JMIR Publications

ABSTRACT

Background:

Stroke has a major impact on global health, causing long-term disability and straining healthcare resources. Generative large language models (gLLMs) have emerged as promising tools to help address these challenges, but their applications and reported performance in stroke care require comprehensive mapping and synthesis.

Objective:

The aim of this scoping review was to consolidate a fragmented evidence base and to examine the current landscape, shortcomings, and future directions in the design, reporting, and evaluation of gLLM-based interventions in stroke care.

Methods:

This scoping review, adhering to Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) guidelines and the Population-Concept-Context (PCC) framework, searched six major scientific databases in December 2024 to evaluate gLLM interventions across the stroke care pathway mapping their key characteristics and outcomes.

Results:

Retrospective designs predominated. Key applications of gLLMs included clinical decision-making support (40%), administrative assistance (36%), direct patient interaction (20%), and automated literature review (4%). Implementations mainly employed Generative Pretrained Transformer (GPT) models accessed through task-prompted chat interfaces. Five key challenges were identified from the included studies during the implementation of gLLM-based interventions: ensuring factual alignment, maintaining system robustness, enhancing interpretability, optimizing efficiency, and facilitating clinical adoption.

Conclusions:

The application of gLLMs in stroke care, while promising, remains relatively new, with most interventions reflecting early-stage or relatively simple implementations. Against this backdrop, critical gaps in research and clinical translation persist. To support the development of clinically impactful and trustworthy applications, we propose an actionable framework that prioritizes real-world evidence, mandates transparent technical reporting, broadens evaluation beyond output accuracy, strengthens validation of advanced task adaptation strategies, and investigates mechanisms for safe and effective human–gLLM interaction. Clinical Trial: The review protocol was pre-registered on the Open Science Framework and is available at: https://doi.org/10.17605/OSF.IO/J36WV.