Random Forest Algorithm Based on Speech for Early Identification of Parkinson's Disease.

Area of Science:

• Neurology
• Speech Science
• Machine Learning

Background:

• Parkinson's disease (PD) diagnosis can be challenging in early stages.
• Speech impairments are common early symptoms of PD.
• Objective diagnostic tools are needed to aid clinical assessment.

Purpose of the Study:

• To evaluate the effectiveness of using acoustic speech features and random forest (RF) classification for early PD identification.
• To compare the diagnostic accuracy of the RF algorithm with that of experienced neurologists.

Main Methods:

• Extraction of various acoustic parameters (prosodic and segmental features) from speech signals.
• Application of the random forest (RF) classification algorithm to identify early-stage PD patients.
• Comparison of RF algorithm accuracy against neurologist judgments based on auditory tests.

Main Results:

• The RF algorithm demonstrated superior accuracy in identifying early-stage PD patients compared to neurologists.
• Speech signal analysis using RF provides a highly accurate method for PD detection.
• The proposed method offers a significant improvement over traditional diagnostic approaches.

Conclusions:

• The random forest algorithm applied to speech analysis is a highly effective and accurate method for the early diagnosis of Parkinson's disease.
• This approach serves as an efficient auxiliary tool, enhancing diagnostic capabilities in clinical settings.
• Speech-based analysis holds significant potential for non-invasive and early detection of PD.