Internal Tool for Biometric–Flare Pattern Validation (2025) A dashboard designed to visualize health trends and build a flare-up “fingerprint”. The dashboard allows users to find correlations between biometrics signals(HRV, HR, sleep) and chronic pain flare-ups.
This dashboard was built to visualize and use data to find patterns in patient responses to external stimuli.
The goal of the dashboard was to visualize and time-aligned trends across multiple data types: biometrics, weather, and self-reported data. This would support hypothesis testing, model refinement, and insight validation before exposing them in the consumer app.
Designed and built the tool end-to-end using Python + Streamlit Engineered ingestion of biometric, weather, and self-reported data Created a clean UX for cross-metric comparison over time Integrated filters to align around flare-up dates or biometric shifts
Parse Apple HealthKit data and external weather sources using time-normalized sequences Structured flare-up reports as labeled windows (ex., 2 days before/after a pain spike) Developed multi-chart views to explore 1. HRV suppression before flare-ups 2. Sleep duration dips before high pain scores 3. Environmental triggers like sudden humidity or pressure drops Added export functions to support further statistical testing or ML model tuning
Discovered repeatable correlations between barometrci pressure and increase in next-day pain. Pre-trigger warning cards have validated correlations for trigger warnings Built trust with users and improved accuracy of model
Time labeling(temporal alignment) is a critical factor in building a multi-modal system The dashboard allowed for quicker improvements then end-user app changes Data visualization helped surface deeper insights to improve the model