👤 Lindsay C Hanford

We tested whether spontaneous speech acoustics provide a scalable digital marker of biologically defined Alzheimer's disease (AD) risk. Forty-nine cognitively unimpaired older adults were stratified within APOE genotype into Low-, Moderate-, and High-Risk groups based on log₁₀-transformed plasma p-tau217. Acoustic features were extracted from spontaneous speech and entered into multiclass SVM classifiers with leave-one-out cross-validation, with and without genetic-algorithm feature selection and age. Parallel models using neuropsychological measures were evaluated for comparison. Feature contributions were interpreted using SHAP. Speech-based models substantially outperformed cognition-only models and exceeded chance performance for three-group classification (33.3%), achieving up to 77% accuracy compared with 47% for neuropsychological models. SHAP analyses identified a compact, stage-dependent acoustic signature dominated by voice-quality, spectral-envelope, and formant-bandwidth features, with age contributing secondary effects. Spontaneous speech acoustics capture p-tau217/APOE-defined AD risk despite preserved cognition, supporting speech as a scalable, biologically grounded biomarker for preclinical AD risk stratification. Show less

Alzheimer's disease (AD) is the most prevalent form of dementia and a major public health challenge. In the absence of a cure, accurate and innovative early diagnostic methods are essential for proactive life and healthcare planning. Speech metrics have shown promising potential for identifying individuals with mild cognitive impairment (MCI) and AD, prompting investigation into whether speech motor features can detect elevated risk even prior to cognitive decline. This preliminary study examined whether speech kinematic features measured during a color-word interference task could distinguish cognitively normal APOE-ε4 carriers (ε4 Sixteen cognitively normal older adults ( Although no group differences reached statistical significance after accounting for multiple testing, several features showed moderate effect sizes. The optimal SVM model achieved 87.5% cross-validated accuracy (precision 88.9%, sensitivity 88.9%, specificity 85.7%) using three features: (1) lip movement duration during the pre-interference segment, (2) average lip speed during interference, and (3) the change in lip movement range from pre- to during-interference segments (ΔDuring-Pre). These findings suggest that lip kinematic responses to mild cognitive-motor interference may capture subtle neuromotor differences associated with APOE-ε4 status in cognitively intact older adults. The identified features point to potential alterations in anticipatory motor planning, interference susceptibility, and articulatory adaptability in ε4 Show less