Predicting Age From Behavioral Test Performance for Screening Early Onset of Cognitive Decline

Area of Science:

• Neuroscience
• Cognitive Psychology
• Biostatistics

Background:

• Aging is associated with neuronal slowing and cognitive changes, varying individually.
• Assessing cognitive changes across the lifespan is complex, often requiring multiple tests.
• Current methods have limitations in comprehensively evaluating cognitive domains.

Purpose of the Study:

• To improve early diagnosis of cognitive decline by estimating its onset in healthy individuals.
• To predict an individual's age group using behavioral and psychophysiological tests.
• To compare predicted cognitive age with chronological age for accelerated aging detection.

Main Methods:

• Utilized publicly available datasets (POBA, SSCT).
• Employed Pearson correlation, Kruskal-Wallis test, clustering, feature selection, and classification methods.
• Assessed relationships between age and test results, identified cognitive decline onset, and predicted age groups.

Main Results:

• Psychophysiological test results showed a U-shape across the lifespan, peaking after 35 years and declining post-55-60.
• Cognitive test performance variance showed a linear age-related trend from adolescence.
• Classification models achieved high performance in predicting subject age groups.

Conclusions:

• Machine learning (ML) models show promise as computer-aided detectors for neurocognitive decline.
• Classification models can effectively predict age-related cognitive changes.
• Combining multiple cognitive and psychophysiological tests may enhance ML model accuracy for reliable detection.