Digital health sensors capture behavioral and physiological signals continuously outside clinical settings. Combining speech/language models, wearable gait data, passive cognitive probes, and sleep monitoring may detect subtle pre-symptomatic decline. Validation against longitudinal biobank data with PET/CSF ground truth is needed.
Can speech, gait, retinal imaging, sleep, and smartphone data detect neurodegeneration 5-10 years before diagnosis?