Tufts University Medical Center Malden, MA, United States
Erik A. Holzwanger, MD1, Mohammad Bilal, MD2, Jeremy Glissen Brown, MD3, Ishita Barua, MD3, Mandeep S. Sawhney, MD3, Douglas Pleskow, MD3, Tyler M. Berzin, MD, FACG3 1Tufts University Medical Center, Malden, MA; 2Minneapolis VA Health Care System, Minneapolis, MN; 3Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA
Introduction: In recent years there have been several published clinical trials evaluating the role of various aspects of artificial intelligence (AI) in gastroenterology. New guidance from the CONSORT (Consolidated Standards of Reporting Trials) and SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) Steering Group represents an important step towards standardizing the quality of AI trial reporting. However, all currently available clinical trials in gastroenterology were performed before these guidelines were developed. Therefore, we sought to evaluate the adherence to the SPIRIT-AI standardized reporting guidelines in AI clinical trials in gastroenterology.
Methods: Multiple databases were searched (from inception to June 2021) and clinical gastroenterology trials applying AI and machine learning (ML) were identified. Only prospective trials were included. The studies were evaluated to determine if they met each of the 13 core recommendations of the SPIRIT with AI extension guidelines. Each recommendation was given 1 point if they met criteria and 0 points if they did not. Proportion of adherence to the SPIRIT-AI recommendations was reported as percent value.
Results: A total of eight prospective clinical trials evaluating the role of AI/ML in gastroenterology were identified. Six of these focused-on AI for colon polyp/adenoma detection, while one was pertaining to Barrett’s esophagus and another evaluating blind spots during esophagogastroduodenoscopy. The adherence to the SPIRIT-AI extension ranged from 84.6% to 100%, with five of the studies scoring 100%. The mean was 95.62%. The most commonly missed recommendations out of the 13 included specifying the procedure for acquiring and selecting the input data for the AI intervention along with specifying the output of the AI intervention.
Discussion: Our analysis reveals that all of the previously published artificial intelligence related clinical trials in the field of gastroenterology have high adherence to the SPIRIT-AI extension guidelines. This shows that previous trials adhered to high-quality methodological standards and that the SPIRIT-AI extension guidelines are practical and relevant to current clinical trial methodology.
Disclosures: Erik Holzwanger indicated no relevant financial relationships. Mohammad Bilal indicated no relevant financial relationships. Jeremy Glissen Brown indicated no relevant financial relationships. Ishita Barua indicated no relevant financial relationships. Mandeep Sawhney indicated no relevant financial relationships. Douglas Pleskow: Boston Scientific – Consultant. Fuji – Consultant. Medtronic – Consultant. Olympus – Consultant. Tyler Berzin: Boston Scientific – Consultant. Fuji – Consultant. Medtronic – Consultant.
Erik A. Holzwanger, MD1, Mohammad Bilal, MD2, Jeremy Glissen Brown, MD3, Ishita Barua, MD3, Mandeep S. Sawhney, MD3, Douglas Pleskow, MD3, Tyler M. Berzin, MD, FACG3. P1467 - Adherence to the SPIRIT-AI Standardized Reporting Guidelines in Artificial Intelligence Clinical Trials in Gastroenterology, ACG 2021 Annual Scientific Meeting Abstracts. Las Vegas, Nevada: American College of Gastroenterology.
