The math for pancreatic cancer screening has always been a tough pill to swallow. With an annual incidence of just 13 per 100,000 in the general population, traditional screening tests simply can’t function effectively at such a low prevalence. Fortunately, a new strategy is emerging: risk enrichment.
During the AJR Forum on Opportunistic Screening, Michael Rosenthal, MD, PhD, of Dana-Farber Cancer Institute outlined how we can utilize AI to flip the script on these stats. Instead of searching for a needle in a haystack, first, let AI shrink the haystack.
High-Risk Filter: The goal? Move the screening population from a risk of 13 per 100,000 more toward a risk profile of 0.5 to 1 per 100—a hundred-fold intensification. This mirrors the risk levels seen in known familial and genetic cohorts, where screening is already proven to save lives. Dr. Rosenthal describes a multi-layered “filter” approach:
- Top Layer (low cost/low risk)—using both EMR and opportunistic imaging analysis to filter out low-risk individuals at the average-risk pool level
- Deepening the Filter—as the pool narrows, clinicians can move toward more intensive and specific tools
- Targeted Surveillance—final groups identified then receive direct surveillance: blood testing, stool testing, active imaging
AI Advantage: AI thrives in this “top-of-the-funnel” environment. And by analyzing vast amounts of data from existing records and imaging, it can provide the insight needed to identify high-risk subgroups…minus the cost and invasiveness of primary screening.
Bottom Line: In harnessing AI to identify these high-intensity risk groups within gen pop, finally, we’ll make active pancreatic cancer screening a clinical reality.
Leave a Reply