Noticias

AI-selected burnout support for nurses: what is known?

11:01 pm 1 November 2026 0 Comments

Many nurses are asked to try a wellbeing app, and some of those apps now say they use artificial intelligence. A South Korean study reported lower burnout among nurses using an app that selected cognitive behavioural therapy (CBT) based modules for them. It does not show that AI is better than human support or therapy, and its reported numbers conflict, so the findings are best treated as preliminary.

What the researchers did

The study was published on 3 July 2026 in JMIR mHealth and uHealth. The app, called Nurse Healing Space, used an algorithm to select CBT-based modules matched to each participant's burnout profile, job stress and coping characteristics. The options included mindfulness meditation, storytelling and reflective writing, laughter therapy, and acceptance and commitment therapy.

The app selected modules for people. It was not described as chatbot therapy, so this is not a study of an AI therapist.

Nurses in the intervention group were asked to complete three sessions a week, each lasting 10 to 15 minutes. The comparison group received online information about burnout management. That means the app was compared with information, not with a therapist, and not with modules chosen by a person.

Illustration for the article: AI-selected burnout support for nurses: what is known?
Illustrative image. AI-generated editorial illustration.

What they found

The numbers of participants do not line up, and the difference is worth setting out clearly. The abstract says 125 nurses were enrolled and randomised, with 62 assigned to the intervention and 63 to the control group. The Methods section instead says 162 eligible participants were randomised, and that 137 completed both follow-ups. The paper then says 125 people were included in the final analyses after 12 outlier cases were removed. That last step is consistent, since 137 minus 12 is 125. What remains unresolved is the number originally randomised: 125 in the abstract against 162 in the Methods.

The researchers reported group effects for three kinds of burnout: client-related, personal and work-related. The reported F statistics were 7.548, 6.533 and 38.194 respectively. They found no significant differences between the groups for coping strategies, job stress or stress response.

An important qualification is that the paper reported no significant group-by-time interaction. That interaction is the statistical test of whether one group changed more over time than the other. The paper did report improvements within groups. But a change within one group does not by itself show that the group improved more than the comparison group. The results should therefore not be turned into a claim that AI selection produces a faster or better improvement.

The paper also described qualitative findings. Some participants were receptive to the app and reported greater self-awareness and reflection on coping strategies that they might not have chosen for themselves. That gives a window into how people experienced the app. It is not evidence that the app works for everyone, or that it can stand in for professional care.

What this does and does not show

The study lasted four weeks and relied on self-reported outcomes, so it cannot say whether any change would last. The paper itself points to sustainability as a question for future research.

Several design limits make the findings less secure. There was no intention-to-treat analysis, which means the analysis did not keep everyone in the group to which they were originally assigned. Job stress and stress response differed between the groups at the start. The paper notes the possibility that people knew which group they were in, and there were no objective measures of how much participants used the app.

There was also no human-selection comparator. Nobody compared the algorithm's choices with modules chosen by a clinician, a nurse manager or the participant. So even if a nurse found the app helpful, the study cannot show that the AI selection was responsible, or that it did better than a person choosing. The two groups also received quite different things, which is relevant when judging the result.

Burnout is shaped by staffing, rosters, workloads and workplace culture, which a phone app cannot change. Health systems and workplace rules differ between countries, and this study of South Korean nurses is not Australian advice. This article is general education, not advice, and it does not recommend for or against any app or tool.

If burnout is affecting you or someone you work with, a digital programme is at most one possible source of support. It is not an established replacement for professional care. If you are in distress, in Australia Lifeline is available on 13 11 14, and 000 is the number for an emergency.

Questions to ask

  • If an app uses an algorithm to choose what I see, what is it basing that choice on, and what evidence shows that choosing this way helps?
  • Did the research explain clearly how many people were randomised, how many finished, and how the results were analysed?
  • If my burnout comes mainly from workload, rosters or staffing, what changes beyond personal skills are needed, and who could I raise this with at work?

Source: https://mhealth.jmir.org/2026/1/e85986

AI-generated editorial illustration. The scene is fictional and does not depict a study participant or demonstrate treatment effectiveness. This article summarises published research in our own words and is general information, not personal health advice.

Comments are disabled

Buscar
Archives
Recent entries
Categories