Projects
Intelligence from Population-Scale Health Insurance Data
Public health insurance systems record, transaction by transaction, the real patterns of illness, access, provider behaviour, and fiscal exposure across a population. At sufficient scale, these administrative records become one of the most powerful sources of health system intelligence available to government—if they can be properly structured, secured, and analysed.
DPIC, in partnership with the Department of Health & Family Welfare and the Indian Institute of Science, is building the data architecture and analytical systems needed to realise this potential—transforming Odisha’s health insurance claims data from an administrative record into a strategic resource for evidence-based health system management.
Curating Population-Scale Health Data
This initiative draws on a multi-year administrative dataset covering 2018 to 2025: over 5 million insurance transactions across more than 2 million beneficiaries, generating in excess of 15 terabytes of structured and unstructured medical records. These data include detailed claims histories, treatment packages, beneficiary demographics, hospital-level transactions, and reimbursement flows.
DPIC has designed data pipelines to clean, standardize, anonymize, and link these records — creating a unified, longitudinally consistent foundation. By connecting high-volume transaction data with associated medical records at this scale, the system enables analysis that moves well beyond transactional reporting toward population-level health intelligence.
Secure AI Infrastructure
DPIC is establishing a secure analytics environment with privacy-preserving anonymization pipelines, controlled access protocols, and sandbox testing infrastructure for sensitive health data.
Within this environment, AI tools — including predictive modelling, NLP-based document extraction, and anomaly detection — are being developed and evaluated. The applications under development span disease trend analysis, utilization forecasting, fraud detection, and fiscal risk modelling. DPIC’s approach prioritises rigorous testing and validation before any application moves toward operational use — ensuring that AI tools are assessed for accuracy, stability, and equity implications in a controlled setting.
Health System Intelligence Applications
By structuring and analysing millions of transactions and terabytes of medical records, DPIC is developing decision-support capabilities that enable:
- Identification of emerging disease patterns and geographic health burden shifts.
- Detection of over- and under-utilised treatment packages, informing benefit design.
- Fraud and anomaly detection to strengthen fiscal discipline across the insurance system.
- Equity diagnostics mapping access and utilisation disparities across districts and demographic groups.
The objective is to move Odisha’s health insurance system from retrospective claims processing to anticipatory, intelligence-driven health system management.
The Value of Population-Scale Health Insurance Data
Most health systems operate with fragmented, incomplete, or delayed data — making it difficult to anticipate pressures, identify inequities, or allocate resources efficiently. Odisha’s insurance claims data, structured and analysed at population scale, offers a rare opportunity to change this. By investing in the data architecture, secure infrastructure, and analytical tools needed to unlock this resource, the state is building durable capacity to manage its health system more responsively and more equitably.