Artificial Intelligence Sparse-Data Estimation of Postpartum Hemorrhage Risk Under Variable Blood Bank Inventory Logging Completeness: A Causal Transportability Evaluation of Healthcare Transfusion Readiness

Postpartum haemorrhage remains a leading cause of maternal mortality, and the readiness of a facility's blood bank determines whether haemorrhage becomes survivable. Artificial intelligence models that estimate haemorrhage risk from routine clinical records are increasingly proposed as triage aids, but their transportability across facilities is threatened by a neglected measurement problem: blood bank inventory logs are recorded with variable completeness, and the missingness is not random with respect to the very readiness conditions the model is meant to predict. The framework separates the clinical risk model from the logging process, defines a completeness-adjusted transport functional, and derives identification conditions under which the target-facility risk can be recovered from source-facility data. A design-based evaluation characterises the operating characteristics of the before-after-control-impact estimator used to validate transport under a declared variance structure (between-cluster SD = 6.0, residual SD = 4.5, intraclass correlation = 0.64). Power was estimated across 2000 design cells spanning cluster counts of 8, 16, 24, 32 and effect sizes of 3, 5 and 8 percentage points.54 clusters are required to detect a 5-percentage-point effect at 80% power, and that the intraclass correlation of 0.64 dominates the design's sensitivity. The contribution is a transportability framework that treats logging completeness as a causal variable rather than a nuisance, together with a design-theoretic account of what such evaluations can and cannot establish.