Good doctors do not simply tell patients what to do. They explain the options, listen to the patient’s values and concerns, and work towards making a shared decision with their patients. This is called shared decision-making, or SDM. SDM is a key part of patient-centred care, but opportunities for structured SDM training are limited, both at medical school and in clinical practice.
Traditional clinical communication training often uses simulated patients, actors who play the role of a patient. This works well, but it is expensive and hard to offer at scale. For many educational institutions this is very much a premium option. A new study from the Tübingen Institute for Medical Education in Germany tested an alternative using AI. It was published in JMIR Medical Education in July 2026.
Medical students and practising doctors had a 20-minute written chat with an AI-simulated patient. After the chat, they received automated feedback on how well they had involved the patient in the decision. The researchers had three aims. They wanted to check how closely the AI’s ratings matched ratings by human experts. They also compared the performance and attitudes of students and doctors. Finally, they measured whether participants felt more confident about their communication skills after the practice.
Participants found the AI-simulated conversation authentic and potentially useful for SDM training. The authors say the study helps to test OPTION-12, a scale that measures how far clinicians involve patients in decisions, in consultations with an AI patient. They also suggest that chatbot-based training can give precise, structured feedback on SDM performance.
The authors are careful about their conclusions. This was a feasibility study, so an early study that tests whether an approach can work. They note that the design and the small number of participants limit what we can conclude and that larger studies with control groups are needed to show whether the approach is effective.
The Tübingen study is part of a fast-growing body of research on AI-simulated patients. SLC’s article The AI Patient Will See You Now reviews several recent studies. In one randomised trial, an AI patient called SOPHIE helped learners show more empathy and explain things more clearly in conversations about serious illness. In a UK study, medical students who practised with a voice chatbot as the patient felt more confident about difficult conversations.
A second SLC article, AI standardised patients: a five-agent approach, describes a study from China in which five AI components share the work of “being a patient”: one manages emotions, one holds the case details, and others interpret questions, write replies and check them. This gave more accurate and consistent answers than a single AI model.
SLC’s own AI Care Communication Trainer, presented by Chris Moore at the recent English for Healthcare conference in Kraków, uses similar ideas. Overseas-born care workers practise everyday conversations, such as talking about pain or falls, with a voiced AI service user. An AI tutor then scores their language, empathy, clarity, professionalism and active listening.
AI patients have clear attractions for communication skills teaching. Learners can practise as often as they like, at any time, without the pressure of a live audience. They get immediate feedback. For learners working in a second language, this kind of low-stress practice may be especially valuable.
There are some points to keep in mind. The conversation in this study was written, not spoken, and real consultations also depend on tone of voice, body language and silence. The study was also carried out in German with German-speaking participants. We do not yet know how well these tools work for second-language speakers, or for the cultural side of communication. The research so far highlights that AI practice is likely to work best alongside human teaching, not instead of it.
Back to Menu ↩ Metaphors can tell us a lot about how people think. A new study, published in Acta Psychologica, uses metaphors to explore
Back to Menu ↩ In many countries, a large part of the healthcare workforce has trained overseas. In the UK for example, international medical graduates
Back to Menu ↩ Every day, healthcare workers meet patients who do not share their language or their cultural background. Language is only part of
