Research Round up

22 April, 2026

In this edition, I continue my round up of recent research that’s been published on stoma care and experiences. I’ll be discussing three studies, all of which focus on the role of Artificial Intelligence (AI) in stoma care. Some of you may not be too familiar with the use of AI, whilst others may use AI technologies on a daily basis. It seems inevitable that AI will play an increasing role in our lives over the coming years, and this includes its use in health spheres.

Turkish Study: AI Generated Patient Educational Materials for Ostomates

The first study I’ll share was carried out in Turkey and looked at whether AI ‘chatbots’ can be used to produce good quality patient education materials for ostomates. A chatbot is a software application that uses generative AI systems (that is, they can use what they’ve learned to produce fresh content) and are capable of maintaining a conversation that mirrors human interaction. The researchers in this study wanted to find out if the patient education materials produced were relevant to ostomates’ needs and easy to read and understand.

‘ChatGPT’ was used, a popular, freely available chatbot produced by an OpenAI. The researchers first instructed ChatGPT to carry out a literature review to identify the information needs of ostomates, through which the following topics were identified: stoma care, nutrition, daily life, psychological support and stoma-related problems. ChatGPT was then asked to create educational content covering these five topics. Finally, it converted this content into a flipbook, including pictures and diagrams, that could be read on a screen or printed. The patient education flipbook was then assessed by an expert group of 10 specialist stoma and wound care nurses, using a patient education assessment tool to help them.

ChatGPT was found to provide generally accurate and comprehensive information without any significant medical errors. The information was scored highly on how easy it was to understand (82% on the patient education assessment tool) and on its ‘actionability’ (85%), that is, whether it gives clear, concrete steps the patient can follow in their stoma management. Ease of reading was judged based on guidance from the American Medical Association, which states that patient materials should be targeted at a 12-year-old reading age. The Chat GPT-generated material was found to be pitched at a higher reading age level than this recommendation.

The authors concluded that AI-assisted patient education material for ostomates includes accurate information that’s easy to understand and put into action, though some people may find it difficult to read. However, a key aspect missing from this study‒ which the authors do acknowledge ‒ is that ostomates themselves were not asked what they thought of this information, nor were they asked how they feel about information coming from a chatbot, which seems like an important oversight. Nevertheless, given that the future role of AI in providing health information seems somewhat inevitable, it’s at least reassuring that the information provided to ostomates is of a good standard.

Australian Study: Chatbot Responses to Common Questions on Stoma Management

The second study is on a similar topic. Researchers in Australia were also interested in whether AI chatbots can provide patient education for ostomates, but rather than producing education materials, they looked at the responses chatbots provide to common questions about stoma management. They also wanted to compare the responses from different chatbots. They included two different versions of ChatGPT (an older version, and a newer, expanded version); ‘Gemini’, a chatbot produced by Google; and ‘CoPilot’, which is made by Microsoft, but is powered by OpenAI (who you’ll remember from above also produce ChatGPT).

The questions that the researchers asked each chatbot were based on the 10 most frequently asked questions in stoma nurse appointments at an Australian hospital. This included, “what should I avoid eating or drinking?”; “why is my stoma bag not sticking properly?” and “when can I get my stoma reversal?”. Two researchers independently assessed the responses and compared their assessments, with a third researcher available to resolve any disagreements. They used different assessment tools to decide the quality of responses, how reliable the responses were, and ease of reading.

For all four AI chatbots, their responses to the questions were judged as mostly easy to read, clear and relevant to the questions posed. However, the researchers noted that the reliability of responses was difficult to appraise, because the chatbots usually don’t mention where on the internet they got their information from. In comparing the four chatbots, it was found that CoPilot and the newer version of ChatGPT (version 4.0) produced responses that had better readability, reliability, and overall quality than the other two. ChatGPT and Gemini were found to use a more professional tone in their responses, which the researchers speculate could lead to them being seen as more formal sources of information, but also felt that they could lack empathy. It was found that all four chatbots consistently advised patients to seek guidance from healthcare professionals, therefore framing their responses as a supplement to, not a substitute for, medical advice.

The researchers concluded that AI chatbots are a promising source for responding to ostomates’ questions but are not a substitute for professional advice. They also propose that advancements are needed to allow chatbots to provide more personalised responses.

Once again, a key absence in this study is the view of ostomates themselves on the usefulness of these responses. Also, the questions posed seem to relate primarily to experiences of recent stoma surgery. It would be interesting for future research to explore how AI chatbots respond to questions from people who have lived with their stoma for a longer period.

Dutch Study: Predictive Machine Learning to Support Surgeons’ Decision-Making

The final study I’d like to share is a little different, focusing on how AI can help surgeons’ decision-making (and this one does involve seeking the views of ostomates). Researchers in the Netherlands looked at Predictive Machine Learning. In simple terms, this is where a computer learns patterns from past data to make future predictions. The study investigated whether Predictive Machine Learning could be used to help prevent colorectal anastomotic leakage (CAL) in colorectal cancer patients. CAL happens when, after surgery to join two parts of the colon or rectum, the new connection (anastomosis) doesn’t fully heal. This can mean intestinal content leaks into the abdomen. Predictive Machine Learning can be used during surgery to help the surgeon predict whether this problem will occur. If it is shown that CAL is likely, the surgeon can carefully choose where to place the stoma to reduce the likelihood of this leakage from happening after surgery.

The researchers wanted to find out what people with colorectal cancer thought about Predictive Machine Learning being used to help surgeons. They carried out online interviews and focus groups via ‘Zoom’ with 19 people from five different countries ‒ the UK, Netherlands, Germany, France and Sweden (although it’s unclear if groups were held with people from each country separately, or a mix of people from different countries).

Everyone who took part expressed optimism that Predicative Machine Learning could be better at assessing the risk of CAL than surgeons by themselves, and could therefore improve surgical decision making. However, some of the participants stressed that the AI model needs to learn from, or be ’trained on’, high quality, up-to-date datasets, to make sure it’s as accurate as possible.

Participants also expressed some concerns that Predictive Machine Learning could reduce the human element of patient care. They wanted surgeons to integrate the AI insights with their knowledge and understanding, but for the surgeon to be the one that makes the final decision. Whilst some participants felt it would be important for patients to be given clear information about how Predictive Machine Learning is used, others did not feel that there is a need for the patient to be told. They made comparisons with other fields such as radiology, where AI may be used in medical scans without the patient being aware. It may be interesting for a future study to explore surgeons’ views on this topic also, particularly in light of the patient findings.

Whilst the inclusion of AI technology was largely seen by participants as being a positive addition to colorectal surgery, this study shares similarity with the others discussed, in that AI isn’t seen as being able to take the place of healthcare professionals’ advice or expertise (or not yet at least). Whilst AI may represent the future of healthcare, it seems that in relation to colorectal surgery and stoma management, the human element is still important in the present.

References

• Hmido SB, Rahim HA, Ploem C, Haitjema S, Damman O, Kazemier G, Daams F. Patient perspectives on AI-based decision support in surgery. BMJ Surgery, Interventions, & Health Technologies. 2025 Apr 2;7(1):e000365.
• Lim B, Lirios G, Sakalkale A, Satheakeerthy S, Hayes D, Yeung JM. Assessing the efficacy of artificial intelligence to provide peri‐operative information for patients with a stoma. ANZ Journal of Surgery. 2025 Mar;95(3):464-96.
• Yüceler Kaçmaz H, Kahraman H, Akutay S, Dağdelen D. Development and Validation of an Artificial Intelligence–Assisted Patient Education Material for Ostomy Patients: A Methodological Study. Journal of Advanced Nursing.