Lex Fridman Podcast Regina Barzilay: Deep Learning for Cancer Diagnosis and Treatment
22 snips
Sep 23, 2019 Regina Barzilay, a professor at MIT and a leading expert in deep learning for oncology, discusses her groundbreaking work in cancer diagnosis and treatment. She shares insights on how literature shapes scientific perspectives and the importance of empathy in research. The conversation touches on the challenges of data access for deep learning, innovations in drug discovery, and the evolution of natural language processing. Regina also reflects on personal missions and the search for meaning in life, emphasizing authenticity in the face of societal expectations.
AI Snips
Chapters
Books
Transcript
Episode notes
Data Donation
- Gathering large medical datasets is challenging due to privacy regulations and institutional incentives.
- Implement a system for patients to easily donate data for research, similar to organ donation.
Modernizing Data Exchange
- Current medical data exchange is outdated, relying on physical media like CDs.
- A centralized, cloud-based system where patients own and control their data would revolutionize healthcare data sharing.
Improving Risk Assessment
- Current breast density assessments are imprecise, classifying a large percentage of women as high-risk without clear guidance.
- Deep learning models can offer significantly more accurate breast cancer risk prediction by identifying subtle patterns.
