AI-Enabled Opportunistic Screening

Gaurav Singal MD moderated a panel with Bea Bakshi MD, Alexandra Goncalves, Faisal Mahmood PhD and Nandini Nayer on an increasingly important use of healthcare AI: finding disease signals in data that already exists before patients enter a formal screening pathway.

Bakshi noted that many cancers are discovered because a patient presents with a problem, not because routine screening found the disease first. Records, symptoms and prior encounters may contain signals that are easy to miss in a short primary care visit. Goncalves described a similar lesson from an AI program designed to identify hypertrophic cardiomyopathy. Some apparent false positives turned out to have other significant cardiac conditions, showing that a model can surface clinically important risk even when it misses the intended label.

Mahmood and Nayer pushed the discussion toward what happens next. As models get better at identifying high-risk patients, health systems need the capacity to act on those findings without turning every alert into another specialist referral. The next phase of opportunistic screening will depend as much on redesigning the downstream clinical pathway as on improving the algorithm itself.