Wolfpack Vitals

Inspiration

People with type-2 diabetes are 1.7 times more likely to suffer cardiovascular death. Accurate diagnosis of diabetes is a necessity to prevent avoidable deaths. There are few applications on the market that automatically collect and use AI/ML to analyze user data through non-intrusive means to detect prediabetes.

What it does

Wolfpack Vitals would read streamed heart rate, interbeat interval, glucose concentration, and electrodermal activity data from a wearable device to make a confident diagnosis about a users tendency to prediabetes. A Databricks agent has been trained on the labeled data of 15 patients. Wolfpack Vitals calls Databricks to provide context to Gemini alongside user data for Gemini to make an accurate prediction.

How we built it

The Four Aces split into four separate roles. Aaron handled our API calls, Kenan designed our UI, Christian set up the Databricks and local schemas, and Rudra was flex and graphic design.

Challenges we ran into

1) PhysioNet has terrible download speed, wasting four hours of time in the hackathon. 2) Learning Databricks took valuable time.

Accomplishments that we're proud of

1) Our data-gathering technique. 2) The UX of the application. 3) The wireframe logo for Wolfpack Vitals. 4) Our use of Databricks.

What we learned

1) Think outside the box. 2) Better utilization of APIs. 3) Collaboration methods across concentrations. 4) How little sleep the human body really needs.

What's next for Wolfpack Vitals

Depending on feedback, we may implement: actual connectivity with wearable devices, cross-platform support, and lifestyle recommendations.

Built With

• android-studio
• codex
• cursor
• databricks
• fastapi
• gemini
• html5
• kotlin
• python
• sqlite