How to Actually Use AI for Risk Stratification in Healthcare

theprescription.substack.com · 3,398 words · saved by 1 readers

AI is taking over healthcare. Suddenly, everyone is trying to build AI, implement AI tools, evaluate them, or become an AI influencer on LinkedIn. Despite all this hype, most people don't actually understand how AI works. They either view it as a magic solution—dump data into ChatGPT and solve everything—or dismiss it entirely as overhyped nonsense. The reality is more nuanced. AI is genuinely transformative, but only when you understand which problems it solves well versus where it struggles, how to evaluate AI products effectively, and what implementation actually requires. I think not enough engineers are involved in the discussion around how AI is built, implemented, and regulated in healthcare. After 15 years of building ML systems—from founding Avon Health (the first AI-native EMR to receive ONC certification) to deploying models at Microsoft, PathAI, Carnegie Mellon, and Stanford—I've seen the gap between what works in practice versus what sounds good in whitepapers. I hold a pa

How to Actually Use AI for Risk Stratification in Healthcare A non-engineering guide written by an engineer Maitreyee Joshi Jun 10, 2025 14 2 Share AI is taking over healthcare. Suddenly, everyone is trying to build AI, implement AI tools, evaluate them, or become an AI influencer on LinkedIn. Despite all this hype, most people don't actually understand how AI works. They either view it as a magic solution—dump data into ChatGPT and solve everything—or dismiss it entirely as overhyped nonsense. The reality is more nuanced. AI is genuinely transformative, but only when you understand which prob

How to Actually Use AI for Risk Stratification in Healthcare

related reading