MedInsight: A Multi-Source Context Augmentation Framework for Generating Patient-Centric Medical Responses using Large Language Models

TL;DR

MedInsight is a novel retrieval augmented framework that augments LLM inputs (prompts) with relevant background information from multiple sources that generates enriched, patient-specific responses tailored for healthcare applications such as diagnosis, treatment recommendations, or patient education.

Abstract

Large Language Models (LLMs) have shown impressive capabilities in generating human-like responses. However, their lack of domain-specific knowledge limits their applicability in healthcare settings, where contextual and comprehensive responses are vital. To address this challenge and enable the generation of patient-centric responses that are contextually relevant and comprehensive, we propose MedInsight:a novel retrieval augmented framework that augments LLM inputs (prompts) with relevant background information from multiple sources. MedInsight extracts pertinent details from the patient's medical record or consultation transcript. It then integrates information from authoritative medical textbooks and curated web resources based on the patient's health history and condition. By constructing an augmented context combining the patient's record with relevant medical knowledge, MedInsight generates enriched, patient-specific responses tailored for healthcare applications such as diagnosis, treatment recommendations, or patient education. Experiments on the MTSamples dataset validate MedInsight's effectiveness in generating contextually appropriate medical responses. Quantitative evaluation using the Ragas metric and TruLens for answer similarity and answer correctness demonstrates the model's efficacy. Furthermore, human evaluation studies involving Subject Matter Expert (SMEs) confirm MedInsight's utility, with moderate inter-rater agreement on the relevance and correctness of the generated responses.

Figures (8)

• Figure 1: MedInsight's context augmentation approach for generating patient-centric responses. Patient context from medical transcripts and prompt is augmented with relevant medical knowledge from authoritative sources into a comprehensive context input to the language model, enabling personalized patient-centric response generation.
• Figure 2: The illustration emphasizes response generation using LLMs with and without external knowledge. FIg. \ref{['fig:simple_llm']} shows response generation without external knowledge, Fig. \ref{['fig:simple_rag']} depicts a simple RAG that utilizes an single external source whereas Fig. \ref{['fig:our_rag']} depicts our approach where we combine multi source context to generate personalized response.
• Figure 3: Detailed architecture of MedInsight Framework
• Figure 4: Illustration the transformation of the doctor-patient interaction into the patient's specific context: A represents the interaction between the doctor and patient, B denotes an unstructured medical transcript, and C indicates the annotated structured representation of the patient's unique context.
• Figure 5: An illustrative example of retrieving medical knowledge that is tailored to a patient's unique context when responding to a medical query.