Truace
Subscribe
60/100

Moderate evidence

Cite
More
Export CSVExport JSONEmailLinkedInX

Neurosurgery residents trained with an AI virtual patient scored higher in a 157-resident study

Study title: Application of an Artificial Intelligence Simulated Patient Model in Teaching Neurosurgical Decision-Making

Evidence type
Observational study
Sample
n = 157
Sources
1 sourcePeer-reviewed
Published
2 min read
Truvace record
TRV-2026-2459cite or check it

Study title: Application of an Artificial Intelligence Simulated Patient Model in Teaching Neurosurgical Decision-Making

A retrospective study of 157 neurosurgery residents found that those trained with an AI-simulated patient scored significantly higher on a surgical decision-making exam than those taught by traditional case discussion (p < 0.001). The 72 residents in the AI group trained between January 2025 and April 2026.

Read the study abstract

AIM: Training in neurosurgical decision-making remains a challenging component of standardized residency training. Traditional teaching models are limited in their ability to simulate the uncertainty and complexity encountered in real-world clinical scenarios. As a novel educational tool, the potential value of artificial intelligence (AI)-simulated patient models in surgical decision-making training requires further evaluation. This study aimed to explore the application of an AI-simulated patient model in neur…

The gain

In a retrospective study of 157 neurosurgery residents, the 72 who trained with a DeepSeek-based AI virtual patient scored higher on a surgical decision-making exam than the 85 taught by traditional case discussion.

In brief

The AI group also scored higher on all five exam dimensions: localization diagnosis, surgical approach, risk identification, emergency management and surgical timing. More of them reached the excellent threshold of over 85 points, and they gave higher satisfaction scores.

The groups were not randomized. The control group of 85 residents trained earlier, from January 2023 to December 2024, so the authors describe the AI model as associated with better performance rather than proven to cause it.

Main points

  1. The study covered 157 residents in one neurosurgery department's standardized residency training, from January 2023 to April 2026.
  2. The experimental group (72 residents) trained with a virtual patient system powered by the DeepSeek large language model.
  3. The control group (85 residents) received traditional case discussion teaching from 2023 to 2024.
  4. The virtual patient supported dialogue-driven history taking, image interpretation, surgical planning and risk prediction with real-time feedback.
  5. Decision-making was assessed with an objective structured clinical examination across five dimensions.
  6. Scores in all five dimensions were higher in the AI group (all p < 0.05).
  7. A higher share of the AI group scored above 85 points in each dimension, and satisfaction scores were also higher (p < 0.001).

The full story

The study was run in a neurosurgery department's residency programme. Residents were split by when they trained: those from January 2023 to December 2024 got traditional case discussion, and those from January 2025 onward used the AI-simulated patient.

The AI system let residents take a history through dialogue with a virtual patient, interpret images, plan surgery and predict risk, with feedback in real time. It was built on DeepSeek, a general-purpose large language model.

After teaching, an objective structured clinical examination scored residents on localization diagnosis, surgical approach, risk identification, emergency management and surgical timing. The AI group beat controls overall and on each dimension, and more of them exceeded 85 points. A questionnaire also showed higher teaching satisfaction in this group.

Talk it overFour questions for a class, a team or the dinner table

  1. Who gets this benefit, and who was left out of the study or story?

    The gain In a retrospective study of 157 neurosurgery residents, the 72 who trained with a DeepSeek-based AI virtual patient scored higher on a surgical decision-making exam than the 85 taught by traditional case discussion.

  2. How do we know? What can this kind of evidence show, and what can’t it?

    The evidence This rests on an observational study with a sample of 157, from Annali Italiani di Chirurgia.

  3. What does it leave open? What else would you want to know before acting on it?

    The record’s own caveat The study was retrospective and compared two cohorts trained in different periods rather than randomized groups, so the authors can report only an association between the AI model and better performance.

  4. Have you seen AI used this way where you study, work or live? Did it go the same way?

The questions are the same on every record. The details under them are taken from this record as written.

Sources

1 on file
  1. Annali Italiani di Chirurgiadoi:10.62713/aic.4849
    Peer-reviewed2026-11-10

The debate

More from Health