Speech-based digital cognitive assessment for clinical trials: Detecting cognitive impairment stages and AD biomarker relations across European cohorts

Introduction: Early detection of Alzheimer's disease (AD) is critical for timely intervention as disease-modifying treatments emerge. Speech-based digital biomarkers offer scalable options for remotely capturing speech-derived functional changes associated with early cognitive decline, but validation across real-world populations remains limited. Methods: We evaluated the speech biomarker for cognition (SB-C), an automated speech-derived measure associated with cognitive status, in 736 participants across five European cohorts (Barcelonaβeta Brain Research Center's Alzheimer's at-risk cohort, European Prevention of Alzheimer's Dementia Scotland, Dementia Study of Cognitive and Biomarker Dynamics, Longitudinal Cognitive Impairment and Dementia Study, and Biomarkers for Identifying Neurodegenerative Disorders Early and Reliably [BioFINDER-Primary Care]). Participants completed verbal learning and semantic fluency tasks via automated phone or app-based platforms. SB-C performance was compared to Mini-Mental State Examination, Clinical Dementia Rating, Preclinical Alzheimer Cognitive Composite 5, and cerebrospinal fluid amyloid beta and phosphorylated tau181 biomarker status. Results: SB-C significantly differentiated cognitively unimpaired and impaired groups (P < 0.001), correlated with standard cognitive measures, and showed moderate-to-high area under the curve (0.56-0.82) for classifying biomarker positivity, with strongest results in BioFINDER-Primary Care. Discussion: SB-C is a scalable, remote speech-derived marker associated with cognitive status and AD biomarker group differences.