VR-FlexAI: A Multimodal Approach for Assessing Cognitive Flexibility in Aging

Cognitive flexibility is a key component of executive functioning and is particularly vulnerable to age-related cognitive changes and mild cognitive impairment. However, conventional neuropsychological assessments often rely on structured, decontextualized tasks that may fail to capture the adaptive demands of everyday life. The VR-FlexAI project addresses this limitation by proposing a multimodal, process-oriented approach to the assessment of cognitive flexibility in aging. The project develops ACTA-VR, an immersive virtual reality environment designed to evaluate adaptive behavior during realistic daily tasks characterized by interruptions, implicit rules, competing goals, and incomplete information. Cognitive flexibility is operationalized through task switching and set shifting, assessed respectively through the Adaptive Savory Breakfast task and the Shifting Food Categorization task. Behavioral digital biomarkers extracted from the virtual environment are integrated with verbal digital biomarkers derived from speech and language analysis, including semantic organization, speech rate, pauses, and prosodic features. These data are combined with cognitive measures from neuropsychological testing and analyzed using machine learning and natural language processing techniques. The project includes an initial usability and feasibility phase with cognitively healthy older adults, followed by a preliminary validation study involving healthy participants, individuals with suspected executive difficulties, and patients with mild cognitive impairment. By integrating behavioral, linguistic, and cognitive indicators, VR-FlexAI aims to provide a more ecologically valid and clinically informative assessment of cognitive flexibility, supporting early detection, individualized profiling, and longitudinal monitoring of cognitive decline.