SASunday A. Adetunji

Research programme

Reproductive/perinatal epidemiology, maternal safety, and clinical AI evaluation.

A coherent programme connecting clinical obstetric experience, causal inference, survey-weighted epidemiology, electronic health records, model validation, and implementation-ready evidence systems.

Maternal Safety & Rescue Systems

Research on the critical interval where preventable maternal deterioration can be recognized, escalated, and rescued.

  • Obstetric rescue timing and delay compression
  • Severe maternal morbidity surveillance
  • Structural inequity in delivery hospitalizations
  • Failure-to-rescue epidemiology

Reproductive & Perinatal Epidemiology

Population and survey-based work on reproductive access, facility birth, intrapartum care, and newborn survival.

  • Digital reproductive access cascade
  • ANC continuity and facility delivery
  • Pregnancy intention and preterm birth
  • Facility care gaps and early neonatal outcomes

Clinical AI & Learning Health Systems

Evaluation of prediction models under real clinical constraints, with attention to leakage, calibration, workload, and deployment.

  • Maternal early-warning systems
  • EHR-based rescue surveillance
  • Leakage-proof AI benchmarking
  • Conformal selective triage

Global Maternal Health & Equity

Cross-country and nationally representative analyses focused on health-system performance and inequitable maternal-risk gradients.

  • DHS multicountry reproductive equity studies
  • Institutional delivery and maternal mortality
  • Country-level causal panels
  • Effective coverage beyond contact metrics

Measurement, Bias & Validity Science

Methods that distinguish signal from artifact in administrative, survey, EHR, and text-based biomedical data.

  • Misclassification-aware phenotyping
  • Missingness and MNAR sensitivity
  • Quantitative/probabilistic bias analysis
  • Trial generalizability audits

Biostatistics Education & Reproducibility

Teaching and reproducible-methods work translating complex statistical models into transparent applied workflows.

  • Longitudinal linear mixed models
  • Reproducible R/Python/SAS workflows
  • Survey-weighted causal templates
  • Open teaching modules

Output distribution

Research streams represented in the current evidence archive.

The portfolio remains expandable through the JSON-based CMS while preserving a stable public architecture.