The article discusses the development of both rule-based systems and Large Language Model (LLM)-based systems to effectively identify mentions of social support (SS) and social isolation (SI) categories in clinical notes. The study details the creation of a lexicon, annotations, and a system description that enhances transparency and classification accuracy. Despite initial challenges with certain machine learning models due to low data incidence, a semi-automated method utilizing fine-tuned LLM 'FLAN-T5' was successfully implemented, providing a promising avenue for clinical text analysis.
In developing our rule-based and LLM-based systems for identifying social support and isolation in clinical notes, we ensured full transparency and targeted classification.
Despite initial attempts with SVMs and BERT models to identify social support and isolation categories, they proved inadequate due to the limited mentions found in the clinical corpus.
Collection
[
|
...
]