Ragil Yulianto

Smart Prescription Reader: Enhancing Accuracy in Medical Prescriptions

Introduction

Smart prescription reader: enhancing accuracy in medical prescriptions . Enhance medical prescription accuracy with a Smart Prescription Reader. This study uses CNN & MobileNet to convert doctor's handwritten scripts from a BD dataset into readable text via a web app.

Abstract

Reading a doctor's handwritten prescription is a challenge faced by most patients and some pharmacists, which in some cases can lead to negative consequences due to misinterpretation of the prescription. The "Doctor's Handwritten Prescription BD Dataset" on Kaggle contains segmented images of handwritten prescription words from BD (Bangladesh) doctors. This dataset, intended for machine learning applications, includes 4,680 individual words segmented from prescription images. This study introduces a Handwriting Recognition System using Convolutional Neural Network (CNN) developed to identify text in prescription images written by doctors and convert the cursive handwriting into readable text. Two models were evaluated in this study: CNN and MobileNet. Based on the experiments, MobileNet showed better results compared to CNN alone. From the dataset of 4,680 words, 3,120 were used for training, 780 for testing, and 780 for validation. The study achieved a training accuracy of 97%, a testing accuracy of 88%, and a validation accuracy of 83%. The developed model was successfully implemented in a web application

Review

This study addresses a critical and pervasive issue within healthcare: the misinterpretation of handwritten medical prescriptions, which can have serious negative consequences for patient safety. The authors propose a "Smart Prescription Reader" designed to enhance accuracy by converting doctors' cursive handwriting into legible text. Utilizing the "Doctor's Handwritten Prescription BD Dataset," a valuable resource of segmented words from Bangladeshi doctors, the research aims to develop a robust handwriting recognition system. The clear articulation of the problem and the immediate relevance of a technological solution make this a timely and potentially impactful contribution to improving healthcare delivery and patient outcomes. The methodology involves developing a Convolutional Neural Network (CNN)-based handwriting recognition system, with two primary models, CNN and MobileNet, evaluated for performance. The abstract effectively details the dataset's composition (4,680 individual words) and its strategic partitioning into training (3,120), testing (780), and validation (780) sets. A key finding is MobileNet's superior performance compared to a standalone CNN, although specific metrics beyond overall accuracy are not provided in the abstract. The reported accuracies—97% for training, 88% for testing, and 83% for validation—demonstrate a promising level of performance for the MobileNet model, especially given the inherent challenges of diverse handwriting styles. The successful implementation of the developed model into a web application further underscores the practical readiness of this research. Overall, this work presents a promising step towards mitigating a significant challenge in medical practice. The use of a specialized dataset of BD doctors' prescriptions is a strength, ensuring relevance to a specific context, though it also raises questions about generalizability to other regions. While the abstract effectively highlights the system's development and performance, a full paper would benefit from deeper analysis of misclassification types, a comparative discussion of MobileNet's specific advantages, and a more detailed exploration of the web application's features and user experience. Nevertheless, the study offers a valuable contribution to the field of medical informatics and machine learning applications in healthcare, with clear potential for real-world impact on prescription accuracy and patient safety.

Full Text

Comments

You need to be logged in to post a comment.