The National Health Insurance Authority (NHIA) of Ghana operates an AI-enabled claims integrity analytics system designed to detect potentially fraudulent, abusive, or erroneous health insurance claims submitted by credentialed healthcare providers under the National Health Insurance Scheme (NHIS). The system is described in secondary reporting quoting NHIA officials as using artificial intelligence and machine learning to analyse electronic claims data, identify outliers and anomalous trends, and support vetting before reimbursement is authorised (Graphic Online, 2024, citing NHIA Deputy Director of Quality Assurance William Omane-Adjekum). The AI analytics layer sits atop the NHIA's broader digital claims processing infrastructure, which has been operational since 2013 when the Authority piloted its electronic claims (e-claims) system as a replacement for paper-based claims submission (Apaak et al., 2022, PMC9086605).
The underlying digital infrastructure centres on the CLAIM-it platform, a four-module claims management application developed by the NHIA in collaboration with PharmAccess Group. CLAIM-it comprises a claims entry module that enforces NHIA claims generation rules and protocols, a receiving and aggregation system, a claims adjudication module, and a Regional and District Health Director reporting module (Apaak et al., 2022, PMC9534449). Healthcare providers submit claims electronically either through the CLAIM-it web-based tool or via a standardised XML interface that links directly to provider health management information systems. As of available reporting, approximately 2,188 out of over 4,500 credentialed providers (approximately 48.6%) submit claims electronically, with the NHIA setting targets to phase out manual claims processing entirely (GNA, 2024). Claims are processed at four Claims Processing Centres (CPCs) located in Accra, Cape Coast, Kumasi, and Tamale, which serve all regions of the country (Apaak et al., 2022, PMC9086605).
The AI fraud detection component is reported to analyse the electronic claims payloads, which include structured data on facilities, patient encounters, services rendered, medicines dispensed, tariff codes, dates, and provider and patient identifiers conforming to the NHIA e-claims interface. The system is described as using claims-integrity analytics to flag claims that exhibit patterns consistent with fraud, abuse, or billing errors. According to NHIA officials quoted in media coverage, the AI monitors insurance claims to determine outliers and trends, enabling faster and more accurate vetting of claims before reimbursement decisions are made (Herald Ghana, 2024; Graphic Online, 2024). The specific model family, algorithm architecture, and whether the system operates truly in real time or near-real-time batch processing have not been confirmed in recoverable primary NHIA sources.
Flagged claims are reviewed by NHIA Claims and Quality Assurance teams within the Clinical and Compliance Audit framework. The system operates as an advisory and triage tool: AI-generated flags inform but do not determine final audit or payment decisions. Final decisions on claim validity, reimbursement adjustments, or sanctions rest with NHIA auditors who conduct targeted audits based on the AI-generated risk signals. The existence of formal appeal and adjustment pathways is documented: in a notable 2023 case, the NHIA annulled a clinical and compliance audit adjustment of GHS 1,199,841.02 imposed on La Polyclinic following an appeal by the facility's management, and additionally committed to refund GHS 288,809.14 that had been previously deducted from the facility's claims (Modern Ghana, 2023). This demonstrates that the audit process, which the AI system feeds into, includes functioning grievance and redress mechanisms.
The NHIA has reported significant financial outcomes from the AI-enabled fraud detection system. In 2023, the Authority saved approximately GHS 9.5 million (Ghanaian Cedis) from fraudulent claims identified through AI analytics, according to Deputy Director of Quality Assurance William Omane-Adjekum (Graphic Online, 2024). The digital payment platform infrastructure was credited as making this fraud exposure possible by providing the necessary structured data for AI analysis. The NHIA, which serves approximately 30 million subscribers with 17.9 million active members under the leadership of CEO Dr. Da-Costa Aboagye, processes annual claims ranging between GHS 20-30 million from service providers (Herald Ghana, 2024).
The NHIA has also reportedly invested in staff capacity building for the AI system: one NHIA article title indicates that the Claims Directorate trained staff on the use of AI in claims management, but the underlying official page is no longer recoverable as substantive content. The available evidence therefore supports the existence of staff training activity more weakly than it supports the core fraud-savings claim. By contrast, the benefit-cost analysis of the broader e-claims system that underpins the AI layer is well documented in peer-reviewed literature. That analysis found electronic claims processing more efficient than manual processing, with e-claims rejection rates of 3% at district hospitals compared to 10% for paper claims, and 1% at regional hospitals compared to 6% for paper claims (Apaak et al., 2022, PMC9086605). These lower rejection rates reflect the e-claims system's better ability to detect errors in submitted claims at the point of submission, before they reach the AI analytics layer.
The regulatory framework governing data protection in the context of this AI system is the Ghana Data Protection Act, 2012 (Act 843), which established the Data Protection Commission as an independent statutory body mandated to protect the privacy of individuals and regulate the processing of personal information. The Commission maintains the Data Protection Register and ensures compliance with data protection provisions. The extent to which the NHIA has conducted formal data protection impact assessments or AI-specific risk assessments for the fraud detection system is not documented in available sources. Overall, the case remains in the tracker because the existence of AI-enabled fraud analytics is credibly reported and fits the documented digital claims infrastructure, but the loss of recoverable primary NHIA pages means the file should be read as a likely rather than fully confirmed account of the system's technical operation.