MedGemma vs GPT-4：开源与闭源模型的零样本医学图像疾病分类对比
MedGemma vs GPT-4: Open-Source and Proprietary Zero-shot Medical Disease Classification from Images
AI 导读
一项研究对比了经 LoRA 微调的开源模型 MedGemma-4b-it 与未微调的专有模型 GPT-4 在六种疾病图像分类上的表现。MedGemma-4b-it 平均测试准确率达 80.37%，高于 GPT-4 的 69.56%，且在癌症、肺炎检测等高风险任务中灵敏度更高。结果表明领域特定微调对减少临床场景中的模型幻觉至关重要。
正文 · 原文
Abstract:Multimodal Large Language Models (LLMs) introduce an emerging paradigm for medical imaging by interpreting scans through the lens of extensive clinical knowledge, offering a transformative approach to disease classification. This study presents a critical comparison between two fundamentally different AI architectures: the specialized open-source agent MedGemma and the proprietary large multimodal model GPT-4 for diagnosing six different diseases. The MedGemma-4b-it model, fine-tuned using Low-Rank Adaptation (LoRA), demonstrated superior diagnostic capability by achieving a mean test accuracy of 80.37% compared to 69.56% for the untuned GPT-4. Furthermore, MedGemma exhibited notably higher sensitivity in high-stakes clinical tasks, such as cancer and pneumonia detection. Quantitative analysis via confusion matrices and classification reports provides comprehensive insights into model performance across all categories. These results emphasize that domain-specific fine-tuning is essential for minimizing hallucinations in clinical implementation, positioning MedGemma as a sophisticated tool for complex, evidence-based medical reasoning.
| Comments: | Accepted for publication in the Journal of Machine Learning and Deep Learning (JMLDL). 9 pages, 9 figures, 10 tables |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2512.23304 [cs.CV] |
| (or arXiv:2512.23304v2 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2512.23304 arXiv-issued DOI via DataCite |
|
| Journal reference: | Journal of Machine Learning and Deep Learning, Vol. 2, No. 2, pp. 1-9, 2025 |
| Related DOI: | https://doi.org/10.64820/AEPJMLDL.22.1.9.122025
DOI(s) linking to related resources |
Submission history
From: Md. Sazzadul Islam Prottasha [view email]
[v1]
Mon, 29 Dec 2025 08:48:36 UTC (892 KB)
[v2]
Tue, 27 Oct 2026 21:44:29 UTC (827 KB)
来源：arXiv cs.AI · arxiv.org