Building Structured Personal Health Records from Photographs of Printed Medical Records

Area of Science:

• Health Informatics
• Computer Vision
• Document Analysis

Background:

• Personal health records (PHRs) promote patient-centered care by enabling patient access to health information.
• Accessing electronic health records is challenging in China, where patients typically receive printed medical documents.

Purpose of the Study:

• To develop a practical method for extracting structured data from images of printed medical records.
• To facilitate the creation of personal health records (PHRs) from accessible paper documents.

Main Methods:

• An optical character recognition (OCR) pipeline was employed for text recognition from document photos.
• Image pre-processing and multi-engine OCR synthesis were used to handle low image quality and complex content.
• Annotation algorithms were developed to identify document types, entities, and their correlations for structured PHR creation.

Main Results:

• The proposed approach successfully extracted structured data from real-world printed medical records.
• The system demonstrated effectiveness and applicability in digitizing and structuring patient medical information.

Conclusions:

• The developed method provides a viable solution for building personal health records (PHRs) from printed medical documents captured via mobile phones.
• This approach can improve patient engagement and healthcare management by making medical information more accessible.