ArticleSurgical and radiologic anatomy : SRA2025
System-based comparison of the knowledge level of popular AI chatbots on human anatomy: a multiple-choice exam analysis of GPT-4.1, DeepSeek, Co-Pilot, and Gemini models.
Article in Surgical and radiologic anatomy : SRA, 2025. The graph could read no effect estimate from its abstract, so it casts no vote on the map. Cited by 2 papers.
What it found
Each row is one number read from the abstract, on the scale the paper reported it, with its interval. Left of the dashed line favours the treatment, right favours the comparator. Under each row is the sentence it came from. New to these charts? A ten-minute tutorial.
The abstract states no effect estimate the extractor could read, or names no intervention and outcome on the map, so this paper lights no cell and moves no belief. It is still indexed, cited and linked below.
The trial behind it
Trials whose registry record cites this paper, or whose number appears in the abstract. A trial that started after this paper was published is citing it as background, not reporting it.
Neither the registry nor the abstract names a trial number. If this is a trial report, that itself is worth knowing.
Who cites it
2 citing papers in PubMed.
- Performance of Seven Large Language Models on Anatomy Examination Questions.Clinical anatomy (New York, N.Y.) · 2026Article
- Comprehensive Evaluation of Large Language Models on Four Core Medical School Courses: A Cross-Sectional Comparative Study.Advances in medical education and practice · 2026Article
Corrections and comments
PubMed lists nothing against this paper. Absence here is not a guarantee, only a check that was made.
Authors and funding
3 authors.
Funding
No grant is acknowledged in the PubMed record.
Abstract
purposeThis study aimed to comparatively assess the knowledge level of AI-based chatbots on human anatomy systems using multiple-choice questions, and to analyze their potential contribution to medical education.
methodsSeventy multiple-choice questions covering seven major anatomical systems (musculoskeletal, respiratory, circulatory, digestive, urinary, genital, and nervous systems) were translated in accordance with Terminologia Anatomica and presented to GPT-4.1, DeepSeek, Co-Pilot, and Gemini. Questions were selected from first- and second-year medical student exams and distributed based on item difficulty index (Pj). All bots were tested under the same conditions to minimize bias. Success rates and statistical differences were evaluated using Kruskal–Wallis and Cochran’s Q tests. The relationship with item difficulty was assessed using point biserial correlation.
resultsGPT-4.1 showed the highest accuracy (95.7%), followed by Co-Pilot (94.3%), DeepSeek (92.9%), and Gemini (91.4%). System-based results showed Co-Pilot reached 100% in musculoskeletal, and GPT-4.1 reached 90% in nervous system questions. All bots scored 100% in respiratory and circulatory systems. In other systems, success rates ranged from 80 to 100%. No significant correlation was found between item difficulty and chatbot accuracy.
conclusionChatbots achieved high accuracy in anatomy questions, but there were notable differences across systems and areas. While their supportive role in medical education is increasing, expert supervision is still recommended. These results show that AI-based systems can serve as complementary educational tools, but further improvement is needed for complete reliability.
Indexed as
Identifiers
41233619What OpenQuestion holds
Registered trials
Read under generation 80e0d062 · epoch 390. Bibliography from PubMed, PubMed Central and OpenAlex; grants from NIH RePORTER; trial links from ClinicalTrials.gov; estimates, votes and beliefs from the OpenQuestion graph.