SKAI Builds NHIS System to Detect Fraudulent Employee Enrollment, Uses Graph DB to Flag Cases

Sky Worldwide (SKAI) has completed the development of a system for the National Health Insurance Service (NHIS) to detect fraudulent enrollment as an employee subscriber, using domestically developed graph database technology. The system identifies suspected cases of fraudulent enrollment that are difficult to detect from individual records alone by analyzing relationships among data. SKAI plans to use this project as a springboard to expand its data analytics and artificial intelligence (AI) transformation (AX) business with public institutions.

SKAI said on November 10 that it had developed the detection system with its partner The Sky Lab as part of the NHIS project to rebuild an integrated collection information system for digital transformation. The system integrates collection operations for National Health Insurance, the National Pension, Employment Insurance, and Industrial Accident Compensation Insurance. The domestically developed graph database AgensGraph was used to build a fraudulent enrollment detection model and a self-inspection system.

Fraudulent enrollment as an employee subscriber involves registering as an employee subscriber without actually working, in order to avoid high local subscriber premiums. Although individual enrollment records alone make it difficult to distinguish fraudulent cases from legitimate ones, linking and analyzing information such as assets, income, and the reported size of a workplace can reveal combinations of information that do not match.

According to the NHIS, 9,202 people were caught fraudulently enrolled as employee subscribers from 2023 through 2025, and approximately KRW 66.6 billion in local subscriber premiums was assessed retroactively. The NHIS believes that many cases have gone undetected, as the methods used are becoming increasingly sophisticated.

The system was designed to select the data fields needed for analysis from collection data, organize them into a knowledge graph, and analyze relationships among the information using AgensGraph. In addition to the existing rules, 18 new detection scenarios were created. Staff can adjust on-screen criteria involving dozens to hundreds of fields to identify people for further review, then refer the final list to regional officials for on-site inspections.

A key feature is that staff can check why a person was flagged and what evidence supported the decision. Records show which combinations of information met the detection criteria, helping staff make assessments. If new methods of fraudulent enrollment emerge, the system can be updated by adding detection criteria and links between data.

SKAI also applies graph database technology to investigations and telecommunications network analysis, in addition to verifying health insurance eligibility. It has built similar systems using relationship analysis for the National Tax Service and the National Police Agency. The National Police Agency's system for searching for and tracking drug offenders uses a 3D display to show links between individuals who appear as buyers and sellers in different cases, among more than 140,000 people under monitoring. The Seoul Metropolitan Government's integrated optical communications network control system maps the connections among optical cables and equipment linking 25 district offices.

The company sees graph databases as a foundational technology for AI analysis as well. The NHIS system is also structured to allow AI analysis capabilities to be added to the graph data in the future. SKAI plans to expand its public-sector AX business by integrating AgensGraph, which connects data relationships, with Ontovia, its AI solution.

In 2020, SKAI donated some of AgensGraph's graph modules to the Apache Software Foundation under the name Apache AGE. The company said the technology is used as a standard graph database feature in Microsoft Azure.

Shin Jaehyeok, CEO of Sky Worldwide, said, "The global graph database market is growing rapidly because, in the AI era, competitiveness depends not on the volume of data but on the ability to read the relationships hidden within it. Building on graph technology proven in the public sector, we will grow into a key public-sector AX company that connects everything from data to AI."

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