•  
  •  
 

MEDICAL AI CHATBOT

Abstract

Long wait times in clinics delay patient care and reduce efficiency. To address this issue, I am developing an AI chatbot designed to streamline the patient intake process. The chatbot allows patients in waiting rooms to enter their symptoms and responds with progressively more refined questions. This process makes gathering sensitive information easier, since patients may feel more comfortable disclosing details to a chatbot. Before the appointment begins, the physician can review a chatbot-generated summary of the patient’s responses, enabling more focused and efficient care. The prototype is built using Streamlit, an open-source Python framework that makes it simple to create interactive web applications. It operates through a Retrieval-Augmented Generation (RAG) pipeline, which retrieves relevant information from a predefined document set and combines it with a large language model to produce accurate and consistent answers. Unlike unrestricted chatbots, this design ensures reliability and allows medical practices to upload their own materials for domain-specific customization. The interface is minimalist and user-friendly, with a sidebar for easy navigation and conversation logging to track persistent symptoms across visits. Currently in development, the chatbot can process PDF documents, create a local vector database, accept user questions, fetch relevant knowledge from the database, and generate context-based responses. While still in its prototype stage, these steps demonstrate the system’s ability to handle structured medical information reliably. Although not intended as a diagnostic tool, this chatbot has strong potential to improve patient–provider communication, reduce waiting times, and adapt across healthcare settings, including mobile and emergency applications.

This document is currently not available here.

Share

COinS