SoftTiger is a groundbreaking clinical large language model designed to structure clinical notes into valuable data for healthcare workflows. It addresses the challenge of unstructured clinical narratives hindering intelligentization in healthcare. By collecting and annotating data for critical subtasks like international patient summary, clinical impression, and medical encounter, SoftTiger fine-tunes a state-of-the-art LLM using public and credentialed clinical data. The model excels in supporting basic tasks such as abbreviation expansion and temporal information extraction before progressing to more complex clinical tasks like impression and encounter summary. SoftTiger outperforms popular open-source models, GPT-3.5, and is comparable to Gemini-pro, showcasing its potential in healthcare digitalization. The release of SoftTiger models at scales of 13 billion and 70 billion parameters, along with datasets and code, aims to make a significant contribution to the healthcare industry.
To Another Language
from source content
arxiv.org
Key Insights Distilled From
by Ye Chen,Igor... at arxiv.org 03-05-2024
https://arxiv.org/pdf/2403.00868.pdfDeeper Inquiries