A Distributed Cognition Approach to Agentic Self-Medication System in Low-Resource Healthcare Environments

Self-medication is common in regions with limited access to healthcare, yet it poses significant risks when combined with low health literacy and the circulation of counterfeit drugs. Existing digital self-medication tools often rely on static databases and rigid input formats, limiting their ability to interpret natural-language symptom descriptions and provide context-aware guidance. This study proposes a distributed cognition–based multi-agent system that integrates large language models (LLMs) with specialized AI agents to support self-medication decision making. The system processes symptom narratives, infers potential conditions, retrieves pharmaceutical knowledge from telemedicine sources, and generates structured medication guidance, including dosage instructions and potential side effects. The architecture distributes reasoning tasks across agents responsible for input classification, symptom analysis, pharmaceutical recommendation, and response synthesis. Evaluation using the RAGAS framework demonstrates high context recall and strong response relevance, while expert validation confirms the clinical appropriateness of the generated recommendations. This study contributes to the information systems literature by demonstrating how distributed cognition and multi-agent architectures can enhance AI-assisted healthcare decision support in resource-constrained environments.