Automatic Detection of Cognitive Impairment with Virtual Reality

Area of Science:

• Neuroscience
• Psychiatry
• Computer Science

Background:

• Cognitive impairment significantly impacts daily functioning and safety in neuropsychiatric conditions.
• Early diagnosis and intervention are crucial for managing these conditions.
• Virtual Reality (VR) presents a novel technological approach for cognitive assessment.

Purpose of the Study:

• To introduce and validate new VR-derived measures for assessing cognitive performance.
• To demonstrate the correlation between these VR measures and established clinical cognitive assessments.
• To explore the utility of VR in identifying cognitive deficits in individuals with psychosis.

Main Methods:

• An immersive VR environment (VStore) was utilized for a simulated supermarket shopping task.
• Spatio-temporal behavioral data were collected from participants (psychosis patients and controls).
• Four key features (route optimality, proportional distance, execution error, hesitation) were extracted using mathematical models and mapped to validated cognitive scores via linear regression.

Main Results:

• The hesitation score emerged as the most statistically significant VR-derived feature.
• A multiple linear regression model combining all VR features yielded significant results (R2 = 0.369, F-Stat = 7.158, p = 0.000128).
• These findings indicate a strong correspondence between VR-derived behavioral metrics and clinical cognitive assessments.

Conclusions:

• VR-based behavioral analysis, particularly hesitation scores, can effectively measure cognitive performance.
• This VR approach shows potential for the recognition and diagnosis of cognitive impairment in neuropsychiatric disorders.
• VR offers a promising, objective tool for enhancing cognitive assessment in clinical settings.