AI Medical Document Generation

Author

Maniteja Parchuri, Southern Illinois University CarbondaleFollow

Date of Award

5-1-2025

Degree Name

Master of Science

Department

Computer Science

First Advisor

Henry, Hexmoor

Abstract

Clinical documentation is an essential component of modern healthcare, yet it remains a time-consuming and manual process that often contributes to clinician burnout and delays in patient care. This project introduces the AI-Based Medical Document Generation System, a comprehensive solution that automates the generation of structured clinical reports using advanced techniques in Natural Language Processing (NLP), Machine Learning (ML), and Retrieval-Augmented Generation (RAG). The system accepts clinical input from doctors and nurses in both text and voice formats. Voice input is transcribed using the Web Speech API, and the resulting text undergoes preprocessing before being summarized using models such as T5, BART, and GPT-2. These summaries are then used to retrieve semantically similar historical patient cases through TF-IDF and FAISS, which provide relevant clinical context for report generation. The summarization and retrieved data are combined and sent to ChatGPT via the OpenAI API to produce a professional, structured medical report. Additionally, the system includes a Naive Bayes–based disease prediction module, which provides real-time diagnostic support and medication suggestions. The platform is developed using React for the frontend and FastAPI for the backend, ensuring performance, modularity, and ease of deployment. Evaluation results show high summarization accuracy (ROUGE, BLEU), strong disease prediction performance, and positive feedback from user testing. This project demonstrates a practical, scalable approach to automating clinical documentation, reducing administrative workload, and enhancing the quality and efficiency of patient care.