Abstract
This dissertation explored the predictive relationship between physiological and voice biomarkers and cognitive changes in middle and later adulthood. Study 1 investigated allostatic load (AL), a composite measure of multisystem physiological dysregulation, as a predictor of cognitive changes over approximately 10 years in healthy middle-aged and older adults from a subsample of the national Midlife in the United States (MIDUS) study. While previous research has primarily focused on cross-sectional analyses, this study investigated the longitudinal relationship of AL to cognitive performance over a decade. Findings revealed that higher AL was associated with greater declines in cognitive domains such as verbal fluency, inductive reasoning, and processing speed. These results suggest that identifying early signs of preclinical physiological risk factors in midlife could present opportunities to reduce or delay cognitive declines. Study 2 evaluated the role of prosody voice biomarkers as predictors of ~10-year cognitive changes within a subsample of the MIDUS study participants. Previous research by Mahon & Lachman (2022) found that prosody voice measured at the same time as the cognitive outcomes was related to changes in word list recall, verbal fluency, and attention switching/inhibitory control reaction time. This study analyzed prosody voice metrics 8 years before the cognitive outcomes. Results indicated that higher jitter, pitch, number of voice breaks, as well as lower shimmer and pulse, predicted cognitive changes over a decade in word list recall, working memory, and attention switching/inhibitory control reaction time. Natural and easily accessible voice metrics are a valuable complement to physiological biomarkers, offering a non-invasive approach to detecting early cognitive declines and mitigating the risk of impairment through longitudinal monitoring. Study 3 investigated the relationship between prosody voice biomarkers, derived from spontaneous speech, and cognitive diagnostic assessments in clinical and unimpaired samples from the University of Pittsburgh Alzheimer’s Disease Research Center. In this sample, higher jitter, pulse, amplitude, pitch, and number of voice breaks were associated with concurrent and future cognitive impairment diagnoses over 3 years. Across the findings of Studies 2 and 3, high jitter was the most consistent predictor of cognitive declines and impairment, controlling for demographics and neurological disorders. Assessing prosody voice may facilitate the early identification of dementia risk, enabling timely intervention and treatment implementation. In summary, this dissertation highlights the potential of physiological and vocal biomarkers as early indicators of cognitive change, offering a practical framework for identifying at-risk individuals and supporting interventions to slow or prevent cognitive impairment.