Pedi-Medi-Instructor: Pediatric Medicine Information and Instruction System

Abstract

This study proposes a two-part system to enhance medication safety and user experience, especially for pediatric patients. The Data Acquisition System for Medicine Identification (DAMI) trains an AI model with comprehensive images of medications captured from various angles. The Medicine Identification and Instruction System (MIIS) leverages this model to identify medications and prescriptions, even with imperfect placement, and delivers audio instructions on usage. Designing the structured prescription provides highly accurate medicine text extraction, medicine type classification, and proper consumption unit extraction. While currently limited to pediatrics and challenged by handwriting variations, the system's robustness to user errors, digital twin monitoring, and planned improvements for broader use and recognition accuracy position it as a promising tool for medication management.