Artificial Intelligence Oral Cancer Screening Under Variable Intraoral Camera White Balance Drift: A Color-Constancy Evaluation of Healthcare Referral Concordance

Intraoral cameras used in oral cancer screening produce images whose colour balance drifts with illumination and sensor ageing, and convolutional classifiers trained on colour-stable images may therefore issue referral decisions that depend on camera state rather than on tissue. the present analysis formalises that dependence as a colour-constancy failure and evaluates, by design analysis rather than by observation, how much referral concordance a screening pipeline can be expected to retain under white-balance drift. A before-after-control-impact design is characterised under a declared variance structure with between-cluster standard deviation 6.0, residual standard deviation 4.5 and intraclass correlation 0.64; power was estimated across 2000 replications per design cell for cluster counts of 8, 16, 24 and 32 and effects of 3, 5 and 8 percentage points.54 clusters are required to detect a 5-percentage-point concordance effect at 80 per cent power, that the minimum detectable effect falls steeply as clusters increase, and that the intraclass correlation of 0.64 dominates the precision of any referral-concordance estimate. The contribution is a design-level account of what colour-constancy evaluation can and cannot establish about artificial-intelligence oral cancer screening, together with the sample sizes that such an evaluation would require.