Alternate fluency in Parkinson's disease: A machine learning analysis.

Area of Science:

• Neuroscience
• Cognitive Psychology
• Computational Neuroscience

Background:

• Parkinson's Disease (PD) is associated with executive function deficits.
• Assessing these deficits often involves complex cognitive tests.

Purpose of the Study:

• To investigate performance changes in extra-dimensional shifting in PD patients.
• To implement a novel analysis using alternate phonemic/semantic fluency tests.
• To develop machine learning models for classifying PD patients based on fluency test performance.

Main Methods:

• Utilized a novel alternate phonemic/semantic fluency test.
• Applied machine learning (ML) for high-accuracy classification of PD patients.
• Evaluated classification accuracy based on semantic fluency, phonemic fluency, and a shifting index.

Main Results:

• ML models achieved classification accuracy between 80% and 90%.
• Semantic fluency test was the most efficient predictor (86.96% accuracy).
• Phonemic fluency test and shifting index also showed significant classification accuracy (80.43% and 83.69%, respectively).

Conclusions:

• ML analysis of verbal fluency effectively identifies executive deficits in PD.
• The findings suggest a reliable method for detecting cognitive changes in PD patients.
• This approach offers good out-of-sample generalization for clinical application.