A single brief dynamic amplitude-modulated envelope-following response plus machine learning reads out cochlear neural degeneration in gerbils and transfers to human listeners

dAM EFRs revealed selective deficits at fast modulation rates without threshold elevation in Mongolian gerbils with histologically verified CND. Conventional EFR protocols are too slow for clinical use; this work sweeps the full modulation spectrum in a single brief stimulus, compressing the neural-degeneration readout into one measurement. CND in the animals was histologically verified, and the deficit appeared as modulation-rate selectivity rather than a change in hearing threshold, pointing to synapse and auditory-nerve-fiber loss rather than reduced peripheral sensitivity.

A machine-learning classifier distinguished young from middle-aged animals with high accuracy, and its most informative feature, power near 400-500 Hz, tracked synapse counts. Linking that feature to synapse counts moves the classification result from an age label toward cochlear neural health itself, providing an interpretable physiological anchor. The evidence rests on the correspondence between histological synapse counts and the EFR feature; the abstract reports high accuracy but gives no specific values.

Applied without retraining to 56 human listeners, the gerbil-trained classifier separated age groups above chance. Direct cross-species transfer suggests the readout captures a relatively conserved neurophysiological signal rather than a species-specific artifact. The human sample is 56 listeners and the result is above chance rather than high accuracy, making this an initial cross-species validation.

The result is aimed at audiology and otology settings that need an objective assessment of cochlear neural health rather than hearing threshold alone, and at research and clinical workflows that want to screen for neural degeneration with one brief stimulus; the animal-side conclusion rests on Mongolian gerbils with histologically verified CND, and the human-side conclusion rests on 56 listeners with a non-retrained cross-species application, so the current scope is a candidate marker and screening tool for CND rather than a replacement for full diagnosis.

A careful reader would still watch: the human side reports only above-chance age-group separation, so its effect size and individual-level usability remain unclear; whether the link between power near 400-500 Hz and synapse counts also holds in humans; and the classifier separates age groups rather than directly histologically confirmed CND, so the inference from an age label to a neural-health readout still needs validation in humans. The loaded text is an incomplete abstract plus a competing-interest statement, missing figures, sample details, and statistical values, and those gaps limit further judgment of evidential strength.