Use of artificial intelligence–enhanced virtual patients in educational approaches to medical interview training: a systematic review
BMC Medical Education. 2026-03-04; 26(1):
DOI: 10.1186/s12909-026-08804-9
Gilbert V(1)(2), Philip P(3), Llorca PM(4), Samalin L(4).
Author information:
(1)Department of Psychiatry, CHU Clermont-Ferrand, University of Clermont
Auvergne, CNRS, Clermont Auvergne INP, Institut Pascal (UMR 6602),
Clermont-Ferrand, France. .
(2)Centre Hospitalier Et Universitaire, Service de Psychiatrie B, 58 Rue
Montalembert, Clermont-Ferrand, 63000, France. .
(3)CHU Bordeaux, Service Universitaire de Médecine du Sommeil, University of
Bordeaux, SANPSY, UMR 6033, Bordeaux, France.
(4)Department of Psychiatry, CHU Clermont-Ferrand, University of Clermont
Auvergne, CNRS, Clermont Auvergne INP, Institut Pascal (UMR 6602),
Clermont-Ferrand, France.
BACKGROUND: Effective communication and empathy are essential for
professional–patient relationships, patient care, and the therapeutic alliance.
With the increasing use of artificial intelligence-assisted virtual patients
(AI-VPs), new opportunities to improve interview training are observed. This
systematic review assesses the educational use, benefits, and limitations of
AI-VPs in clinical interview training for healthcare professionals and students.
METHODS: Following PRISMA recommendations, the review involved a systematic
search in PubMed, Google Scholar, and PsycINFO. Eleven studies met the inclusion
criteria. Methodological quality of papers was assessed using the MMAT 2018.
RESULTS: The findings suggest that AI-VPs can improve empathy, communication,
and clinical reasoning, as well as promote perceived self-efficacy, reduce
anxiety, and enhance engagement through immersive experiences. However,
questions remain about the transferability of these skills to clinical practice.
The studies show a lack of consistent theoretical frameworks and standardized
outcome measures. While feedback is widely used, it is not formalized. Technical
limitations, such as issues with speech recognition and nonverbal communication,
can disrupt immersion.
CONCLUSION: AI-VPs represent a promising educational tool for initial interview
training, as it creates learning environments that are scalable, standardized,
and interactive. A step-by-step approach, beginning with standardized patients
and progressing to real patients, may optimize skill acquisition. Robust
longitudinal studies and cost–benefit analyses are needed to support these
findings. Although AI-VPs can improve training in interpersonal and
communication skills, further work is needed to validate its clinical impact and
educational design.
DOI: 10.1186/s12909-026-08804-9
PMCID: PMC13067467
PMID: 41781962
Conflict of interest statement: Declarations. Ethics approval and consent to
participate: Not applicable. Consent for publication: Not applicable. Competing
interests: The authors declare no competing interests.