Augere Medical comments on potential AI-assisted colonoscopy “deskilling”

Artificial Intelligence (AI) during colonoscopy is an emerging technology with the aim to improve the detection rates of polyps during screening and exams for colorectal cancer. Since 2018, different AI systems for this clinical indication have been cleared by several regulatory agencies including the Pharmaceuticals Medical Device Agency (PMDA) in Japan, the US Food and Drug Administration (FDA) and different European Notified Bodies.

Augere Medical (“Augere”) was founded in 2018 based on research by the Oslo University Hospital, Ullevål, and the National Hospital (“Rikshospitalet”) in Norway and in cooperation with Simula Research Laboratory. The company has developed the PolypAID™ platform which was cleared for European market introduction in August 2024.

The over-reaching aim for Augere is to improve patient care following the “do-no-harm” principle. We are dedicated to transparent AI by improving patient outcomes, using AI to augment, not replace, the skills of clinicians and by providing proper training for clinicians using our AI tools. Read our complete transparent AI manifest here.

Since 2015, there has been a dramatic increase in the number of publications for AI assisted colonoscopy. More than 1 500 articles have been published of which more than 30 publications are randomized clinical trials (RCT’s) and half of these are multi-centre RCT’s. Total studied patients in the RCT’s are more than 30 000 patients.

The key findings from these RCT’s are that the adenoma detection rate (ADR) is increased rather than decreased by using AI support and the overall increase is between 8% to more than 20%.

In Norway, Haukeland University Hospital has published data documenting a ~20% increase in ADR during AI-assisted colonoscopies with the interesting finding that younger clinicians see a faster increase in their ADR compared to more experienced gastroenterologists. These findings indicate that AI supports a rapid performance improvement. In Augere’s own clinical experience, we have observed a clear change in physician behaviour when using AI by more precise investigations and longer scope withdrawal times indicating improved diagnostic quality.

The study published in the Lancet and commented in the NRK article is a retrospective, observational study in four Polish centres between September 2021 and March 2022 in 1 443 patients.

Augere finds the observation of “deskilling” of physicians before and after introducing AI-assisted colonoscopies interesting but more speculative than substantiated by robust clinical data. We therefore agree with the authors of the study that “interpretation of these data is challenging”. Furthermore, the study design is not prospective and does not document ADR before AI use, during AI use and potentially after ending AI use and final conclusions for a potential “deskilling” process can therefore not be made.

Augere believes that continued physician education is warranted to improve patient outcomes in this emerging field. We remain committed to improving our technology and educational programs to be a leader in the field of AI assisted diagnosis of gastrointestinal diseases and welcome the continued scientific debate.