Problem Statement
Interpreting Doctor Notes using Deep Learning Techniques
A method for digitising handwritten prescriptions, doctor notes, and lab results that is also capable of tightly integrating with other healthcare systems to provide seamless digitization and data flow. To make digitization much easier, standardised forms can be made machine readable and supported in a variety of regional Indian languages.
PS Number: PSAIML006
Domain Bucket: Artificial Intelligence
Category: Software
Dataset : NA
The ability to express language, ideas, and thoughts through handwriting. It has become widely established over the years that doctors are notorious for having unreadable cursive handwriting. The datasets used in this study are examples of doctors’ cursive handwriting that were gathered from several hospitals and clinics in Taytay, Rizal, Quezon City, and Metro Manila. In this paper, we show the constructed Deep Convolutional Recurrent Neural Network Handwriting Recognition System.
Background of the Problem
Solution to digitise the handwritten prescriptions, doctor notes, lab reports, which can also help to integrate tightly with other healthcare systems for seamless digitization and data flow. Standardised forms can also be made machine readable with support for multiple local Indian languages to make digitization much simpler.
Objective
Handwriting is a skill to express thoughts, ideas, and language. Over the years, medical doctors have been well-known for having illegible cursive handwritings and has been a generally accepted matter. The datasets used in this paper are samples of doctors cursive handwriting collected from several clinics and hospitals of Metro Manila, Quezon City and Taytay, Rizal. In this paper, we present the Handwriting Recognition System using Deep Convolutional Recurrent Neural Network that is developed
Summary
The study successfully implemented the hybrid model in web and mobile application and based on the testing there are more correctly identified prescriptions than the misidentified. There are 389 images correctly identified out of the 540 input images. The testing through the mobile application yielded a 72% accuracy