Artificial Intelligence Ultrasound Interpretation for Placental Abruption Under Variable Portable Probe Acoustic Output Drift: A Test-Time Calibration Evaluation of Healthcare Referral Urgency

Portable ultrasound probes used in obstetric triage drift in acoustic output as transducers age and batteries discharge, and an artificial intelligence interpreter trained on stable cart-based images may therefore misclassify placental abruption at the point of care. the present analysis treats test-time calibration as a design problem for referral urgency rather than for image fidelity, and asks how much calibration is needed before an interpreter's urgency calls become safe under drift. A formal model of acoustic output drift is developed, a calibration operator is defined that rescales log-intensity features at inference, and the operating characteristics of the resulting referral rule are characterised under a declared variance structure with between-cluster SD 6.0, residual SD 4.5 and intraclass correlation 0.64. Power was estimated across 2000 design cells for cluster counts of 8, 16, 24 and 32 and effects of 3, 5 and 8 percentage points; 56 clusters were required for 5 percentage points at 80% power. Drift acts as a multiplicative bias on the decision statistic, that uncalibrated referral urgency is biased toward under-triage at low output and over-triage at high output, and that test-time calibration restores the decision threshold only when the drift estimate is itself unbiased. The contribution is a calibration-aware formulation of referral urgency that separates acoustic drift from pathological signal and specifies the cluster counts at which the separation can be evaluated.