Artificial Intelligence Sequential Decision Optimisation for Postpartum Haemorrhage Triage Under Variable Vital-Sign Missingness: A computed Evaluation of Healthcare Maternal Safety

Postpartum haemorrhage remains a leading cause of maternal mortality, and early triage decisions are often made with incomplete vital-sign data. The framework combines a constrained partially observable Markov decision process with a missingness-aware belief update, and it is benchmarked against complete-case and mean-imputation policies. Using a before-after-control-impact design with 8, 16, 24 and 32 clusters, effect sizes of 3, 5 and 8 percentage points, and 2000 replications per cell, the replication estimated power and minimum detectable effect under a variance structure with between-cluster SD = 6.0, residual SD = 4.5 and intraclass correlation = 0.64. Power reached conventional levels only at the larger cluster counts and effect sizes, and the minimum detectable effect at 80% power was 5 percentage points with 52 clusters. The results indicate that missingness-aware sequential policies can improve triage safety relative to complete-case analysis, but that the cluster counts needed to detect realistic improvements are substantial. The contribution is a formalisation of missingness-aware triage as a constrained sequential decision problem and a design-analysis template for evaluating such policies before field deployment.