Daijiworld Media Network - New Delhi
New Delhi, Aug 13: As Artificial Intelligence (AI) continues to gain traction in healthcare, a new study has raised red flags about its unintended consequences. Published in The Lancet Gastroenterology and Hepatology, the study suggests that doctors who regularly rely on AI-assisted colonoscopies may lose their edge in detecting precancerous colon growths — when performing procedures without AI.
Colonoscopy is a vital tool in cancer prevention, allowing doctors to spot and remove adenomas before they turn cancerous. AI tools have been praised for boosting adenoma detection rates, but concerns are now surfacing over their long-term effect on medical skills.
“This is the first study to indicate that AI might actually reduce a clinician’s ability to perform a crucial task,” said lead author Dr Marcin Romanczyk of the Academy of Silesia, Poland. “The findings are concerning, especially as AI adoption in healthcare is accelerating.”
The study was conducted between September 2021 and March 2022 at four colonoscopy centres in Poland. Researchers reviewed 1,443 colonoscopies performed by 19 experienced endoscopists (each with more than 2,000 procedures to their credit).
They compared detection rates before and after AI was introduced:
• Before AI exposure: Adenoma detection rate was 28.4% (226 out of 795 procedures).
• After regular AI use: That dropped to 22.4% (145 out of 648), reflecting a 20% relative and 6% absolute decline in detection performance during non-AI-assisted colonoscopies.
• With AI assistance: The detection rate was 25.3% (186 out of 734 procedures).
Researchers stressed that while AI still improved overall detection during assisted procedures, its regular use appeared to erode doctors’ ability to function independently.
The study also acknowledged its limitations, such as being observational in nature, which means other factors might have influenced outcomes.
Dr Romanczyk urged the medical community to explore AI’s broader implications, cautioning that skill degradation could affect more fields as AI becomes a staple in diagnostics and treatment.