A Novel Method for Parkinson's Disease Diagnosis Utilizing Treatment Protocols

Area of Science:

• Neurology
• Artificial Intelligence
• Biomedical Engineering

Background:

• Neurodegenerative disorders like Parkinson's disease (PD) affect individuals regardless of sex.
• Accurate and timely diagnosis of PD is crucial for effective management.
• Traditional diagnostic methods can be time-consuming and costly.

Purpose of the Study:

• To develop a deep learning model for predicting Parkinson's disease severity.
• To leverage voice analysis for non-invasive and efficient PD diagnosis.
• To provide medical decision support for physicians using AI-driven insights.

Main Methods:

• A dataset of voice recordings from 253 participants was analyzed.
• Data preprocessing and methodical sampling were employed for data balancing.
• Feature selection based on label influence was used to create data groups.
• Classification algorithms including Decision Trees (DT), Support Vector Machines (SVM), and k-Nearest Neighbors (kNN) were evaluated.
• The Support Vector Machine (SVM) technique was selected for model development, utilizing 45% of the data.

Main Results:

• The developed deep learning model achieved an 86% performance accuracy.
• The model demonstrated exceptional outcomes across various project aspects.
• Voice analysis proved effective in supporting Parkinson's disease diagnosis.

Conclusions:

• The study successfully developed an AI model for Parkinson's disease severity prediction.
• The model offers significant potential for medical decision support in clinical practice.
• Voice data analysis presents a promising, cost-effective approach for PD diagnosis.