AI-assisted modeling of attentional control for intervention engagement
Cognitive training is one of the most used cognitive interventions, aiming for preventing or slowing cognitive decline in older adults. However, its effect is inconsistent, often due to the lack of effective engagement. Here we developed a novel AI-assisted multimodal framework, called Attentional Control Index using Digital measures (AID), which combines video-based facial expression analysis with ECG-derived heart rate variability (HRV) to assess the availability and allocation of attentional resources during cognitive training. Using two independent datasets of older adults at elevated risk for cognitive impairment, we demonstrated that AID accurately predicts both behavioral markers of engagement and cognitive outcomes from cognitive training. Additionally, we identified interpretable facial expressions and HRV features that reflect attentional control. These findings establish a robust, generalizable digital marker for assessing, and potentially in the future, enhancing effective engagement in cognitive training by focusing on the real-time dynamics of individual users' attentional control.