Back to the newsroom

Press Release

Korea Deep Learning completes OCR for automatic digitization of medical records with Kyung Hee University Hospital at Gangdong

Korea Deep LearningPublished

Korea Deep Learning completes OCR for automatic digitization of medical records with Kyung Hee University Hospital at Gangdong
In November 2023, Korea Deep Learning and Kyung Hee University Hospital at Gangdong finished developing an OCR system that reads, item by item, medical records whose formats differ from hospital to hospital.

In November 2023, Korea Deep Learning and Kyung Hee University Hospital at Gangdong completed the development of an Intelligent Document Understanding OCR-Based Automatic Medical Record Digitization System. It is optical character recognition (OCR) specialized for medicine, in which AI works out for itself how to read medical records whose formats differ from hospital to hospital, and reads and stores them item by item. The system was a project under the AI Voucher Support Program, carried out by a consortium Korea Deep Learning formed with Kyung Hee University Hospital at Gangdong.

Medical records whose formats differ from hospital to hospital

When patients move to a new hospital, they bring the medical records issued by their previous hospital. But the format of medical records varies from hospital to hospital, so at the new hospital, medical staff and employees had to read the documents one by one and copy what they needed into the system. The more directly the information is used in treatment, such as prescriptions, diagnoses and test results, the more it matters to copy it accurately, but when formats differ, the same item appears in a different place and with different wording in each document.

What Kyung Hee University Hospital at Gangdong needed was not a rule-based system fitted to fixed formats but OCR that looks at a document and decides for itself.

DEEP OCR that works out the format and reads item by item

Korea Deep Learning specialized its own DEEP OCR algorithm for medical records. In the system, AI works out for itself the format of each hospital's medical records, and reads and stores the information item by item. In prescriptions, it separates drug names, doses and directions for use into distinct items, and in medical records, it structurally separates diagnoses, findings and test results.

The company also built safeguards to reduce errors in the recognition results. Korea Deep Learning developed an automated data post-processing system that minimizes recognition errors, cutting the parts that people have to review and correct again. It also applied training fitted to medical document data and a design fitted to the hospital's environment and computer systems.

Quickly understanding and automatically storing records from other hospitals

As a result, Kyung Hee University Hospital at Gangdong can now quickly understand and automatically store the contents of medical records that newly arriving patients bring from other hospitals. Structured patient records can become the basis for later uses such as medical data analysis and diagnostic support.

Medical OCR from three straight years in the AI Voucher program

The project is a result of the 2023 AI Voucher Support Program, for which Korea Deep Learning was selected as a provider for the third year in a row in September 2023. The AI Voucher Support Program is a government program that gives demand companies up to KRW 300 million to build a supplier's AI solution fitted to their own work. In 2022, Korea Deep Learning also carried out a project digitizing handwritten waybills with Sunghwa Corporate Express and was named an outstanding supplier. Technology that reads handwriting and inconsistent formats has, in effect, moved from logistics to medicine.

With the results of this research, Korea Deep Learning plans to show the potential of an intelligent OCR service that recognizes tables and items.

About DEEP OCR

Korea Deep Learning started out by reading documents. Built on KDL's own document-specialized VLM, DEEP OCR is Korea's first VLM-based OCR solution, reading not just the characters but the structure and meaning of a document together. It handles unstructured documents in every kind of layout without building separate training data, and extracts text and key-value data from visual elements such as seals, signatures, tables, numbers and symbols. From loan documents in finance to administrative documents in the public sector and handwritten waybills in logistics, documents people used to copy out by hand become business data through DEEP OCR. More information is available on the DEEP OCR page at koreadeep.com.

About Korea Deep Learning

Korea Deep Learning (KDL), founded in 2019, is an enterprise AI agent company. Built on KDL Frontier, its own vision-language model (VLM) specialized for documents, KDL makes AI agents that take on and complete the document work of businesses and public institutions. Its main products are DEEP OCR, DEEP Parser, DEEP Agent and DEEP Ops.

Source

  • Korea Deep Learning blog 2026-01-14