Deciphering AI in Oncology: A Conversation with Dr. Helena Vance
Dr. Vance argues that the sheer volume of genomic data generated today has surpassed the cognitive bandwidth of human practitioners. "We are no longer looking for needles in haystacks," she notes. "We are looking for specific molecules in a haystack that is itself made of other needles." The discussion traverses the mechanics of Large Language Models in parsing medical literature and how specialized vision transformers are revolutionizing histopathology.
The future of oncology isn't about replacing the radiologist — it's about giving the radiologist a co-pilot trained on every paper ever published.
Trust remains the final frontier. Soma and Vance tackle the "black box" problem, discussing Explainable AI frameworks that allow clinicians to see the why behind a risk score. They explore case studies where AI-augmented workflows reduced diagnostic error rates by 14% in early-stage lung cancer detection.
View Full Transcript
[00:00] Sri: Welcome back to Newsense. Today I'm joined by Dr. Helena Vance from Alpha Health…
[02:14] Helena: Thanks Sri. Yes, the conversation around oncology AI has really shifted in the last 18 months — partly because the models are better, but mostly because the validation data is better…
Deep Dive Resources
The companion materials for Episode 42.