Enhancing Medical Readiness with LLMs: A Low-Resource OTC Support Bot for Deployed Units
James Paul Tan
a,b
,
Margrette Yebes
a,c
,
Michael Ralph Estrada
a,c,d
,
Paula Marielle S. Ababao
e,f
,
Gabriel Avelino Sampedro
a
a Convergent Technologies Research Laboratory, Manila, Philippines
b College of Engineering, Samar State University, Samar, Philippines
c Gokongwei College of Engineering, De La Salle University, Manila, Philippines
d Computational Science Research Center, University of the Philippines Diliman, Quezon City, Philippines
e Innovation and Research Office, FEU Institute of Technology, Manila, Philippines
f Electrical Engineering Department, FEU Institute of Technology, Manila, Philippines
2025 International Conference on Mobile, Military, Maritime IT Convergence (ICMIC), (2025), pp. 229-232
Abstract: In military and expeditionary (maritime) health care environments where isolation, safety, limited personnel, and resource constraints can threaten the delivery of frontline health care, access to timely and knowledgeable medical assistance can be extremely valuable. The purpose of this paper is to investigate the use of large language models (LLMs) in military and maritime health care environments by creating an AI-powered, Over-the-Counter (OTC) Medication Assistance Bot using the Mistral-7B model. The bot is intended to be deployed within tactical, or even shipboard systems, and it would empower autonomous, in-the-moment recommendations of medications for specific symptoms, while mitigating the risks associated with deploying personnel self-medicating through potential non-fundamental use errors. For this work, we employed Low-Rank Adaptation (LoRA) to fine-tune the system, and the bot was trained on a specific dataset derived from material on pharmacological sources, contextualized for medical practices in the Philippines. Based on evaluation the model achieved an average F1-score of 0.7296, which is above the 0.60-0.70 expected levels of performance for medical dialogue systems. The research shows promise for the model as it enhances combat and maritime healthcare readiness by providing consistent, low-bandwidth, and local medical assistance when connected medical supervision may not be immediately available.