Creative Commons source reading · Machine learning for performance
Artificial Intelligence for Diagnosis of Esophageal Manometry: A Narrative Review
Robalino Gonzaga E; Madej JJ; Ahuja N
Originally published in Curr Gastroenterol Rep · 2026-08-10
From the article
Selected source text. JournalOS removes markup and shortens excerpts where indicated. The full article provides the complete context.
Abstract
AbstractPurpose of ReviewHigh-resolution manometry (HRM), the gold standard for diagnosing esophageal motility disorders, remains constrained by inter-rater variability, limited access to expert esophagologists, and inadequate training infrastructure. Many gastroenterology fellowship programs provide insufficient motility training, and dedicated expert centers remain geographically concentrated. Emerging Artificial intelligence (AI) tools have the potential to automate diagnosis, and augment clinician interpretive capacity and democratizing expert-level motility assessment.Recent FindingsTwenty-two studies encompassing over 5,000 patients were synthesized across three overlapping developmental phases: early machine learning for feature extraction and classification; deep learning deployed toward automated pattern recognition and…
Conclusion
Artificial intelligence has demonstrated compelling technical performance in HRM-based Chicago Classification automation [11, 16]. AI achieves near-expert diagnostic accuracy in controlled research settings and offers consistent, reproducible interpretation superior to human inter-rater agreement in certain tasks [6, 9]. Yet the current evidence base does not yet support clinical implementation. The complete absence of external validation [9], the lack of prospective outcome data [2], and the systematic selection bias toward high-volume expert centers [7, 12] are fundamental prerequisites for responsible clinical translation. Beyond diagnostic capabilities, AI holds transformative potential as an access and training equalizer, extending expert-level manometry interpretation to the community centers, training programs, and global…
Editorial reading guide
- What data created and tested the model, and was evaluation independent?
- What is the cost of a false positive, false negative or out-of-distribution decision?
