Applications of Natural Language Processing and Large Language Models for Social Determinants of Health: Protocol for...

Area of Science:

• Public Health
• Computational Linguistics
• Health Informatics

Background:

• Natural Language Processing (NLP) offers innovative methods for analyzing social determinants of health (SDOH) in textual data.
• Existing literature on NLP for SDOH is fragmented across disciplines, necessitating a consolidated overview.
• Identifying knowledge gaps and future research directions is crucial for advancing this field.

Purpose of the Study:

• To systematically review and highlight NLP techniques, including large language models, applied in SDOH research.
• To consolidate existing knowledge on NLP applications for SDOH.
• To inform future research by identifying gaps and proposing research questions.

Main Methods:

• A comprehensive search strategy was executed across seven major databases (PubMed, Web of Science, IEEE Xplore, Scopus, PsycINFO, HealthSource: Academic Nursing, ACL Anthology).
• Studies published in English between 2014 and 2024 were included.
• Independent screening by three reviewers, with conflict resolution by two additional reviewers, and a forward search of cited studies were employed.

Main Results:

• The search strategy was executed in August 2024 across the specified databases.
• Title and abstract screening was underway as of December 2024.
• Full results are anticipated for peer-review publication in early 2025.

Conclusions:

• This systematic review will provide a comprehensive analysis of NLP applications for SDOH tasks across diverse textual datasets.
• The evaluation of methodologies, tools, and outcomes will pinpoint current knowledge gaps.
• Findings will guide future research, aiming to enhance NLP models for SDOH and improve public health outcomes.