KOR-001

National Health Insurance Service (NHIS) – Machine Learning-Based Fraud Detection and Prediction System

Download PDF
Korea, Rep. East Asia & Pacific High income Full Production Deployment Confirmed

National Health Insurance Service (NHIS); Ministry of Health and Welfare (MOHW)

At a Glance

What it does Anomaly and change detection — Compliance and integrity
Who runs it National Health Insurance Service (NHIS); Ministry of Health and Welfare (MOHW)
Programme National Health Insurance Service (NHIS) – Machine Learning-Based Fraud Detection and Prediction System
Confidence Confirmed
Deployment Status Full Production Deployment
Key Risks Model-related risks
Key Outcomes Between 2014 and 2021, detected 567,820 cases of fraud equivalent to approximately USD 174 million in claim value.
Source Quality 5 sources — Government website / press release, Report (multilateral / development partner), Report (government / official)

The National Health Insurance Service (NHIS) of the Republic of Korea operates a machine learning-based fraud detection and prediction system designed to identify irregularities and forecast potential misuse in healthcare provider billing across the country's universal health insurance programme. South Korea's National Health Insurance system covers virtually all residents, with healthcare services delivered predominantly by private-sector providers who submit claims to NHIS for reimbursement. The fraud detection system targets healthcare facilities that are not established in accordance with current regulations, for example those operated by unqualified persons seeking to maximise profit through fraudulent insurance claims, with a particular focus on long-term care benefits and health insurance eligibility fraud.

NHIS began applying artificial intelligence to fraud detection from 2020, building upon earlier rule-based detection approaches. The system operates as a hybrid detection architecture that combines traditional rule-based screening with AI predictive models to identify information with a high probability of fraud. The machine learning component employs a range of supervised learning techniques, including deep learning, random forest, gradient boosting, logistic regression, and support vector machine (SVM) algorithms. These models are trained on NHIS healthcare big data, which includes socio-demographic variables, disease and treatment history records, and provider billing patterns. The system generates ranked alerts for investigator review, flagging cases with the highest probability of fraudulent activity for priority examination by human investigators.

The transition from experiential, manual screening approaches to data-driven management represented a significant operational shift for NHIS. Previously, fraud detection relied heavily on the experience and intuition of individual investigators reviewing claims data. The machine learning system enables proactive data analysis to identify and manage fraudulent activities systematically rather than reactively. The system processes administrative health insurance claims, provider billing records, and beneficiary identifiers, with beneficiary data pseudonymised in accordance with the Personal Information Protection Act (PIPA). PIPA, one of the strictest data protection frameworks globally, classifies health data as sensitive information under Article 23 and imposes stringent requirements on data controllers, including mandatory designation of a Chief Privacy Officer, data protection impact assessments, and penalties for non-compliance including corrective orders, penalty surcharges, and potential criminal prosecution.

In terms of documented outcomes, the fraud detection system has demonstrated substantial results. Between 2014 and 2021, NHIS detected 567,820 cases of fraud, equivalent to approximately 174 million United States dollars in claim value. In 2023, NHIS enhanced its fraud detection system to combat evolving forms of fraud, uncovering 18.534 billion Korean won (approximately 14 million USD) in unjust enrichment through the machine learning-based detection and prediction system. These results reflect a significant curbing of financial leakage in long-term care financing and healthcare claims processing.

The human oversight model follows a human-in-the-loop paradigm. Investigators manually review and adjudicate all cases flagged by the system before any enforcement action or sanction is imposed. The AI system outputs inform compliance investigations but do not autonomously impose sanctions or alter entitlements. This approach ensures that algorithmic outputs serve as decision support tools rather than automated decision-makers, reflecting the high-criticality nature of the decisions involved, which can affect healthcare provider licensing, financial penalties, and access to the national health insurance system.

The system operates within a governance framework defined by several key pieces of legislation. The Personal Information Protection Act (PIPA) governs the processing of personal and sensitive health data, requiring pseudonymisation for analytical purposes and imposing strict data handling obligations. The Digital Government Act and the National Data Innovation Strategy provide the broader policy framework for data-driven governance and digital transformation across Korean government agencies, including NHIS. NHIS is also subject to internal audit controls and Ministry of Health and Welfare data-ethics oversight.

The technical infrastructure is reported to operate on NHIS on-premise systems within national government cloud architecture, with data maintained under domestic jurisdiction. No public disclosure has been made of specific third-party vendors or commercial software partners involved in developing or maintaining the machine learning models, and the technical architecture of the AI pipeline has not been publicly documented in detail. The International Social Security Association (ISSA) has recognised the NHIS fraud detection system as a good practice example for data-driven innovation in social security across the Asia and Pacific region, documenting the system in multiple publications and good practice awards.

Classifications follow the DCI AI Hub Taxonomy. Hover over field labels for definitions.

