Convolutional Neural Networks for Handwritten Text Recognition of Medical Prescription

Abstract

Converting handwritten prescriptions into electronic format offers several advantages and is crucial for modern healthcare systems. It is essential nowadays because of some factors such as – Legibility and Accuracy: Handwritten prescriptions can be challenging to read and interpret; accessibility and Portability: Electronic prescriptions can be easily stored; Decision Support Systems: By digitising prescriptions, healthcare systems can integrate them with electronic health records (EHRs) and utilise decision support systems. Convolutional neural networks (CNNs) are a class of deep learning algorithms that have proven effective in extracting handwritten text from various documents, including medical prescriptions. By leveraging CNNs for handwritten text extraction, healthcare systems can automate the process of digitising prescriptions, reducing manual effort and potential human errors. This enables seamless integration with electronic systems, facilitating better patient care and overall healthcare management. In this paper, we have trained the CNN model for different parameters and observed the accuracy and loss for various parameters. We got a maximum training accuracy of 89% and a maximum testing accuracy of 70%.