Artificial Intelligence Aids Providers in Maintaining Colonoscopy Quality

A study published in JAMA Network Open found that artificial intelligence (AI)-based systems successfully assisted providers in late-day colonoscopy sessions, indicating their potential to improve endoscopist performance and support adenoma detection.

According to the study, the time of day is a factor that influences the quality of adenoma detection, which takes place during colonoscopies. Although previous research indicates that AI can improve adenoma detection rate (ADR), providers are unclear regarding its ability to improve ADR at different times throughout the day.

To gauge a sense of how the use of AI systems affects the common, time-related decline in ADR, researchers conducted a cohort study of two randomized controlled trials (RCTs) from Renmin Hospital of Wuhan University.

The study population comprised a total of 1,780 patients. Researchers assigned patients undergoing colonoscopy to one of two groups: an AI-assisted group or an unassisted group. There were two study periods, one extending from June 18, 2019, to Sept. 6, 2019, and another from July 1, 2020, to Oct. 15, 2020.

Researchers compared the ADR rates of early and late colonoscopy sessions per half day, both pre-and post-AI intervention.

A total of 1,041 procedures took place in early sessions, which included 357 in the non-AI group and 684 in the AI group. These sessions took place between 8:00 am and 10:59 am or 1:00 pm and 2:59 pm.

A total of 739 procedures occurred during the late sessions, 263 of which were not assisted by AI and 476 that included AI assistance. These sessions occurred from 11:00 am to 12:59 pm or 3:00 pm to 4:59 pm.

Researchers noted that the non-AI group reported an early session ADR of 13.73 percent, which was noticeably higher than that of later sessions, which reported 5.7 percent. However, researchers did not find any significant difference in ADR between early and late sessions following the implementation of the AI system.

They also found that the AI displayed more advanced assistance capabilities in later sessions as opposed to earlier sessions. The AI system showed the highest assistance capability in the last hour per half day, "which demonstrated the potential benefit of using AI to maintain high quality colonoscopy procedures," the study authors wrote.

Based on these findings, researchers noted that AI systems have the potential to ensure colonoscopy quality. Further, these types of tools may be able to elevate endoscopist performance within large screening programs and at centers with high workloads. 

However, some limitations of this study include the validation of results at only a single center, the lack of acknowledgment of differing environments, diets, and lifestyles, and the need for regional samples to determine the applicability of findings.

AI is increasingly being applied to gastroenterology procedures.

In August 2022, New Jersey-based Hackensack University Medical Center announced that it would use funding from the Health Equity Assistance Program to install AI-assisted colonoscopy technology to assist underserved communities.

Known as GI Genius, the tool aims to detect colorectal polyps to support early cancer detection and clinical decision-making. The hospital planned to use the tool to increase screening rates in an effort to mitigate colorectal cancer among underserved communities.