to Boon Logic.
      • Jackson (Object Mapping)
        • Serializing JSON in Boon Logic API requests and responses
    • JDBC (MySQL Connector)
    • MariaDB
      • For running the server-side database

Boon Amber REST API: docs.boonlogic.com/docs/Amber_REST.html