More Than 5 Million HIRA Reduction Cases Used for AI Training

Artificial intelligence (AI) technology will be introduced to address the issue of reduced reimbursement claims at medical institutions.

SumMeditech

SumMeditech

View original image

SumMeditech, a company specializing in medical claim review, and MsysTech, a medical IT specialist, announced on January 22 that they plan to apply AI to their pre-claim review solutions, "ROI" and "ROILite," during the second half of this year. The two companies are currently conducting final beta testing (POC) with medical institutions.


According to SumMeditech, the AI-powered program is characterized by its intelligent system that analyzes and predicts complex patterns. Previous reduction prevention programs were limited to checking only predefined rules.


By utilizing deep learning-based pretext analysis, the accuracy of the review process has been enhanced. The AI goes beyond simple automation, performing self-analysis, decision-making, and verification. This approach ensures the stability and scalability of the program and strengthens the overall quality of service.


Lee Byungseol, CEO of SumMeditech, stated, "This technological advancement signifies a transition to an intelligent system in which AI analyzes and predicts complex patterns, greatly improving the detection rate of reductions and omissions."


The AI solution not only checks for rule violations but also learns the reduction causes and assessment criteria of the Health Insurance Review and Assessment Service (HIRA). For example, it can issue advance warnings for prescriptions that are medically appropriate but are likely to be reduced due to a lack of necessity during the review process.


Medical professionals can receive real-time notifications about reduction risks or the need to supplement medical records at the time of prescription, thereby improving clinical efficiency. On the administrative side, the performance of analyzing reduction causes by department, filing objections, and resubmitting claims has improved by more than 30% compared to the existing ROI solution.


Technically, more than 5 million reduction case records accumulated by HIRA over the past 10 years have been used as training data. The machine learning engine automatically identifies reduction patterns by medical specialty, diagnosis, and procedure.


Kim Myunghwan, CEO of MsysTech, explained, "By leveraging transformer-based natural language processing technology, the system analyzes the appropriateness of medical records and learns the assessment criteria of reviewers to precisely predict the likelihood of reductions."



The system is built on Oracle and MariaDB, enabling high-performance processing that completes verification within 0.3 seconds of prescription entry. In addition, updates to HIRA's monthly notifications are automated, and the system is configured to continuously learn from new reduction cases to maintain prediction accuracy.


This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing