Artificial Intelligence Maternal Hemorrhage Risk Prediction Under Variable Community Blood Pressure Cuff Calibration: A Measurement-Error-Aware Evaluation of Healthcare Referral Urgency

Community blood pressure cuffs used in maternal haemorrhage risk prediction are rarely calibrated to a common reference, and the resulting measurement error propagates into artificial intelligence (AI) referral urgency scores in ways that standard validation metrics do not capture. No observational data are used or claimed. The analysis specifies a variance structure with between-cluster standard deviation 6.0, residual standard deviation 4.5 and intraclass correlation 0.64, and evaluates power across 2000 replications per design cell for cluster counts of 8, 16, 24 and 32 and effect sizes of 3, 5 and 8 percentage points. Under this structure, 54 clusters are required to detect a 5 percentage point improvement at 80% power. The framework distinguishes calibration error from biological variation, formalises the attenuation of referral-urgency discrimination under additive cuff bias, and shows why cluster-level calibration audits must be embedded in the evaluation design rather than treated as a preprocessing step. The contribution is a design-level account of how measurement error in community blood pressure measurement constrains what AI maternal haemorrhage prediction can be shown to achieve.