Generalize HEAR observability scoring to additional heart-rate estimators

Establish whether the HEAR observability score remains useful for heart-rate estimators beyond the estimators evaluated in the study.

Background

HEAR learns an observability score from simulated phase spectra and uses that score to select measurements for heart-rate estimation. The study demonstrates that the score improves selective performance for HEAR’s own heart-rate head and several evaluated external estimators, including FFT-based methods, autocorrelation, harmonic product spectrum, and variational mode decomposition.

The paper explicitly limits this conclusion to the evaluated estimators and identifies the usefulness of the score for other estimators as unresolved. Establishing this would determine whether the learned measurement-quality signal generalizes beyond the specific downstream methods tested.