A Decodability-Grounded Audit of Clinical Prediction Claims in PPG Foundation Models

What does held-out clinical prediction from photoplethysmography (PPG) reveal about physiology? We audit frozen representations through concept decodability, covariate-adjusted and incremental prediction, and post-erasure predictability after probe refitting, with encounter-matched ECG comparisons. Across four encoders and $37,784$ MC-MED encounters, AnyPPG decodes pulse-derived concepts well. For potassium on $6,348$ matched encounters, its correlation falls from $0.21$ to $0.05$ after age and renal adjustment, whereas ECGFounder retains $0.32\to 0.27$ (paired adjusted difference $+0.22$; 95% CI $[0.19,0.26]$). Creatinine likewise attenuates under target-derivative-free adjustment. On a separate $5,160$-encounter complete-case cohort, adding ECGFounder to the same age, estimated glomerular filtration rate (eGFR), and blood urea nitrogen (BUN) baseline raises potassium correlation by about $0.087$ under ridge and nonlinear predictors; direct AnyPPG increments have confidence intervals including zero. Erasure loss gives the opposite modality ordering from interval decodability: after rank-five subspace removal and refitting, AnyPPG loses $0.0156$ in adjusted correlation versus ECGFounder's $0.0081$. AnyPPG starts from a much lower adjusted correlation; its loss is near the heart-rate control. Together, these results identify adjustment-sensitive PPG potassium prediction and stronger retained ECG association, while decodability and erasure capture complementary aspects of the representations.