AI Tool Predicts Barrett’s Esophagus Recurrence with High Accuracy

Why It Matters

Risk‑adjusted surveillance can catch recurrent disease earlier while sparing low‑risk patients from unnecessary procedures, potentially lowering mortality and healthcare costs.

Key Takeaways

• AI model >90% accurate predicting Barrett’s recurrence.
• Analyzes data from 2,500+ treated patients.
• Identifies length, weight, age, sessions, pathology as risks.
• Recurrence occurs in ~30% patients, average two years.
• Personalized surveillance could cut unnecessary endoscopies.

Pulse Analysis

Barrett's esophagus remains the sole known precursor to esophageal adenocarcinoma, a malignancy with dismal survival rates. Endoscopic eradication therapy (EET) has become the standard of care for high‑grade dysplasia and early cancer, yet post‑treatment surveillance still follows a one‑size‑fits‑all schedule. Clinicians lack reliable tools to differentiate patients who are likely to experience recurrence from those who can safely extend follow‑up intervals, leading to both missed early lesions and excess endoscopic procedures.

The newly published AI model leverages more than 2,500 patient records, integrating demographic, endoscopic, and histologic variables into a predictive algorithm. By achieving greater than 90% accuracy in both internal and external validation sets, the tool outperforms traditional risk assessment methods. It flags high‑risk individuals based on measurable factors such as segment length, body mass index, age, number of ablation sessions, and baseline dysplasia grade, while also estimating the probable time to recurrence, typically around two years after EET.

If incorporated into clinical workflows, this decision‑support system could transform Barrett's surveillance. High‑risk patients would receive intensified monitoring, enabling earlier therapeutic intervention before malignant transformation. Conversely, low‑risk patients could benefit from reduced endoscopy frequency, decreasing procedural burden, anxiety, and costs. Ongoing international collaborations aim to validate the model across diverse populations, a critical step toward regulatory approval and widespread adoption in gastroenterology practices worldwide.