Deep Learning for Parkinson's Disease Diagnosis: A Graph Neural Network (GNN) Based Classification Approach with Graph Wavelet Transform (GWT) Using Protein-Peptide Datasets | JoVE Visualize

Area of Science:

Neurology
Machine Learning
Data Science

Background:

Parkinson's Disease (PD) is a neurological disorder with significant motor and non-motor symptoms.
Current PD assessment relies on the Movement Disorder Society Unified Parkinson's Rating Scale part III (MDS-UPDRS-III), but its measurement instability hinders accurate prediction and tracking.
This limitation necessitates advanced methods for reliable PD severity assessment.

Purpose of the Study:

To develop an improved method for predicting Parkinson's Disease severity using Graph Wavelet Transform (GWT) and Graph Neural Network (GNN).
To enhance the accuracy and reliability of PD prediction and tracking, overcoming the limitations of current assessment scales.
To facilitate the development of individualized treatment plans for PD patients through more precise data analysis.

Main Methods:

Proposed a novel approach combining Graph Wavelet Transform (GWT) for weighted feature extraction with Graph Neural Network (GNN) for classification.
Utilized GWT to calculate weighted correlations within patient data, enhancing predictive patterns.
Trained machine learning algorithms, specifically GNN, to predict the MDS-UPDRS-III score for PD tremors based on extracted features.

Main Results:

The proposed GWT-GNN model achieved a Mean Squared Error (MSE) of 0.1796 and a Root Mean Squared Error (RMSE) of 0.2845 in predicting PD severity.
Demonstrated significant improvements in prediction accuracy compared to state-of-the-art methods: 27.66% over DNN, 54.11% over ANFIS + SVR, and 0.71% over Mixed MLP.
The model effectively predicts motor and MDS-UPDRS scores, indicating its potential for clinical application.

Conclusions:

The developed GWT-GNN strategy proves highly effective for predicting Parkinson's Disease severity.
This approach offers a more reliable method for PD assessment, potentially improving patient outcomes.
The findings support the use of advanced machine learning techniques in neurological disorder research and personalized medicine.