PulseAI

PulseAI is an AI agent pipeline that turns raw health data into a plain-English brief. A routing agent first detects whether the uploaded CSV represents an individual patient or a population, then runs the appropriate analysis: for an individual, it checks each of 20 biomarkers (including Apple Watch metrics) against sex-stratified clinical reference ranges and flags anything abnormal; for a group, a hypothesis-generating agent profiles the population and proposes statistically testable hypotheses, which are then run through t-tests, ANOVA, Pearson correlation, and chi-squared tests, alongside K-Means clustering to segment patients into risk tiers and surface demographic health-equity disparities. A live demo is deployed on Streamlit. Built solo at Hook ‘Em Hacks 2026.

Stack: Google Gemini 2.5 Flash, Streamlit, SciPy, scikit-learn, Plotly, pandas, NumPy