MediScan AI: An Intelligent Vision-Based Platform for Prescription Decoding and Comprehensive Medicine Information Delivery

Abstract: In the modern healthcare ecosystem, deciphering handwritten prescriptions remains a critical patient safety challenge, particularly in developing nations and small clinical settings. Studies indicate that up to 61% of prescription errors are attributed to illegible handwriting, leading to severe health and economic consequences.To address this issue, this paper presents MediScan AI, a cloud-deployed web application that utilizes advanced multimodal artificial intelligence to analyze prescription images and deliver patient-friendly medicine information. The system employs a hybrid AI-database architecture, integrating Google Gemini Vision AI for accurate text extraction with a curated local database for validated medical data.A key innovation of the platform is the mandatory extraction and presentation of side effects and overdose risks for all identified medications. Experimental results demonstrate high accuracy across both printed and handwritten prescriptions, with an average processing time under 15 seconds. MediScan AI effectively bridges the gap between complex medical prescriptions and patient understanding, enhancing medication safety and accessibility.

Keywords: Artificial Intelligence, Prescription Decoding, OCR, Google Gemini Vision, Medication Safety, Hybrid Architecture