Artificial Intelligence Improves Prediction of Pancreatic Cancer

FRIDAY, May 12, 2023 (HealthDay News) -- Artificial intelligence (AI) can improve the design of surveillance programs for pancreatic cancer, according to a study published online May 8 in Nature Medicine.

Davide Placido, from the University of Copenhagen in Denmark, and colleagues applied AI methods to clinical data from 6 million patients (24,000 cases of pancreatic cancer) in Denmark (Danish National Patient Registry [DNPR]) and 3 million patients (3,900 cases) in the U.S. Veterans Affairs (US-VA). Machine learning models were trained on the sequence of disease codes in clinical histories, and prediction of cancer occurrence within incremental time windows was tested.

The researchers found that the performance of the best DNPR model had an area under the receiver operating characteristic (AUROC) curve of 0.88 for cancer occurrence within 36 months and decreased when disease events within three months before cancer diagnosis were excluded from training (AUROC [3m] = 0.83); for 1,000 highest-risk patients older than 50 years, the estimated relative risk was 59. Lower performance was seen on cross-application of the Danish model to US-VA data (AUROC, 0.71); performance was improved with retraining (AUROC, 0.78; AUROC [3m], 0.76).

"AI-based screening is an opportunity to alter the trajectory of pancreatic cancer, an aggressive disease that is notoriously hard to diagnose early and treat promptly when the chances for success are highest," a coauthor said in a statement.

Several authors disclosed financial ties to the biopharmaceutical industry.

Abstract/Full Text