BACKGROUND: We developed TalkTracker, a mobile app to enable at-home digital speech assessments. We examined whether self-administered diadochokinetic (DDK) tasks detect subtle cognitive decline in Alzheimer’s risk-enriched cohorts.

METHODS: Eighty-five adults from the WRAP and WADRC (cognitively unimpaired-stable [CUS]=52; subtle-to-mild cognitive impairment [CI]=33; age=68.59±6.44 years) completed weekly DDK (“puh-tuh-kuh” repetition) recordings over 7–8 sessions using TalkTracker. Automated acoustic analysis quantified syllable rate and rhythmicity. Intra-individual variability was measured as standard deviation and coefficient of variation of syllable rate across sessions. Models adjusted for age, sex, education, and plasma p-tau217.

RESULTS: CI individuals produced slower rates and higher intra-individual variability than CUS. Regardless of group, syllable rates increased across sessions, suggesting a robust practice effect.

DISCUSSION: Mobile DDK assessment captured subtle early impairments through slower articulation and greater session-by-session variability, demonstrating the feasibility and sensitivity of at-home speech digital markers for early screening and repeated monitoring of Alzheimer’s-related cognitive change.