Accepted for/Published in: JMIR AI
Date Submitted: May 23, 2025
Date Accepted: Jan 29, 2026
Date Submitted to PubMed: Feb 4, 2026
Augmenting LLM with Prompt Engineering and Supervised Fine-Tuning in NSCLC TNM Staging: Framework Development and Validation
ABSTRACT
Background:
Non-small cell lung cancer (NSCLC) staging is crucial for treatment decisions and prognosis. Given the complex staging criteria, innovative methods are needed to assist physicians.
Objective:
This study aims to develop a robust solution for TNM staging using the GLM-4-Air model.
Methods:
A dataset of de-identified real-world medical imaging reports annotated by senior physicians from 3A-Level hospitals in China was constructed. Based on this dataset, a combined strategy of prompt optimization and Low-Rank Adaptation (LoRA) fine-tuning was developed to enhance the foundational model GLM-4-Air.
Results:
Results showed considerable improvement, with T, N, M, and clinical staging accuracies reaching 92%, 86%, 92%, and 90%, respectively. In comparison, GPT-4o achieved accuracies of 87%, 70%, 78%, and 80% under the same conditions.
Conclusions:
In this study, our model shows superior performance in NSCLC TNM staging. This underscored the effectiveness of our approach and highlighted its potential to enhance diagnostic standardization.
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.