IND-006

AI-Enabled Fraud Detection and Analytics for India's National Health Insurance Scheme (AB PM-JAY)

Download PDF
India South Asia Lower middle income Full Production Deployment Confirmed

National Health Authority (NHA), Ministry of Health and Family Welfare (MoHFW), Government of India

At a Glance

What it does Anomaly and change detection — Compliance and integrity
Who runs it National Health Authority (NHA), Ministry of Health and Family Welfare (MoHFW), Government of India
Programme Ayushman Bharat - Pradhan Mantri Jan Arogya Yojana (AB PM-JAY)
Confidence Confirmed
Deployment Status Full Production Deployment
Key Risks Model-related risks
Key Outcomes As of December 2024: 3,42,988 fraud cases detected; 2.
Source Quality 6 sources — News article / media, Government website / press release, Report (government / official)

The Ayushman Bharat - Pradhan Mantri Jan Arogya Yojana (AB PM-JAY) fraud detection system is an operational AI and machine learning deployment by India's National Health Authority (NHA) designed to proactively detect, prevent and deter healthcare fraud across the world's largest government-funded health insurance scheme. AB PM-JAY provides cashless secondary and tertiary care hospitalisation coverage of up to INR 5 lakh per family per year to over 55 crore beneficiaries from economically vulnerable households. The fraud analytics system operates within a zero-tolerance framework targeting suspect or non-genuine medical treatment claims, impersonation of beneficiaries, and up-coding of treatment packages and procedures (PIB, 2022; MoHFW, 2025).

The institutional architecture centres on a two-tier structure: the National Anti-Fraud Unit (NAFU) established at the NHA for overall monitoring and implementation of the anti-fraud framework, supported by State Anti-Fraud Units (SAFUs) operating at the state level. NAFU coordinates joint investigations with SAFUs across all participating states and union territories. The NHA has issued comprehensive anti-fraud guidelines that govern the investigation, escalation and sanctioning processes (NHA, 2022; PIB, 2022).

The technology stack deployed by NAFU encompasses 57 distinct technologies and analytical methods. These include rule-based triggers that flag transactions meeting predefined suspicious criteria, machine learning algorithms that identify complex patterns of fraudulent behaviour across large volumes of claims data, fuzzy logic for approximate matching and entity resolution, image classification for verifying medical documentation and patient photographs, and de-duplication algorithms to detect multiple claims filed for the same patient or treatment episode. Together, these technologies enable risk-scoring of individual hospitals and claims, identification of outlier billing patterns, and collusion analytics to detect coordinated fraud networks among providers (IBTimes India, 2025; PIB, 2022).

Transaction data flowing through NHA's IT systems is monitored on a near real-time basis through a dedicated dashboard known as the Risk Assessment, Detection and Analytical Reporting (RADAR) system. This dashboard highlights suspicious cases to investigators, enabling rapid triage. States and union territories have access to real-time dashboards that enhance transparency and accountability in claims processing. Regular monitoring and cleansing of databases, combined with additional data analytic techniques, form part of the ongoing surveillance infrastructure (IBTimes India, 2025; MoHFW, 2025).

The NHA engaged five specialist data analytics firms to develop and deploy the fraud analytics solution: SAS Institute, MFX, Optum, LexisNexis, and GreenOjo. These firms were selected from among 24 applicants following a Request for Empanelment (RFE) process. The engagement of multiple analytics providers reflects the scale and complexity of the fraud detection challenge across millions of claims (Medibulletin, 2020; NHA, 2025 RFE).

Aadhaar-based biometric verification of beneficiaries at the time of hospital admission and discharge has been mandated at all private hospitals empanelled under the scheme. This biometric layer serves as a deterrent against impersonation fraud and ensures that the person receiving treatment matches the enrolled beneficiary. All claims additionally require mandatory supporting documentation and on-bed patient photographs before authorisation can proceed (PIB, 2022).

The human oversight model follows a human-in-the-loop (HITL) approach. Algorithmic flags and risk scores generated by the AI systems do not automatically trigger sanctions. Instead, flagged cases are routed through desk audits conducted by medical professionals and, where warranted, field investigations and surprise inspections carried out by State Health Agencies. Only after this human review process are penalties, suspensions, or de-empanelment actions taken against hospitals found to have engaged in fraud or abuse (NHA, 2022; MoHFW, 2025).

