Gait-Based Parkinsonian Tremor Detection Using Smartphone Accelerometry Under Variable Footwear Conditions: A Calibration-Transfer Evaluation of Computer Vision Diagnostic Accuracy

Smartphone accelerometry offers a low-cost route to objective gait assessment for Parkinsonian tremor, but the diagnostic accuracy of computer vision pipelines trained on accelerometer-derived spectrograms is known to degrade when footwear changes between training and deployment. the present analysis treats footwear as a calibration-transfer problem and characterises the operating characteristics of a before-after-control-impact (BACI) evaluation design for detecting that degradation. No observational data are used or claimed. Under a declared variance structure with between-cluster SD = 6.0, residual SD = 4.5 and intraclass correlation = 0.640, power was estimated across 2000 design cells spanning cluster counts of 8, 16, 24 and 32 and effect sizes of 3, 5 and 8 percentage points.54 clusters are required to detect a 5 percentage-point loss of diagnostic accuracy at 80% power, and that the minimum detectable effect falls steeply as cluster count rises from 8 to 32. Kernel density and residual diagnostics confirm that the estimator is well calibrated across the design grid. The contribution is a design-level account of how calibration-transfer studies of gait-based tremor detection should be powered, and a demonstration that footwear-induced accuracy loss is detectable only at cluster counts that most published smartphone gait studies do not reach.