Science Friday Could a ‘digital twin’ help you get better health care?
46 snips
Mar 17, 2026 Caroline Chung, a radiation oncologist and co-director of UT MD Anderson’s Institute for Data Science Oncology, discusses building medical digital twins. She explains how these evolving models could personalize radiation and chemo plans, origins from engineering, data limits, privacy and ownership challenges, and how hybrids of physics and AI might shape clinical decision making.
AI Snips
Chapters
Transcript
Episode notes
Hybrid Models Balance Physics And AI
- Digital twins can be mechanistic, AI-driven, or hybrid; mechanistic models impose constraints to prevent unrealistic outputs.
- Chung highlights physics-informed models help keep predictions grounded compared with unconstrained AI alone.
Using Twins To Target Radiation Within Tumors
- Chung describes using digital twins in radiation oncology to map tumor subregions needing higher dose while sparing normal tissue.
- She explains adapting radiation early in treatment by anticipating aggressive or resistant tumor cells within each patient's tumor.
Trials Are Testing Twin-Guided Radiation Plans
- Clinical trials are already testing differential radiation dosing and MD Anderson is designing trials incorporating digital twins.
- Chung mentions colleague Heiko Enderling has run such trials with clinical partners.