Machine Learning and Eye Movements Give Insights into Neurodegenerative Disease Mechanisms

Area of Science:

• Neuroscience
• Ophthalmology
• Computational Biology

Background:

• Human perception is heavily reliant on vision and eye movements (EMs).
• Neurodegenerative disorders (ND), including Alzheimer's (AD) and Parkinson's diseases (PD), are known to affect cognitive and motor functions.
• Alterations in EMs are observed in ND, potentially explaining associated symptoms.

Purpose of the Study:

• To determine if changes in EM-evoked responses can predict the progression of ND like AD and PD.
• To investigate the utility of EMs as biomarkers for assessing symptom progression and identifying preclinical stages of ND.

Main Methods:

• Review and analysis of psychological, neurological, and EM (saccades, antisaccades, pursuit) test results.
• Application of machine learning (ML) algorithms (e.g., granular computing, Naive Bayes, Decision Trees, logistic regression, SVC, KNC, Random Forest) to predict disease progression.
• Investigation of EM experiments in virtual environments to detect neurodegeneration-related spatial navigation deficits.

Main Results:

• ML algorithms identified significant EM changes correlated with ND symptoms, providing insights into disease mechanisms.
• EMs were confirmed as robust biomarkers for assessing symptom progression in PD and AD.
• Navigation difficulties in 3D space, assessed via virtual EM experiments, were linked to neurodegeneration.

Conclusions:

• EM parameters, combined with clinical symptoms and ML, serve as powerful tools for precision medicine in ND.
• EM analysis holds significant potential for predicting ND progression and detecting preclinical stages of AD and PD.