The article introduces two open-source Natural Language Processing (NLP) systems aimed at detecting social support (SS) and social isolation (SI) within clinical notes of psychiatric patients. Each system employs distinct methodologies—a rule-based approach and a large language model (LLM) approach—both showcasing their respective advantages and limitations in this complex task. Comprehensive annotation guidelines and lexicons for SS and SI are made available to facilitate reproducibility and further exploration of the topic. The research underscores a collaborative effort supported by various NIH grants, emphasizing data transparency and community resource sharing.
The article presents two open-source NLP systems for identifying social support and social isolation in psychiatric patients, highlighting the unique strengths and limitations of a rule-based versus a LLM approach.
Findings indicate that both the rule-based and LLM systems demonstrate effective performance in categorizing social support and social isolation, proving beneficial in clinical note analysis.
Comprehensive annotation guidelines and lexicons for social support and social isolation are provided in the supplementary materials, aiding in reproducibility and enabling further research.
This research received funding from the National Institutes of Health and ensures the availability of all code and resources on GitHub for future use by the scientific community.
Collection
[
|
...
]