AI Tool Predicts Post-Therapy Barrett’s Esophagus Recurrence

Barrett’s esophagus (BE) is the only known precursor to esophageal adenocarcinoma, an aggressive cancer with high mortality. After endoscopic eradication therapy, disease can recur, making long-term surveillance essential yet imprecise. Uniform follow-up schedules may expose low-risk patients to unnecessary procedures while missing early relapse in high-risk patients. To help address this challenge, U.S. researchers have developed an artificial intelligence tool to predict recurrence and estimate its timing.

Developed by investigators from the University of Colorado Anschutz Medical Campus (Aurora, Co), USA) and collaborators across the United States, the machine-learning model analyzes routine clinical variables from patients treated with endoscopic eradication therapy. The study reporting the tool’s development and validation appears in Clinical Gastroenterology and Hepatology, published April 7, 2026. The aim is to support risk-aligned surveillance after therapy for Barrett’s esophagus–related dysplasia and early esophageal adenocarcinoma.

Researchers assembled data from more than 2,500 patients who underwent endoscopic eradication therapy and were followed over time to determine whether Barrett’s tissue, dysplasia, or cancer returned and when. The model was trained to evaluate multiple factors simultaneously, including age, body weight, extent of Barrett’s segment, need for additional treatment sessions, and severity of histologic changes at diagnosis. It learned patterns across these variables that are not easily recognized by clinicians.

Nearly three in 10 patients experienced recurrence after successful treatment, and recurrence typically occurred about two years after therapy. In testing, the model achieved over 90% accuracy in identifying patients likely to recur and in estimating the timing of recurrence. Performance held when evaluated both on patients similar to those in the training cohort and on different patient groups drawn from other sources.

Such predictions could enable personalized follow-up that intensifies endoscopic surveillance for those at higher risk and reduces procedure burden for those at lower risk. A risk-aligned approach could also lessen patient anxiety and improve allocation of endoscopy resources. The tool is positioned as an adjunct to existing post-therapy care pathways.

Planned work will further validate the model using international datasets through collaborations in the Netherlands, the United Kingdom, Belgium, and Switzerland. The goal is to confirm generalizability so the model can be applied broadly. If validated, the approach could offer a reliable, universal aid to guide surveillance after endoscopic eradication therapy for Barrett’s esophagus.

“Early detection of Barrett's esophagus related dysplasia and associated esophageal adenocarcinoma can save lives. Identifying recurrence in the form of BE, BE-related dysplasia and BE-related esophageal adenocarcinoma earlier, especially in high-risk patients who have undergone endoscopic eradication therapy, creates opportunities for timely treatment before cancer develops or progresses,” said Sachin Wani, MD, executive director, Rady Esophageal and Gastric Center of Excellence, University of Colorado Anschutz Cancer Center.

Related Links
University of Colorado Anschutz Cancer Center