Social Protection Functions

Implementation/delivery chain
Accountability mechanisms primaryManagement of contributions and withdrawals
SP Pillar (Primary) The social protection branch: social assistance, social insurance, or labour market programmes. Social insurance
Programme Name National Health Insurance Service (NHIS) – Machine Learning-Based Fraud Detection and Prediction System
Programme Type The type of social protection programme, classified under social assistance, social insurance, or labour market programmes. View in glossary Health Insurance
System Level Where in the social protection system the AI is applied: policy level, programme design, or implementation/delivery chain. View in glossary Implementation/delivery chain
Programme Description South Korea's universal National Health Insurance programme, administered by NHIS under the Ministry of Health and Welfare, covering virtually all residents. The ML-based fraud detection system targets fraudulent billing by healthcare providers, with particular focus on long-term care benefits and health insurance eligibility fraud.
Implementation Type How the AI output is produced: Classical ML, Deep learning, Foundation model, or Hybrid. Affects validation, compute requirements, and governance profile. View in glossary Classical ML
Lifecycle Stage Current stage in the AI lifecycle, from problem identification through to monitoring, maintenance and decommissioning. View in glossary Monitoring, Maintenance and Decommissioning
Model Provenance Origin of the AI model: developed in-house, adapted from open-source, commercial/proprietary, or accessed via third-party API. View in glossary Not documented
Compute Environment Where the AI system runs: on-premise, government cloud, commercial cloud, or edge/device. View in glossary National/government cloud
Compute Provider The specific cloud or infrastructure provider hosting the AI system. NHIS on-premise infrastructure
Sovereignty Quadrant Classification of data and compute sovereignty: I (Sovereign), II (Federated/Hybrid), III (Cloud with safeguards), or IV (Shared Innovation Zone). View in glossary I — Sovereign AI Zone
Data Residency Where the data used by the AI system is stored: domestic, regional, or international. View in glossary Domestic
Data Residency Detail Additional detail on the specific data hosting arrangements and jurisdictions. Data maintained under domestic Korean jurisdiction within NHIS government infrastructure
Cross-Border Transfer Whether data crosses national borders, and if so, whether documented safeguards are in place. View in glossary None
Decision Criticality The rights impact of the decision the AI supports. High criticality requires HITL oversight; moderate requires HOTL; low may operate HOOTL. View in glossary High
Human Oversight Type Level of human involvement: Human-in-the-Loop (active review), Human-on-the-Loop (monitoring), or Human-out-of-the-Loop (periodic audit). View in glossary HITL
Development Process Whether the AI system was developed fully in-house, through a mix of in-house and third-party, or fully by an external provider. View in glossary Fully in-house
Highest Risk Category The most significant structural risk source identified: data, model, operational, governance, or market/sovereignty risks. View in glossary Model-related risks
Risk Assessment Status Whether a formal risk assessment, informal assessment, or independent audit has been conducted for this system. Not assessed

Risk Dimensions

Governance and institutional oversight risks
Operational and system integration risks

Impact Dimensions

Equality, non-discrimination, fairness and inclusion
  • DPIA/AIA conducted
  • Data minimisation controls
  • Human oversight protocol
CategorySensitivityCross-System LinkageAvailabilityKey Constraints
Administrative data from other sectorsPersonalLinks data across multiple systemsCurrently available and usedHealthcare facility registration and licensing data used to identify facilities not established in accordance with regulations
Beneficiary registries and MISSpecial categoryLinks data across multiple systemsCurrently available and usedHealth insurance claims and beneficiary records pseudonymised under PIPA; covers virtually all residents; includes socio-demographic, disease and treatment history variables
Financial and payments data: programme operationsSensitiveLinks data across multiple systemsCurrently available and usedHealthcare provider billing records submitted to NHIS for reimbursement; includes transaction-level claims data across all provider types

Government of Korea (2024). Digital Government Act and National Data-Innovation Strategy. Seoul: Ministry of the Interior and Safety. Available at: https://www.mois.go.kr/eng/ (Accessed 31 Oct 2025).

View source Government website / press release

International Social Security Association (ISSA) (2024). Operation of Machine Learning-Based Fraud Detection and Prediction System – National Health Insurance Service (Republic of Korea). Geneva: ISSA. Available at: https://www.issa.int/gp/259276 (Accessed 24 Mar 2026).

View source Report (multilateral / development partner)

ISSA (2024) 'Detecting fraud in health care through emerging technologies', issa.int. Available at: https://www.issa.int/analysis/detecting-fraud-health-care-through-emerging-technologies (Accessed: 27 March 2026).

View source Report (multilateral / development partner)

ISSA (2024) 'Data-driven innovation in social security: Good practices from Asia and the Pacific', issa.int. Available at: https://www.issa.int/analysis/data-driven-innovation-social-security-good-practices-asia-and-pacific (Accessed: 27 March 2026).

View source Report (multilateral / development partner)

National Health Insurance Service (NHIS) (2025). Illegality and Fraud Information System in NHIS – Fraud Detection. Wonju: NHIS Global Hub. Available at: https://nhis-globalhub.com/images/qr/202502/pdf/1105_14.pdf (Accessed 24 Mar 2026).

View source Report (government / official)
Deployment Status How far the system has progressed into real-world operational use, from concept/exploration through to scaled and institutionalised. View in glossary Full Production Deployment
Year Initiated The year the AI system was first initiated or development began. 2020
Scale / Coverage The scale and geographic or population coverage of the deployment. Nationwide — covers virtually all residents of South Korea through the universal NHI programme
Funding Source The source(s) of funding for the AI system development and deployment. NHIS operational budget (statutory national health insurance financing)
Technical Partners External technology vendors, academic partners, or development partners involved. No public disclosure of vendor or software architecture. ISSA documentation does not identify third-party technology partners.
Outcomes / Results Between 2014 and 2021, detected 567,820 cases of fraud equivalent to approximately USD 174 million in claim value. In 2023, uncovered KRW 18.534 billion (approximately USD 14 million) in unjust enrichment. Significantly curbed financial leakage in long-term care financing.

How to Cite

DCI AI Hub (2026). 'National Health Insurance Service (NHIS) – Machine Learning-Based Fraud Detection and Prediction System', AI Hub AI Tracker, case KOR-001. Digital Convergence Initiative. Available at: https://socialprotectionai.org/use-case/KOR-001 [Accessed: 1 April 2026].

Change History

Created 30 Mar 2026, 08:40
by v2-import (import)