CareChain AI: An AI Powered Universal Health Record System

Abstract

Due to the fact that healthcare data is often spread out across many hospitals (as well as laboratories) and in many different formats (i.e., PDFs, images, handwritten prescriptions), it is often difficult to obtain urgent/critical patient information during emergencies. This fragmentation of data may cause delays in diagnosis and inefficient clinical decision making; additionally, patients are not as engaged in their own healthcare when the data they need to manage their own health is fragmented. Current digital health systems only focus on storing data, without providing the contextual intelligence and safety mechanisms that can assist users in understanding complex medical data. In response to these issues and challenges, this paper presents CareChain AI, an AI-based universal healthcare record management system that serves both to consolidate and intelligently process lifelong patient medical records in a secure (encrypted) and patient-centered manner. CareChain AI integrates Optical Character Recognition (OCR) functions to extract text from various documents, along with a Retrieval-Augmented Generation (RAG) framework to allow for natural language queries from a user-friendly interface. As text is extracted from documents, it is converted into vector embeddings and stored using FAISS indexes for highly efficient document retrieval methods.