Mayo Clinic researchers have published results from a landmark validation study demonstrating that their AI model, the Radiomics-based Early Detection Model (REDMOD), can detect pancreatic cancer on routine abdominal CT scans up to three years before a clinical diagnosis would typically occur. The study, published in the journal Gut and led by Dr Ajit Harishkumar Goenka, analysed 219 CT scans from patients who later developed pancreatic ductal adenocarcinoma alongside 1,243 control scans from cancer-free individuals.
The results are striking: REDMOD identified 73 per cent of very early pancreatic cancers at a median of approximately 16 months before diagnosis, compared to just 39 per cent detected by specialist radiologists reviewing the same scans. In scans obtained more than two years before diagnosis, the AI identified nearly three times as many early cancers that would otherwise have gone undetected. The model also correctly identified 88 per cent of people without cancer, and risk scores showed 90 to 92 per cent consistency across repeat scans — critical for clinical reliability.
What makes REDMOD fundamentally different from traditional diagnostic tools is its approach: rather than looking for visible tumours, it measures hundreds of quantitative imaging features that describe tissue texture and structure, capturing faint biological changes as cancer begins to develop. The model is designed to analyse CT scans already obtained for other clinical reasons — particularly in high-risk patients such as those with new-onset diabetes — and flag elevated risk before any visible mass appears.
For context engineers, this represents one of the most compelling demonstrations of AI's real-world impact beyond software development. The technical architecture mirrors patterns familiar to developers: feature extraction from unstructured data, ensemble classification, and rigorous validation against human baselines. Mayo Clinic is now conducting AI-PACED, a prospective clinical study evaluating how clinicians can integrate AI-guided detection into standard care for patients at elevated risk. Pancreatic cancer has a five-year survival rate below 13 per cent precisely because it is almost always caught too late — if REDMOD proves out in prospective trials, it could fundamentally change outcomes for one of the deadliest cancers.
