Extended Reality, Telementoring, and Artificial Intelligence-Assisted Assessment in Surgical Training: A Narrative Review.

OBJECTIVE: The Halstedian apprenticeship model in surgical training faces escalating pressure from work-hour restrictions, rising operative complexity, and heightened patient-safety expectations. Extended reality (XR) technologies-virtual reality (VR), augmented reality (AR), and mixed reality (MR)-together with telementoring platforms and artificial intelligence (AI)-assisted assessment, are frequently proposed as responses to these pressures. This narrative review examines what the evidence currently supports, for whom, and with what methodological certainty.

DESIGN: A structured narrative review was performed and reported with reference to PRISMA 2020 principles for source flow and to the SANRA framework for narrative-review quality. Titles and abstracts were screened independently by 2 reviewers and disagreements resolved by consensus. Seminal pre-2014 sources were retained through hand-searching of reference lists for conceptual and historical context. Findings were synthesized narratively in 3 stages (independent reading, inductive coding, consensus theme definition); quantitative pooling was not performed.

SETTING: PubMed/MEDLINE, Scopus, and Web of Science were searched for records published between January 1, 2014, and May 31, 2026, using the string: (("extended reality" OR "virtual reality" OR "augmented reality" OR "mixed reality" OR "XR" OR "VR" OR "AR" OR "head-mounted display") AND ("surgical education" OR "surgical training" OR "surgical simulation" OR "operative training" OR "laparoscopic training")) OR (("telementoring" OR "teleproctoring" OR "remote mentoring" OR "telesurgery") AND ("surgical" OR "surgery" OR "laparoscopic")) OR (("artificial intelligence" OR "machine learning" OR "deep learning" OR "computer vision") AND ("surgical assessment" OR "surgical simulation" OR "surgical skills" OR "operative performance")).

PARTICIPANTS: Not applicable at the patient level; the included sources involved medical students, surgical residents, and practicing surgeons across surgical specialties.

RESULTS: Of 8113 records identified, 252 full-text reports were assessed and 67 met eligibility criteria; 50 sources were prioritized for the synthesis: 40 empirical or secondary evidence sources (VR and immersive simulation including haptics, n = 12; AR and MR, n = 7; telementoring and telesurgery, n = 9; AI-assisted assessment and adaptive learning, n = 11; emerging technologies, n = 1) and 10 conceptual, methodological, or regulatory sources. Thirty-six sources fell within the search window and 14 were foundational pre-2014 sources. VR simulation is supported by the strongest evidence base, including randomized trials and pooled analyses showing reduced operative time and intraoperative error and demonstrable transfer to the operating room; participants are predominantly residents and medical students. AR evidence is concentrated in image-guided orthopedic and spinal navigation performed by practicing surgeons, with 1 randomized trial of AR telestration in laparoscopic training. Telementoring evidence relates predominantly to practicing surgeons and proctored credentialing rather than to residency-level graduate medical education. AI-assisted assessment achieves agreement with expert raters in the reported range of 65% to 77% across global skill domains but remains largely single-center and retrospective.

CONCLUSIONS: The evidence supports XR simulation as a mature adjunct for technical skill acquisition at the "Shows How" level of Miller's pyramid, whereas telementoring and AI-assisted assessment remain, in the main, technologies validated outside graduate surgical education. Evidence is strongest at Kirkpatrick Levels 1-2 and weakest at Level 4. The field's principal deficit is no longer proof of concept but implementation science and patient-level outcome validation.