Gordana Petrovska Dojchinovska, Monika Simjanoska Misheva and Kostadin Mishev

ABSTRACT
Integrating advanced open-source large language models, such as
Llama or GatorTron, into the healthcare domain presents a novel
approach to enhancing communication between physicians and
patients. This review paper examines the potential of Llama and
GatorTron to improve patient-provider interactions, focusing on the
models’ ability to process and generate human-like language in real-
time clinical settings. Any large language model (LLM) used in a
clinical setting, especially one that is used to improve patient-doctor
communication, needs to undergo cultural sensitivity training, as
the datasets these models are trained on seem to be diverse and they
include specificities, both linguistic and medical, that may be valid
to one population but completely exclude another. Furthermore,
the model needs to be capable of providing accurate responses that
are context-sensitive and that also align with clinical guidelines
and patient needs. Other things to take into consideration when
fine-tuning LLMs to local medical customs are to adapt them to dif-
ferent geographies and the underlying linguistic demands, to help
them employ ethical and responsible AI practices so that no biases
or stereotypes are perpetuated and that, at the same time, help to
protect patient privacy and data security. Finally, feedback provided
by both the physicians and the patients needs to be incorporated to
further refine any model that is to be used in a medical setting. Over
time, this will help the model to become more nuanced. Common
communication barriers, such as medical jargon, cultural differ-
ences, and patient literacy, which often hinder effective dialogue
are also considered in this review paper, as well as how fine-tuned
LLMs address those issues. By synthesizing current research and
practical applications, this paper aims to provide a comprehensive
understanding of the potential of fine-tuned large language mod-
els to transform healthcare communication, ultimately improving
patient outcomes and satisfaction.