The study, conducted by Chia-Shan Wu of the National Tainan Junior College of Nursing and Shih-Yeh Chen of the Department of Engineering Science at National Cheng Kung University, was designed as an explanatory sequential mixed-methods trial. Between March and April 2026, 106 nursing students from two intact classes took part in a five-week intervention embedded in their psychiatric nursing education. Fifty-six students used an AI-assisted gamified mobile learning system, while fifty students used a comparable gamified system without AI support. The researchers measured three outcomes before and after the intervention: psychiatric nursing knowledge, attitudes toward mental illness as captured by the Mental Illness Clinicians’ Attitudes Scale, and reflective thinking.

The choice of outcomes reflects a long-standing frustration in mental health education. Students can memorize diagnostic criteria and pharmacological mechanisms, yet still carry the stigma, fear, and distance that have historically characterized psychiatric care. The World Health Organization has repeatedly emphasized that person-centred mental health services require practitioners who can engage with patients’ lived experience, not merely manage symptoms. Building that capacity, the researchers argue, requires more than lectures; it requires opportunities to practice perspective-taking and to reflect on one’s own assumptions in a safe environment.

That is where the AI component comes in. In the AI-assisted condition, students interacted with virtual patients and received feedback generated by artificial intelligence, which responded to their choices and prompted them to consider the patient’s point of view. The system was grounded in Self-Determination Theory, a psychological framework that holds that motivation flourishes when learners experience autonomy, competence, and relatedness. Gamified elements such as challenges and progression were intended to sustain engagement, while the AI layer was intended to make each interaction feel responsive and personal, simulating something closer to a real clinical conversation than a fixed multiple-choice scenario.

The quantitative results tell a nuanced story. Psychiatric nursing knowledge increased in both groups, and the researchers found no evidence that the AI-assisted group improved more than the control group on this measure. The effect size was small and statistically non-significant, with a p-value of .579 and a Cohen’s d of 0.11, with a 95 percent confidence interval spanning −0.27 to 0.49. In other words, when it comes to acquiring core psychiatric knowledge, a well-designed gamified learning system may be enough, and the AI layer did not add measurable advantage.

But the picture changed sharply for the more human dimensions of learning. Students in the AI-assisted condition showed significantly greater improvements in their attitudes toward people with mental illness, with a p-value of .041 and an effect size of d = 0.32, with a confidence interval of −0.06 to 0.71. The difference in reflective thinking was even more striking: the AI-assisted group improved substantially more, with p < .001 and d = 0.58, and a 95 percent confidence interval of 0.19 to 0.97, which lies entirely above zero. The researchers analyzed these longitudinal outcomes using generalized estimating equations, a statistical approach suited to repeated measurements in a quasi-experimental pretest–posttest control-group design.

Reflective thinking, the outcome with the strongest effect, is often described as the engine of professional growth in nursing. It is the habit of stepping back from an encounter, examining one’s own reactions and assumptions, and asking what the patient experienced and what could be done differently. The qualitative half of the study, based on semi-structured interviews with eight students, offered a window into why the AI system may have strengthened this habit. Students reported that the AI-generated feedback and the virtual patient interactions helped them think more deeply about their own responses and become more aware of how patients might perceive their words and behavior. In effect, the AI acted as a mirror, returning students’ decisions to them reframed through the patient’s perspective.

This mechanism matters because attitude change is notoriously difficult to teach. Traditional anti-stigma interventions often rely on contact with people with mental illness, which is powerful but logistically demanding and variable in quality. Simulated patients are expensive and inconsistent across settings. An AI-driven virtual patient, by contrast, can be available at any hour, on any phone, and can generate feedback tailored to each student’s specific choices. If such systems can reliably nudge attitudes in a more empathetic direction, they could offer nursing schools a scalable complement to clinical placements, particularly in regions where mental health training resources are scarce.

The authors are careful about the limits of their findings, and the caution is worth emphasizing. The study was a quasi-experiment using intact classes rather than random assignment, which means pre-existing differences between the two groups cannot be fully ruled out. The five-week intervention was short, the sample came from a single institution, and the interview subsample was small. Most importantly, the researchers explicitly note that the study does not establish improvements in person-centred care competence or in actual clinical practice. Improved attitudes and stronger reflective thinking are promising precursors to better care, but the chain from classroom simulation to bedside behavior remains to be demonstrated.

Even with those caveats, the study offers a concrete signal in a fast-moving field. Funded by Taiwan’s National Science and Technology Council through project NSTC 114-2410-H-439-004 and approved by the Human Research Ethics Committee of National Cheng Kung University, the trial suggests that the value of AI in health professional education may lie less in delivering facts, where textbooks and lectures already perform well, and more in cultivating the relational and reflective capacities that have always been the hardest to teach. As generative AI systems grow more capable of sustaining realistic, responsive dialogue, the possibility of giving every nursing student thousands of low-stakes practice conversations with virtual patients moves from speculation toward routine. If future trials confirm that these interactions translate into more compassionate clinical care, the humble gamified phone app tested in Tainan may come to be seen as an early prototype of how machines teach humans to be more human with patients.

Subject of Research: Preparing future nurses for person-centred mental health care through AI-assisted gamified learning: a mixed-methods study

Article Title: Preparing future nurses for person-centred mental health care through AI-assisted gamified learning: a mixed-methods study

Article References: Wu, C.-S., & Chen, S.-Y. (2026). Preparing future nurses for person-centred mental health care through AI-assisted gamified learning: a mixed-methods study. BMC Nursing. https://doi.org/10.1186/s12912-026-05488-w

Image Credits: AI Generated

DOI: 10.1186/s12912-026-05488-w

Keywords: Preparing, future, nurses, person-centred, mental, health, care, AI-assisted, gamified, learning, mixed-methods, scientific research