So I built AIRiskCalc. It's a collection of free health risk assessment tools with AI interpretation.
Currently live with five calculators:
ASCVD Risk (10-year cardiovascular risk based on ACC/AHA 2013 Pooled Cohort Equations) Type 2 Diabetes Risk (ADA assessment model) Cardiovascular Risk (multi-factor evaluation) CVD Risk (cerebrovascular focus) Miscarriage Risk (early pregnancy assessment) Implementation: Next.js 15, Cloudflare Pages. AI layer uses Vercel AI SDK with Claude.
One detail I paid attention to: the ASCVD formula involves logarithmic transformations and coefficient matrices from the original 2013 paper. I transcribed the equations into TypeScript and verified against sample data in the paper's appendix.
Each calculator displays the reference source in code comments. Results include AI-generated explanations of what the numbers mean. No personal data is collected.
The question I'm thinking about: where's the line between helpful AI interpretation and overstepping into medical advice? I want the tool to be useful without replacing professional judgment.
Feedback welcome on this balance.
MrCoffee7•1h ago