AI news ·
AI tool predicts Barrett's esophagus recurrence after treatment with over 90% accuracy
An AI model predicts Barrett's esophagus recurrence after treatment with over 90% accuracy, trained on data from 2,500+ patients. Current protocols treat all patients the same; this tool flags high-risk cases for closer follow-up.
AI Tool Predicts Barrett's Esophagus Recurrence After Treatment
Researchers have developed an artificial intelligence model that predicts with over 90% accuracy whether patients will experience a recurrence of Barrett's esophagus after endoscopic eradication therapy. The findings, published today in Clinical Gastroenterology and Hepatology, could reshape how doctors monitor patients following treatment for this precancerous condition.
Barrett's esophagus is the only known precursor to esophageal adenocarcinoma, an aggressive cancer with high mortality rates. Endoscopic eradication therapy effectively removes abnormal tissue and reduces cancer risk, but recurrence remains common-nearly three in 10 treated patients experience it.
The Problem With Current Surveillance
Current follow-up protocols treat all patients the same regardless of individual risk. Sachin Wani, executive director of the University of Colorado Anschutz Cancer Center's Rady Esophageal and Gastric Center of Excellence, said this approach misses opportunities for early intervention in high-risk patients.
"Current surveillance strategies don't distinguish between patients at high versus low risk," Wani said. "Everyone is followed using the same schedule regardless of their risk."
How the Model Works
Researchers trained the machine-learning tool on clinical data from more than 2,500 patients who underwent eradication therapy and were tracked over time to see if Barrett's tissue returned.
The model analyzes multiple patient factors simultaneously-age, body weight, disease severity, and treatment details-to identify patterns humans cannot easily spot. The analysis revealed five key recurrence risk factors:
- Longer area of Barrett's tissue
- Higher body weight
- Older age
- Multiple treatment sessions needed to remove abnormal tissue
- More advanced cell changes at diagnosis
Patients with these characteristics experienced recurrence about two years after therapy on average.
Testing and Next Steps
The tool was validated on two separate patient groups-one similar to the training dataset and another from different sources. It performed accurately on both, suggesting it will work across different patient populations.
Personalized surveillance could reduce unnecessary procedures for low-risk patients while allowing closer monitoring of high-risk cases. This approach would lower healthcare costs and patient stress.
Collaborators from Johns Hopkins University, Mayo Clinic, Northwestern Feinberg School of Medicine, and eight other institutions contributed data and expertise. The team plans further validation using international datasets from the Netherlands, United Kingdom, Belgium, and Switzerland before applying the tool broadly in clinical settings.
AI for Healthcare applications like this require rigorous AI Research methodology to ensure clinical reliability.