Teaching AI to Predict the Future of Your Brain: Dr Hannah Madan from Prima Mente #423
The Healthtech Podcast - Podcast készítő Dr. James Somauroo
Kategóriák:
Hannah Madan joins us to share a rollercoaster journey of co-founding Prima Mente— an AI-driven neuroscience company building foundations models for the brain. This episode is packed with thought-provoking insights, including:— The biological hypothesis behind Prima Mente that "No one believed was true"— How Hannah made it by doing "just enough", not by being a top student— A baptism by fire— landing her first role as Chief of Staff and flying out for a YC launch party on her 25th birthday— Prima Mente's "Ecosystem plays" and atypical competitive strategies— Hannah's unique focus on the people she works with— deep personal understanding and fostering reflective behaviour — A decentralised vision for the future of healthcareConnect with Hannah: https://www.linkedin.com/in/hannahmadan/Learn more: https://www.primamente.com/Apply to be a guest: www.thehealthtechpodcast.comSubscribe to Healthtech Pigeon 🐦: www.healthtechpigeon.comGet in touch with James: www.jamessomauroo.comChapters:00:00:00 Welcome to the Health Tech Podcast00:01:23 Uncovering Novel Biology in the Neurodegenerative Space00:02:03 Primemente's Integrated Approach: Clinical Studies, Wet Lab, and Machine Learning00:03:22 Early Life 00:04:38 Undergraduate Studies00:08:58 PhD Research: The Link Between Diet, Diabetes, and Bowel Cancer00:12:23 Chief of Staff at Cambridge Cancer Genomics (CCG) and YC00:15:10 CCG's Mission: Liquid Biopsies for Cancer Treatment Monitoring00:19:05 The Road to Prima Mente00:22:19 Establishing Team Culture: Communication, Self-Reflection, and Growth00:27:13 Scientific Hypothesis: Cell-Free DNA and Epigenetics in Neurodegenerative Conditions00:30:03 Technical Approach: Using Transformer Architecture and Foundation Models00:33:47 The Pleiades Model and Diagnostic Accuracy00:34:56 The Sandbox Study: Pushing for Earlier Diagnosis00:37:47 Future Strategy: Focusing on Novel Clinically Actionable Targets
