The integration of artificial intelligence (AI) into colonoscopies is enhancing accuracy, yet a recent study reveals a potential trade-off: doctors’ proficiency may be diminishing as they increasingly rely on AI technology.
Research published in The Lancet Gastroenterology & Hepatology suggests that the routine utilization of AI-assisted colonoscopy systems could result in a 20% decrease in experienced endoscopists’ ability to detect adenomas, which are precancerous growths in the colon, without AI assistance.
The study involved the analysis of colonoscopies performed on 1,443 patients at four centers in Poland, both before and after the implementation of AI over a six-month period. Prior to the introduction of AI, adenomas were detected by clinicians 28.4% of the time. However, three months post-AI implementation, the detection rate declined to 22.4%, marking a notable six percentage-point decrease.
“The findings are worrisome considering the rapid expansion of AI in the medical field,” noted Marcin Romańczyk, co-author of the study from the Academy of Silesia in Poland, emphasizing the necessity for further investigation into AI’s impact on healthcare professionals’ skills in various medical specialties.
Colonoscopy plays a crucial role in preventing bowel cancer by identifying and removing adenomas before they develop into cancer. While AI assistance has demonstrated increased detection rates in several trials, concerns arise from the concept of “deskilling” – the gradual erosion of expertise when clinicians overly depend on automated aids.
The study also casts doubt on previous randomized controlled trials that indicated higher adenoma detection rates with AI support compared to without it.
“It is possible that the non-AI assisted colonoscopy assessed in these trials differs from standard non-AI colonoscopy, potentially influenced by the continuous exposure to AI among endoscopists,” explained Yuichi Mori, another co-author of the study from the University of Oslo in Norway.
The authors caution that the observational nature of the study leaves room for other factors to have influenced the results, underscoring the need for continued research in this area.
