Intelligent Lab Report Analysis: Using MultiAgent Langgraph Framework

Abstract

Medical laboratory reports are essential for clinical decision-making, yet their technical terminology and numerical complexity render them largely inaccessible to patients without medical training. This paper presents an AI-Powered Medical Lab Report Explanation Chatbot that automatically extracts, parses, and explains laboratory test results in plain, patient-friendly language. The system employs a two-layer architecture combining deterministic document processing—PDF text extraction via PyMuPDF with Tesseract OCR fallback, multi-strategy regex parsing, and reference-range rule evaluation—with AI-driven reasoning powered by the Llama 3.3 70B large language model accessed through the Groq API. LangGraph orchestrates the processing pipeline through two compiled state graphs: a report-processing graph for extraction and rule application, and a chat graph for interactive question answering. A Streamlit web interface delivers real-time structured summaries including KPI cards for normal, abnormal, and critical findings, personalised dietary guidance, and an interactive conversational assistant. The system operates entirely in-memory with no persistent data storage, ensuring patient privacy by design. Evaluation on diverse lab report formats demonstrates robust extraction accuracy and clinically grounded, non-diagnostic explanations suitable for patient education.