Enhancing Handwritten Prescription Recognition with AI-Driven OCR

Authors

  • Haseeb Ullah Department of Computer Systems Engineering, University of Engineering and Technology, Peshawar, 25000, Pakistan.
  • Muhammad Tanveer Department of Computer Science, Preston University, Islamabad Campus, Islamabad, 44000, Pakistan.
  • Atif Jan Department of Electrical Engineering, University of Engineering and Technology, Peshawar, 25000, Pakistan.

Keywords:

Deep Learning, TrOCR, Medical Prescription, Handwriting Recognition, Roboflow

Abstract

Accurate interpretation and understanding of medical prescriptions are crucial for healthcare providers to ensure suitable treatment for patients. However, the increasing number of prescriptions and the complexity of pharmaceutical regimens may lead to errors, which could have severe consequences. To overcome this problem, artificial intelligence (AI) can automate tasks such as identifying the correct medication, determining the correct dose, and checking for drug interactions. This makes prescription analysis more accurate and faster. This study presents an AI-driven optical character recognition (OCR) framework that uses TrOCR with Roboflow to convert handwritten prescriptions into a digital format. Our method achieves a Word Error Rate (WER) of 12.5%, a Character Error Rate (CER) of 8.7%, and an Exact Match Accuracy of 81.3%. These results show that the system can accurately transcribe prescriptions and help reduce medication errors, making healthcare workflows safer and more efficient.

Downloads

Published

2025-08-24

How to Cite

Haseeb Ullah, Muhammad Tanveer, & Atif Jan. (2025). Enhancing Handwritten Prescription Recognition with AI-Driven OCR. Journal of Computing & Biomedical Informatics. Retrieved from https://www.jcbi.org/index.php/Main/article/view/1054

Issue

Section

Articles