Multimodal Learning and Intelligent Prediction of Symptom Development in Individual Parkinson's Patients

Area of Science:

• Neuroscience
• Computational Neuroscience
• Biomarkers

Background:

• Neurodegenerative diseases (ND) impact brain function through cell death and compensatory learning, often detected late.
• Current diagnostic methods for ND, like Parkinson's disease (PD), have limitations in early detection and understanding structural effects.
• Early diagnosis is vital for timely monitoring and therapeutic intervention in neurodegenerative conditions.

Purpose of the Study:

• To evaluate the predictive power of reflexive saccade (RS) parameters and patient attributes for diagnosing neurological symptoms in Parkinson's disease.
• To compare the accuracy of measurement-based diagnoses with standard neurological assessments.
• To explore the utility of rough set theory and other machine learning classifiers for PD diagnosis.

Main Methods:

• Measured latency, amplitude, and duration of reflexive saccades (RS) in Parkinson's disease patients.
• Utilized patient age, RS parameters, and 'well-being' scores as condition attributes.
• Applied rough set theory and compared its predictive performance with Naïve Bayes, Decision Trees/Tables, and Random Forests (KNIME/WEKA).

Main Results:

• Patient age and RS parameters achieved an 80% accuracy in predicting neurological symptoms.
• Incorporating 'well-being' scores increased prediction accuracy to 90%.
• Reflexive saccades (RS) were identified as potent biomarkers for assessing symptom progression in PD.

Conclusions:

• Reflexive saccade (RS) measurements offer a valuable, objective tool for early Parkinson's disease diagnosis and monitoring.
• Machine learning algorithms, particularly rough set theory, show significant promise in enhancing diagnostic accuracy for neurodegenerative diseases.
• Integrating diverse patient data, including objective measurements and subjective well-being, can substantially improve predictive models for neurological conditions.