Summary of Enhancing Disease Detection in Radiology Reports Through Fine-tuning Lightweight Llm on Weak Labels, by Yishu Wei et al.
Enhancing disease detection in radiology reports through fine-tuning lightweight LLM on weak labels
by Yishu Wei, Xindi Wang, Hanley Ong, Yiliang Zhou, Adam Flanders, George Shih, Yifan Peng
First submitted to arxiv on: 25 Sep 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: None
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper investigates the application of large language models (LLMs) to the medical domain. The authors aim to overcome limitations, such as model size constraints and the lack of labeled datasets, by fine-tuning a lightweight LLM, Llama 3.1-8B, using synthetic labels. They combine instruction datasets for two tasks and train them jointly. Results show that when using high-quality synthetic labels, LLLA achieves satisfactory performance on disease detection (micro F1 score: 0.91). Conversely, when using low-quality synthetic labels from the MIMIC-CXR dataset, fine-tuned Llama 3.1-8B surpasses its noisy teacher labels (micro F1 score: 0.67 v.s. 0.63) after calibration against curated labels. The findings demonstrate the potential of fine-tuning LLMs with synthetic labels for medical applications. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper looks at how to use special language models to help doctors and hospitals. These language models can be trained on fake data, which makes it easier to use them in real-life situations. The researchers tested a smaller language model called Llama 3.1-8B with fake data from different sources. They found that when the fake data was good quality, the model worked well at detecting diseases (score: 0.91). When the fake data was poor quality, the model still improved its performance after being fine-tuned (score: 0.67 v.s. 0.63). This shows that using language models with synthetic data is a promising way to help doctors and hospitals in the future. |
Keywords
» Artificial intelligence » F1 score » Fine tuning » Language model » Llama » Synthetic data
