Development of OCR Technology Application System for Health Data Recording

Abstract

The shift to digital health records requires advanced technologies to transform medical device data into digital formats. This study created a way to digitize health data from devices that measure blood sugar, blood pressure, and pulse oximeters. It used YOLOv5 to find objects and optical character recognition (OCR) technologies to read text. The solution incorporates a MySQL database for effective data storage and a web application for intuitive data presentation. YOLOv5 was trained on 6,630 photos to effectively detect and evaluate seven-segment displays. A YOLOv5 confidence level of 80.75% and an OCR accuracy of 93.2% were found when testing at different distances (7 to 30 cm) and angles (0º, −35º, −30º left, −30º right) and with different lighting conditions. Well-lit settings yielded optimal performance; however, extreme tilts occasionally led to misreading’s. The technology processed photographs in 10 to 15 seconds, facilitating real-time data conversion and enhancing usability for senior individuals handling daily health information. Even though there were challenges, such as low light and differences between devices, the system showed that it could cut down on mistakes and make healthcare more efficient. Future enhancements will concentrate on sophisticated preparation methods and mistake correction algorithms to guarantee uniform performance. This system provides a strong and scalable solution for digitizing health data, facilitating enhanced electronic health records (EHRs) and individualized healthcare management.