Docbot Announces Results of Study Evaluating Deep Learning Platform for Ulcerative Colitis Published in Journal, Gastroenterology
First to demonstrate deep learning algorithm can be trained to automatically predict levels of ulcerative colitis severity from endoscopic videos
Docbot, Inc., a privately-held company developing leading artificial intelligence (AI)-driven healthcare applications, announced results from a study evaluating machine reading using Docbot’s investigational deep learning AI platform for automated disease activity scoring in patients with ulcerative colitis. The paper, in collaboration with Eli Lilly and Company, titled “Central reading of ulcerative colitis clinical trial videos using neural networks,” was published online today in the leading journal in the field, Gastroenterology.
Endoscopic disease activity scoring in the treatment of ulcerative colitis, an inflammatory bowel disease (IBD) with no known cure, is useful in clinical practice but infrequently done. Endoscopic disease activity scoring is required in clinical trials, but is expensive and time-consuming due to the need to train and employ human central readers.
Recommended AI News: NTT DOCOMO Selects NEC as a Vendor for Standalone 5G Mobile Core
Through data analyzed retrospectively by Docbot’s Ultivision AI platform, the study showed that a deep learning algorithm can be trained to automatically predict levels of ulcerative colitis severity from full-length endoscopy videos. The study showed excellent machine-to-human inter-observer agreement for the Ulcerative Colitis Endoscopic Index of Severity (UCEIS) and the endoscopic Mayo score. The high agreement scores provide promise that a deep learning algorithm may lead to faster, more streamlined scoring of endoscopy videos.
Andrew Ninh, CEO and Co-Founder of Docbot, commented, “We’re excited to announce results of our collaboration for developing an AI for ulcerative colitis with Lilly, who we recognize as a visionary leader in the pharmaceutical industry. We’re pleased to partner and collaborate with their team to spearhead a game-changing effort that may positively impact patients suffering with ulcerative colitis.”
James Requa, Head of AI at Docbot and co-primary author of the study, added, “Publication by Gastroenterology validates our clinical AI research by the scientific community. We believe our study is the first to demonstrate that artificial neural networks can be trained to predict levels of ulcerative colitis severity from routinely obtained full-length endoscopy videos, rather than curated still images, with high inter-rater agreement to human central readers. The study suggests that machine reading of endoscopic videos should be further considered and possibly introduced in a stepwise fashion into ulcerative colitis clinical trials, and is expected to save time and cost, reduce reader variability, and lay the foundations for automated ulcerative colitis disease activity scoring in clinical practice.”
Klaus Gottlieb, a Medical Fellow at Lilly and co-primary author of the study: “We’ve collaborated to develop an algorithm that is easy to use, using colonoscopy videos from 75 sites in 14 countries with no additional burden on the examiner. The human central readers were independent and not part of our research team, eliminating much of the bias seen with single institution work.”
Based on the results from this published study, Docbot and Lilly are continuing to work collaboratively together to explore new ways of automating the process of endoscopic disease activity scoring in ulcerative colitis to elevate clinical care and facilitate clinical research.