PERSiMED: PERSian Intelligent MEDical Document Reader:A Smart NLP-based Framework for Persian Medical Documents

The healthcare industry generates vast amounts of textual data in the form of medical records, prescriptions, and insurance claims. This paper introduces PERSiMED (PERSian Intelligent MEDical document reader), a novel deep learning-based framework designed to automate the extraction, processing, and semantic analysis of Persian medical documents. The proposed system employs a sophisticated pipeline combining Convolutional Neural Networks (CNN) for document preprocessing and Vision-Language (VL) models for intelligent content extraction. PERSiMED addresses critical pain points in the health insurance industry, including fraud detection, claims processing automation, privacy-preserving data management, and longitudinal health record creation. Our framework achieves end-to-end automation from image input through auto-rotation, orientation correction, boundary detection, to semantic annotation and structured HTML output, enabling seamless database integration. This research demonstrates how modern AI techniques can be effectively adapted to handle the unique challenges of Mixed English and Persian script medical documentation.

کلیدواژه ها:

Medical Document Analysis ، Health Insurance Automation ، Fraud Detection ، Vision-Language Models ، Healthcare NL