What if medicine stopped treating you like the average patient, and started treating you like you?

MIT's Prof. Regina Barzilay has spent a decade building the AI algorithms to make that real, with a vision of cancer care where disease is caught far earlier, treatment involves less guesswork, and therapies have less side effects. Her current research takes on some of the hardest cases: metastatic patients in lung, colon, and breast cancer who've run out of options, using AI to predict which therapy is most likely to work for each individual. We get into why today's approach so often gives whole groups of patients the same treatment when only some will benefit, and why getting these tools to actual patients is its own hard problem. She also opens up about her own cancer diagnosis: how it reshaped which problems she thought were worth solving, and how she experienced the limits of treatments built for the average patient.

00:00 Intro01:12 The future of personalized healthcare08:46 Current landscape of personalized treatments12:36 Regina's research: AI for metastatic cancer patients 16:30 Challenges in data and methodology26:42 Interpretability of the models27:40 Validation of the models31:33 Understanding foundation models in biology34:02 Challenges with biology foundation models38:53 The value of personalized treatments for all patients44:29 Regulatory challenges48:14 Regina's personal journey through breast cancer52:30 Overcoming Skepticism in AI Research55:38 The value of building a hospital network58:01 Lightning round01:01:01 Outro