Remote cognitive assessment approaches in the Dominantly Inherited Alzheimer Network (DIAN)

Abstract

AbstractBackgroundThe “one‐shot” paradigm used in traditional in‐clinic cognitive assessments often produces unreliable data due to daily fluctuations in stress, sleep, test anxiety, and fatigue. In DIAN, an international study of familial Alzheimer’s disease (AD), these difficulties are compounded by the scarcity and younger age of the population. We have developed two assessment strategies to address these issues, which we are also validating in the Knight ADRC cohort. The first is a smartphone‐based cognitive assessment platform based on principles from ecological momentary assessment that uses frequent but extremely short (<60 seconds) tests of episodic memory, attentional control, and processing speed administered at random times over 7 days. We examine average performance and variability metrics across 7 days of assessments. The second is a web‐based assessment that focuses on repeated daily administration to measure long‐term memory decay and also skill acquisition learning. The rates of forgetting and rates of learning become the outcome measures.MethodDigital assessments were completed on cognitively normal DIAN participants (n = 56 mutation carriers, n = 60 non‐carriers) and 120 mostly cognitively normal older adults from the Knight ADRC. Follow‐up evaluations were available for approximately 40 DIAN participants. Digital assessments were compared with standard in‐clinic cognitive tests and also with AD biomarkers including amyloid PET, cerebrospinal fluid (CSF) tau, tau PET, and structural MRI.ResultReliabilities of all remote cognitive measures ranged from 0.90‐0.98. All measures (Smartphone and web‐based) correlated with age (r’s 0.22‐0.69), indicating worse performance with increasing age. Correlations with standard in‐clinic measures ranged from 0.24‐0.77, suggesting that remote tests are valid measures of cognition. Test‐retest reliability from repeated smartphone assessments was exceptional for two smartphone measures (ICC’s = 0.89‐0.94). Critically, smartphone measures and web‐based measures demonstrated sensitivity to all AD biomarkers, including amyloid PET, tau PET, CSF tau, and structural MRI. We will present data on relationships with AD biomarkers and data on technical validation of smartphone cognitive measures and adherence strategies for remote studies.ConclusionRemote assessments that rely on short‐term but high frequency testing can be deployed worldwide in adults of all ages and demonstrate excellent reliability and validity compared to standard in‐clinic cognitive assessments.