The enforcement outcomes demonstrate the system's operational scale and impact. As of December 2024, NAFU had processed 6.66 crore claims, identifying 2.7 lakh claims from private hospitals worth INR 562.4 crore as non-admissible on account of abuse, misuse, or incorrect entries. A total of 3,42,988 fraud cases were detected across the scheme. In terms of sanctions, 3,167 hospitals were found guilty of irregularities or violations, 1,114 hospitals were de-empanelled from the scheme entirely, 549 hospitals were suspended, and INR 122 crore in penalties were levied on 1,504 errant hospitals. Over 3 lakh Ayushman beneficiary cards were also disabled where fraud was confirmed. Earlier reporting indicated INR 9.5 crore recovered through the AI/ML-based anti-fraud initiative specifically (MoHFW, 2025; IBTimes India, 2025; GS SCORE, 2022).

The system has been operational since approximately 2020, evolving from an initial rule-based approach to incorporate increasingly sophisticated ML algorithms and the engagement of external analytics firms. Approximately 0.18 per cent of total authorised hospital admissions have been confirmed as fraudulent since inception, though the broader screening identifies a significantly larger pool of suspicious transactions for human review (PIB, 2022; IBTimes India, 2025).

Key challenges include the scale of the programme across India's federated health system with varying state-level capacity, the need for continuous model retraining as fraud patterns evolve, and ensuring that false positive flags do not unduly burden legitimate healthcare providers or delay patient access to covered treatments. The procurement of analytics services through competitive RFE processes suggests ongoing evolution of the technology platform rather than a single settled deployment.

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

Social Protection Functions

Implementation/delivery chain
Accountability mechanisms primaryProvision of payments/services
SP Pillar (Primary) The social protection branch: social assistance, social insurance, or labour market programmes. Social insurance
Programme Name Ayushman Bharat - Pradhan Mantri Jan Arogya Yojana (AB PM-JAY)
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 India's national publicly funded health insurance scheme providing cashless secondary and tertiary care hospitalisation coverage of up to INR 5 lakh per family per year to over 55 crore beneficiaries from economically vulnerable households. Administered by the National Health Authority (NHA) under the Ministry of Health and Family Welfare.
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 Hybrid
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 Commercial/proprietary
Compute Environment Where the AI system runs: on-premise, government cloud, commercial cloud, or edge/device. View in glossary Not documented
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 Not assessed
Hybrid Components Combination of rule-based triggers (predefined fraud criteria), classical ML algorithms (pattern detection, risk scoring, outlier detection, collusion analytics), fuzzy logic (entity resolution and approximate matching), image classification (medical documentation and patient photograph verification), and de-duplication algorithms. 57 distinct technologies deployed across the analytics stack.
Data Residency Where the data used by the AI system is stored: domestic, regional, or international. View in glossary Not documented
Cross-Border Transfer Whether data crosses national borders, and if so, whether documented safeguards are in place. View in glossary Not documented
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 Moderate
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 Mix of in-house and third-party
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. Formal assessment

Risk Dimensions

Governance and institutional oversight risks
Market, sovereignty and industry structure risks

Impact Dimensions

  • Grievance mechanism
  • Human oversight protocol
  • Independent evaluation
CategorySensitivityCross-System LinkageAvailabilityKey Constraints
Administrative data from other sectorsPersonalLinks data across multiple systemsCurrently available and usedHospital and provider metadata including empanelment records, facility characteristics, historical performance; used for risk scoring and outlier analysis
Beneficiary registries and MISSpecial categoryLinks data across multiple systemsCurrently available and usedBeneficiary identifiers linked to Aadhaar biometric records for identity verification at admission and discharge; 24.33 crore Ayushman cards issued; used for de-duplication and impersonation detection
Financial and payments data: programme operationsPersonalLinks data across multiple systemsCurrently available and usedClaims and transaction data from NHA IT systems including treatment codes, billing amounts, hospital identifiers, dates of admission and discharge; 6.66 crore claims processed by NAFU
National ID and biometric databasesSpecial categoryLinks data across multiple systemsCurrently available and usedAadhaar-based biometric verification at private hospitals; biometric data used for beneficiary authentication at point of care
Unstructured and text-based contentSensitiveSingle source (no linkage)Currently available and usedMedical images and on-bed patient photographs required for claims authorisation; used for image classification component of fraud detection

