Research shows that gastroenterologists generally trust and accept the use of AI medical tools in their clinics and hospitals.

Artificial intelligence (AI) is permeating many aspects of healthcare, promising accurate diagnoses, better management decisions, and improved outcomes for both patients and healthcare systems. However, the successful implementation of AI technologies in clinical practice requires trust and acceptance of the use of such tools among healthcare professionals.

Now, using the treatment of gastrointestinal diseases as a case study, an international study led by Nanyang Technological University, Singapore (NTU Singapore) has found that gastroenterologists generally trust and accept AI medical tools.

The paper, “Risk perception, acceptance, and trust in the use of AI in gastroenterology care in the Asia-Pacific region: a web-based survey study,” JMIR AI.

Surveying 165 gastroenterologists and gastrointestinal surgeons across the Asia Pacific region, a research team led by NTU Singapore found that eight in 10 said they would accept and trust the use of AI-powered tools to diagnose and evaluate colorectal polyps – benign tumours in the colon that can become cancerous.

When it comes to using AI to advise endoscopists on whether to remove polyps found in the intestines of patients undergoing colonoscopy, seven in 10 people said they would accept and trust this AI-assisted application.

The researchers found no differences in levels of acceptance between male and female doctors, doctors working in public and private organizations, or doctors working in large hospital departments and small group practices.

But years of experience did matter: While younger physicians were expected to be more open to using technology in clinical decision-making, gastroenterologists with fewer than 10 years of clinical experience perceived AI-enabled medical tools to be more risky than their colleagues with more than 10 years of experience, the study found.

The team of scientists from Singapore, China, Hong Kong and Taiwan said their findings highlight the need for further research into what influences doctors' acceptance of AI in healthcare.

“In this study, we focused on the use of AI in gastroenterology as this speciality, which makes heavy use of image-based diagnosis and surgical or endoscopic interventions, is one area where we believe AI technologies can be readily applied in clinical management,” said Associate Professor Wilson Goh from NTU's Lee Kong Chian School of Medicine (LKCMedicine), who led the study.

“This is one of the earliest reports on AI risk awareness, acceptance and trust among gastroenterologists, with a specific focus on the Asia-Pacific region.”

“Although survey participants perceived certain AI technologies to be risky, most experts still trusted and accepted these applications, highlighting the complex relationship between the complexity and acceptance of AI technologies,” said Associate Professor Goh, who is also co-director of NTU's Centre for Biomedical Informatics.

“AI has the potential to revolutionize healthcare, but its integration into healthcare requires a deeper understanding of the factors that underpin clinician trust and acceptance of AI-powered medical tools.”

“It is intriguing to find that more experienced gastroenterologists have a lower perception of the risks of AI tools,” said Professor Joseph Song, NTU's Senior Vice-Chancellor for Health and Life Sciences and director of the Centre for Medical AI, and one of the study's co-authors, a renowned gastroenterologist.

“As senior gastroenterologists gain more clinical experience in treating colorectal polyps, these clinicians may become more confident in their own medical expertise and practice, and therefore more confident in making clinical decisions when new technologies are introduced.”

Meanwhile, a general lack of trust when there is a discrepancy between AI and human experience could be one of the reasons why less experienced doctors perceive AI as more risky when it comes to invasive surgeries such as removing colon polyps, said Prof Sung, who is also dean of NTU's LKC School of Medicine.

“Giving greater importance to AI, as we have done in NTU LKCMedicine's recently updated curriculum, could help mitigate risk aversion and promote responsible use of AI in clinical practice,” he added.

Professor May O. Lewin, NTU's Wee Kim Wee School of Communication Dean and another study co-author whose research focuses on health communication, suggested future research could incorporate patients' perspectives by assessing the circumstances in which they might have concerns about the use of AI technologies in their health conditions.

She added: “It is important to capture the perspectives of other stakeholders, such as nurses, endoscopy assistants and the general public, to better understand how their views coincide or conflict. This will help us to more practically resolve complex trust and acceptance issues and develop value propositions and effective policies.”

How the study was conducted

For this study, the scientists developed a questionnaire based on questions and statements adapted from validated frameworks and models.

Participants were asked to rate their level of agreement with questions or statements designed to assess their level of trust, acceptance, and risk perception towards the use of AI in gastroenterology.

• Participants were also presented with three different healthcare scenarios in which AI could be applied.
• For detection: To identify the presence of colon polyps and improve the detection rate of polyps that may become cancerous.
• Characterization: Evaluate the pathological properties of polyps and predict the risk of colon polyps becoming cancerous
• For intervention: Endoscopy to guide removal of polyps

For each healthcare scenario, participants were asked to rate the extent to which they agreed with statements such as the following, which assessed their perceived risks and trust in the AI ​​tool:

• I expect that AI diagnosis will entail significant risks.
• I'm ready to try this myself.

They were also asked to rate their beliefs in statements assessing their acceptance of AI tools, such as:

• Do you think machine learning algorithms can, in some cases (like the ones mentioned above), perform better than humans?

Each participant's scores were then tallied and used in statistical analyses to examine how the elements of risk perception, acceptance, and trust interacted with each other.