Noninvasive Electrophysiological Biomarkers of Olfactory Responses Across Cognitive States in Alzheimer Dementia: Cross-Sectional Study

Alzheimer disease (AD) is characterized by progressive cognitive decline, with olfactory dysfunction emerging among its earliest symptoms. Conventional olfactory tests rely on subjective self-reports and patient cooperation, limiting their clinical applicability. Noninvasive olfactory bulb (OB) recordings may provide objective physiological measures for early detection of cognitive decline.

This study investigated the relationship between olfactory dysfunction and cognitive decline across the AD spectrum and evaluated whether noninvasive OB recordings obtained via a wearable electroencephalography-based device could serve as digital biomarkers for early detection and home-based monitoring.

We recruited 71 participants, including individuals who were cognitively normal (CN; n=18), patients with mild cognitive impairment (MCI) subdivided into early MCI (n=30) and late MCI (n=9), and patients with dementia of the Alzheimer type (n=14). OB activity was recorded noninvasively during odor stimulation using a 6-channel wearable electroencephalography system. Behavioral olfactory function was assessed using a culturally adapted 6-item modified Brief Smell Identification Test and compared with electrophysiological responses. Time-frequency analyses identified disease-related components via nonparametric permutation testing (1000 iterations; P<.05). A support vector machine classifier was applied for group discrimination. Structural brain integrity was assessed through diffusion tensor imaging and magnetic resonance imaging, and cognitive performance was evaluated using comprehensive neuropsychological testing. Beta-band power (reflecting top-down cortical feedback) and gamma-band power (representing OB activity) were analyzed to characterize neural communication patterns.

Beta-band power progressively declined from CN to dementia of the Alzheimer type, while gamma-band power decreased and response latency increased beginning in late MCI. Behavioral olfactory testing failed to differentiate CN from early MCI (P=.77), whereas electrophysiological recordings revealed significant differences (P=.02). The support vector machine classifier achieved 83.1% accuracy in distinguishing all groups based on olfactory spectral features. Reduced beta-band power showed a trend-level association with cingulum-hippocampal tract diffusivity (R²=0.28; P=.061), whereas gamma-band power correlated with OB volume in later disease stages (R²=0.21; P<.001). Electrophysiological features explained 91.7% of the variance in cognitive scores (Seoul Neuropsychological Screening Battery II; P<.001). The brief 5 to 10-minute testing protocol, requiring less than 2 minutes to fit the device, supports its feasibility for repeated, home-based assessments.

Bidirectional interactions between the OB and higher-order brain regions appear altered early in the AD continuum, with beta-band changes emerging first and gamma-band alterations occurring later. Digital OB recordings offer objective, noninvasive biomarkers for detecting cognitive decline before overt structural changes. Their portability and short duration make this approach ideal for longitudinal, home-based cognitive assessments within mobile health frameworks and for evaluating therapeutic efficacy in future interventional studies.