TRAI: An AI-Driven Mobile Application to Reduce the Gap Between Triage and Care

The integration of artificial intelligence (AI) into healthcare workflows offers transformative potential to streamline processes and enhance patient experience and outcomes. This paper presents a novel AI-powered chatbot that leverages a large language model (LLM) to conduct context-aware and adaptive conversations for patient triage and information gathering. The system integrates physician-guided prompts, triage notes, and patient history, including electronic health record (EHR) data, to dynamically refine clinical insights through iterative questioning. The chatbot ensures compliance with healthcare data standards while generating structured clinical summaries for physicians. Evaluations across ten gastroenterology/hepatology scenarios, reviewed by independent hepatologists, demonstrated high performance (scores 4.0-5.0/5.0) in documentation quality, assessment accuracy, patient education, and management planning. User ratings closely aligned with physician assessments, underscoring clinical relevance and user satisfaction. Through patient-facing Fast Healthcare Interoperability Resources (FHIR) integration with EHR data, the system reduces redundant history-taking and bridges critical gaps between triage and care. Such systems can streamline previsit workflows, enhance patient-provider communication, and improve clinical efficiency. This work highlights the viability of LLM-driven patient-facing chatbots to improve clinic efficiency, patient engagement, and diagnostic accuracy in healthcare workflows.