OCR + Generative AI Powered Prescription to Explanation System

Abstract

Healthcare systems produce and manage lots of highly sensitive information, especially in the form of prescriptions and other documents related to patients' private data. However, due to the nature of handwritten medical prescriptions and medical jargon, patients often find it difficult to follow medication instructions. Optical Character Recognition (OCR) is still in its nascent phase of digitizing handwritten prescription documents into text. Furthermore, the traditional OCR systems are noisy and difficult to understand when it comes to recognizing the handwriting of doctors to understand medical prescriptions properly. This paper utilizes Generative AI to test the ability of LLMs in conducting OCR on handwritten text and analyze its performance, and merge OCR with Generative AI to analyze and extract medicines from prescriptions in order to give a detailed and specialized description of the medicine to the user

The paper further analyzes some OCR methods for text extraction from medical prescriptions and compares their performance. It has also been demonstrated that the usability of prescription data and the possibility of misinterpretation of medication is decreased when OCR is combined with Generative and Agentic AI. Thus the combination of artificial intelligence and OCR will help improve the understanding of patients and the accessibility of digital healthcare.