Related Experiment Videos
Large Language Model Versus Multidisciplinary Team: A Feasibility Study of Pancreatic Cancer Management
Zhuoran Liu1, Xin Zhao2, Tianyang Mao1
1People's Hospital of Leshan, No.238 Huian street, Shizhong District, Leshan, Leshan, CN.
Journal of Medical Internet Research
|May 26, 2026
Summary
ChatGPT-5.2 accurately categorized pancreatic cancer management in 95.2% of multidisciplinary team (MDT) cases. Further research is needed due to the model
Area of Science:
- Oncology
- Artificial Intelligence in Medicine
- Medical Informatics
Background:
- Pancreatic cancer treatment requires complex multidisciplinary team (MDT) input.
- Evaluating the utility of AI in streamlining MDT decision-making is crucial.
Purpose of the Study:
- To assess the accuracy of a chat-interface AI model (ChatGPT-5.2) in classifying pancreatic cancer management categories.
- To determine the feasibility of using AI for initial MDT case categorization.
Main Methods:
- Analysis of 125 complete structured pancreatic cancer cases from 269 MDT cases.
- Comparison of AI-generated management categories with established MDT classifications.
Main Results:
- The AI model correctly matched the MDT broad initial management category in 119 out of 125 cases (95.2%).
- High concordance observed between AI and MDT initial management categorization.
Conclusions:
- The AI model demonstrates high accuracy in initial pancreatic cancer MDT case categorization.
- Findings suggest potential for supervised exploratory use of AI in oncology MDTs.
- Further validation and version control are recommended for clinical application.
