AI in Oncology: Managing Information Overload for Trainees and Patients

This is the second part of a conversation with Drs Pallav Mehta, Yan Ji, and Kurt Demel. Listen to part 1 here.

In part 2 of this conversation, oncologists discuss how they help both patients and trainees navigate an overwhelming volume of medical information. The speakers describe encouraging patients to consult reliable resources like NCCN's patient site rather than turning solely to AI chatbots, while also acknowledging patients will inevitably seek outside information and clarification from family.

For fellows, the panel emphasizes using training time intentionally: reading about patients in depth before appointments, focusing on core algorithms during subspecialty rotations, and building relationships with mentors who can be lifelong resources. Dr Demel highlights curated tools like condensed literature summaries and topic-specific podcasts as more efficient than reading full journals, especially given the explosive growth in oncology trials and publications over the past two decades alongside rising retraction rates. Dr Mehta describes using voice-based AI during a daily commute to get updated on relevant papers, a habit some fellows have since adopted.

The discussion turns to varying attitudes toward AI among colleagues, with speakers describing a spectrum from enthusiastic adopters to skeptics wary of hallucinations. Dr Ji stresses recognizing "good enough" knowledge limits, using AI to stay current on standard-of-care changes and practice-changing trials, while emphasizing that human collaboration and communication remain essential when navigating complex cases beyond one's expertise.

The conversation closes with reflections on reading outside one's specialty to address patients' broader health needs, including medication interactions and comorbidities. Panelists describe current AI use as still in its "infancy," with promising applications emerging in patient monitoring, wearable data, and predictive analytics. Take-home messages include "cautious optimism" about AI's power and a prediction that oncologists who learn to use AI effectively will have a distinct advantage in the field's future.