Design and Early-Stage Development of a Handwritten Prescription Interpretation App for Bangladeshi Citizens

Abstract

In Bangladesh, handwritten prescriptions remain the conventional method of medical documentation, presenting significant challenges in accurately interpreting medication information. The diverse handwriting styles make these prescriptions difficult to read, potentially compromising patient health. This study introduces a novel solution for digitizing handwritten prescriptions using the User-Centered Design (UCD) methodology. The aim is to develop a mobile application tailored to the needs of Bangladeshi citizens, enhancing access to clear, reliable medication information, reducing errors, and promoting safer healthcare practices. The research objectives include conducting design requirements analysis, need-finding, and developing a high-fidelity prototype. A survey involving 155 participants was conducted to gather insights on the essential features and design requirements of the app. The findings revealed common difficulties faced by participants, such as the inability to read handwritten prescriptions accurately and challenges in understanding the names and dosage instructions of the medicines. Participants highlighted the need for features like daily medication reminders, medication names, automatic prescription scanning, voice and chat-based interactions with a Relational Agent, and user guidelines for prescription scanning. The insights gathered from the survey guided the creation of a low-fidelity paper-based prototype, which was then refined into a high-fidelity digital version of the app. The development process followed a human-centered approach, starting with need-finding analysis and design requirement collection, leading to the creation of both low-fidelity paper-based and high-fidelity digital prototypes. This high-fidelity prototype with an intuitive interface, named “PresCa”, incorporates crucial features such as prescription scanning, scanning guidelines, a report summary, a Relational Agent for assistance, prescription crosscheck by doctors, and the ability to configure medication reminders. These preferences directly informed the design choices and functional priorities of “PresCa”. By focusing on users’ needs, this app aims to empower patients with accurate medication information, reduce the risk of medication errors, and ultimately foster safer healthcare practices. This innovative solution not only addresses the challenges faced by healthcare professionals in Bangladesh but also provides a foundation for future research in healthcare digitization, promoting a more efficient and reliable system for managing prescriptions. The app achieved a mean System Usability Scale (SUS) score of 76.18. This research study illustrates methods for extracting medication information from handwritten prescriptions utilizing a straightforward, deep learning model called TrOCR: Transformer-Based Optical Character Recognition. This pre-trained model can detect complex handwritten language accurately by integrating the Optical Character Recogni- tion (OCR) technique with the Transformer architecture. Additionally, the integration of Machine Learning (ML) techniques in the app improves the accuracy and speed of prescription scanning, making it easier for users to access and understand their medica- tion instructions. The OCR performance of our trained model on the test set showed a Character Error Rate (CER) of 16% and a Word Error Rate (WER) of 36%. With the integration of the Levenshtein distance-based text correction technique, the CER and WER dropped significantly to 7% and 11%, indicating an improvement in the recogni- tion of medication information. Through these advancements, “PresCa” aims to make a significant impact on the healthcare sector in Bangladesh, facilitating better medication management and enhancing overall patient safety.