IBTimes India (2025) 'Government Detects Rs 562.4 Crore in Fake AB-PMJAY Claims, Implements AI-Based Monitoring to Curb Fraud', International Business Times India, [online]. Available at: https://www.ibtimes.co.in/government-detects-rs-562-4-crore-fake-ab-pmjay-claims-implements-ai-based-monitoring-curb-fraud-879401 (Accessed: 24 March 2026).

View source News article / media

Medibulletin (2020) '5 analytical firms look for fraud in Ayushman Bharat PMJAY', Medibulletin, [online]. Available at: https://medibulletin.com/5-analytical-firms-look-for-fraud-in-ayushman-bharat-pmjay/ (Accessed: 24 March 2026).

View source News article / media

Ministry of Health and Family Welfare (2025) 'Update on Strengthening of PM-JAY Implementation' (Rajya Sabha written reply, 19 August). New Delhi: MoHFW/PIB. Available at: https://www.mohfw.gov.in/?q=en/press-info/9166 (Accessed: 31 October 2025).

View source Government website / press release

National Health Authority (2024) Anti-Fraud Framework Practitioners' Guidebook: Ayushman Bharat Pradhan Mantri Jan Arogya Yojana. New Delhi: NHA. Available at: https://cdnbbsr.s3waas.gov.in/s3169779d3852b32ce8b1a1724dbf5217d/uploads/2024/09/20240924831436164.pdf (Accessed: 24 March 2026).

View source Report (government / official)

Press Information Bureau (2022) 'Anti-fraud system for India's National Health Insurance Scheme (AB-PMJAY)'. New Delhi: PIB, Ministry of Health and Family Welfare, 2 August. Available at: https://pib.gov.in/PressReleseDetailm.aspx?PRID=1847423 (Accessed: 31 October 2025).

View source Government website / press release

Press Information Bureau (2023) 'Use of AI for checking frauds under AB-PMJAY' (Rajya Sabha written reply, 8 August). New Delhi: PIB, Ministry of Health and Family Welfare. Available at: https://www.pib.gov.in/PressReleasePage.aspx?PRID=1946706 (Accessed: 24 March 2026).

View source Government website / press release
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. National: covers 55+ crore beneficiaries across all participating states and union territories. NAFU has processed 6.66 crore claims. 24.33 crore Ayushman cards issued as of August 2023.
Funding Source The source(s) of funding for the AI system development and deployment. Government of India (central and state governments fund AB PM-JAY)
Technical Partners External technology vendors, academic partners, or development partners involved. SAS Institute, MFX, Optum, LexisNexis, and GreenOjo engaged as analytics providers via competitive RFE process. Selected from 24 applicants.
Outcomes / Results As of December 2024: 3,42,988 fraud cases detected; 2.7 lakh claims worth INR 562.4 crore identified as non-admissible from 6.66 crore processed claims; 3,167 hospitals found guilty of irregularities; 1,114 hospitals de-empanelled; 549 hospitals suspended; INR 122 crore in penalties levied on 1,504 hospitals; 3+ lakh Ayushman cards disabled. Approximately 0.18% of total authorised admissions confirmed as fraudulent since inception.
Challenges Scale of programme across India's federated health system with varying state-level capacity; continuous model retraining needed as fraud patterns evolve; risk of false positives burdening legitimate providers or delaying patient access; ongoing procurement cycles for analytics services indicate platform still evolving.

How to Cite

DCI AI Hub (2026). 'AI-Enabled Fraud Detection and Analytics for India's National Health Insurance Scheme (AB PM-JAY)', AI Hub AI Tracker, case IND-006. Digital Convergence Initiative. Available at: https://socialprotectionai.org/use-case/IND-006 [Accessed: 1 April 2026].

Change History

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