Gastroenterology & Artificial Intelligence: 3rd Annual Global Artificial Intelligence Summit | September 2021-5- Utilizing AI and Machine Learning in IBD Diagnosis and Prognostication

5- Utilizing AI and Machine Learning in IBD Diagnosis and Prognostication

Pdf Summary

This document discusses the utilization of Artificial Intelligence (AI) and Machine Learning (ML) in the diagnosis and prognostication of Inflammatory Bowel Disease (IBD). The author, Dr. David T. Rubin, highlights the potential benefits of AI and ML in various aspects of IBD management, including diagnosis, prognosis, treatment selection, treatment response, disease monitoring, care delivery, and clinical trial efficiency.<br /><br />The current management of IBD is characterized by delays and disease progression, with patients experiencing symptoms, contacting their doctors, waiting for appointments, and starting therapy. The STRIDE 2 treatment targets for both Crohn's Disease and Ulcerative Colitis emphasize the importance of achieving symptomatic remission, mucosal healing, and absence of disability.<br /><br />Various applications of AI in IBD include diagnostics, treatment response prediction, prognostication, case delivery, and understanding the disease's pathogenesis. Multiple data sources, such as age, laboratory values, biomarkers, imaging data, and endoscopy findings, can be incorporated into machine learning models to predict outcomes.<br /><br />Deep learning models have been trained using capsule endoscopy images to identify and classify Crohn's Disease lesions. Class activation maps help validate the model's ability to detect clinically relevant endoscopic features. Another study used deep learning models to detect intestinal strictures from capsule endoscopy images and evaluate their ability to distinguish between different pathology groups.<br /><br />A deep neural network called DNUC predicts histologic and endoscopic remission in Ulcerative Colitis. Machine learning algorithms have been developed using routine laboratory results to predict remission status in patients on thiopurines.<br /><br />The Mucosal Healing Index, which applies AI to mucosal healing markers, demonstrates high accuracy regardless of disease location. Additionally, the UChicago IBD Biosensor Study collects biosensor data, quality of life data, and clinical data to monitor and manage IBD.<br /><br />Preliminary evidence suggests that Fitbit step measurements can be predictive of flares in IBD.<br /><br />Overall, the use of AI and ML in IBD management has the potential to streamline disease monitoring and improve patient outcomes.

Keywords

Artificial Intelligence

Machine Learning

Inflammatory Bowel Disease

Diagnosis

Prognostication

IBD Management

Treatment Selection

Treatment Response

Disease Monitoring

Clinical Trial Efficiency