From single cells to hospital wards: how is AI actually changing medical research and care today? I’m joined by Marianna Rapsomaniki (Assistant Professor, Biomedical Data Science Center, CHUV–UNIL; leads the AI for Biomedicine group focused on tumor heterogeneity) and Jean Louis Raisaro (Assistant Professor of Biomedical Informatics & Data Science, CHUV–UNIL; Head of the Clinical Data Science Group working on trustworthy, privacy‑preserving clinical AI). What we cover - Generative AI in everyday clinical work: what’s genuinely useful now (notes, documentation, admin) vs. decision support. - Risks & ethics: hallucinations, automation bias, and how to protect critical thinking (including in literature reviews). - Interpretability vs. reliability: when explanations matter and when robust performance is the priority. - Marianna’s research: single‑cell & spatial omics, organoids, and multimodal models; using generative methods to simulate drug effects and predict immunotherapy response from tissue topology. - Data quality at scale: batch effects, foundation‑model fine‑tuning vs. training from scratch, and compute constraints in public research. - Jean Louis’s work: real‑time sepsis risk (neonates & adults), turning routine hospital data into actionable alerts and quality‑of‑care metrics. - LLMs in the clinic: clinician‑led evaluation and local alignment (RAG, tool use, agentic patterns), and surfacing uncertainty. - What it takes to reach the bedside: governance, validation across sites, interoperability, and evolving regulation. - The road ahead: why better data and human–AI collaboration will decide outcomes—both for research and for patients. Here is how to get more content like this: Substack of this podcast: https://theaiimpactpodcast.substack.c... My Linkedin: / quentin-gallea-phd My instagram: / the_ai_impact_podcast My TikTok: / the_ai_impact_podcast