Clinical Manifestations.

Area of Science:

• Neurology
• Neuroscience
• Artificial Intelligence in Medicine

Background:

• Vascular Cognitive Impairment (VCI) is a leading cause of dementia, often co-occurring with Alzheimer's Disease (AD) pathology.
• Accurate differentiation of MCI subtypes (Vascular, AD, Mixed) is crucial for effective diagnosis and treatment.
• Existing diagnostic methods can be complex, highlighting the need for accessible screening tools.

Purpose of the Study:

• To develop and validate digital games capable of detecting Mild Cognitive Impairment (MCI).
• To differentiate between subtypes of MCI, specifically Vascular, Alzheimer's, and Mixed types.
• To leverage machine learning for classifying MCI pathologies based on cognitive game performance.

Main Methods:

• Analysis of neuropsychological and cognitive data from 389 subjects (Cognitively Normal and MCI).
• Quantification of White Matter Hyperintensities (WMH) load via FLAIR imaging and amyloid levels via plasma amyloid beta 42/40 ratio.
• Development of 4 digital games to assess cognitive domains and machine learning models to differentiate MCI subtypes based on game features.

Main Results:

• A combined game score accurately detected MCI from Cognitively Normal (AUC=0.89) and differentiated Vascular from non-Vascular MCI (AUC=0.88).
• The system achieved high accuracy in distinguishing Mixed MCI (AUC=0.9) and Alzheimer's MCI (AUC=0.9) within their respective groups.
• Validation on 230 new subjects confirmed the digital games' efficacy in detecting and classifying MCI subtypes.

Conclusions:

• The ReCOGnAIze app demonstrates significant potential for early detection of MCI.
• The tool effectively differentiates between Vascular, Alzheimer's, and Mixed MCI subtypes.
• These findings support the use of the ReCOGnAIze app to guide clinical decisions and further diagnostic evaluations.