Dysarthria Speech Detection Using Convolutional Neural Networks with Gated Recurrent Unit.

Area of Science:

• Neurology
• Artificial Intelligence
• Speech Pathology

Background:

• Increasing prevalence of neurological diseases like Parkinson's, stroke, and cerebral palsy due to population growth and aging.
• Dysarthria is a common symptom in these neurological conditions, significantly impacting disease management and patient quality of life.
• Timely detection and intervention for dysarthria are critical to prevent symptom exacerbation and psychological/physiological distress.

Purpose of the Study:

• To propose an advanced deep learning model for the accurate detection of dysarthria.
• To evaluate the efficacy of an integrated Convolutional Neural Network (CNN) and Gated Recurrent Unit (GRU) model for dysarthria detection.
• To compare the performance of the proposed CNN-GRU model against existing research models.

Main Methods:

• Development of an integrated deep learning model combining Convolutional Neural Networks (CNNs) and Gated Recurrent Units (GRUs).
• Utilizing the CNN-GRU architecture for the classification and detection of dysarthria.
• Comparative analysis of the proposed model's accuracy against other machine learning and deep learning approaches.

Main Results:

• The proposed integrated CNN-GRU model achieved a high accuracy rate of 98.38% in dysarthria detection.
• The CNN-GRU model demonstrated superior performance compared to other research models evaluated in the study.
• The findings indicate the model's effectiveness in identifying dysarthria with remarkable precision.

Conclusions:

• The integrated CNN-GRU model presents a highly accurate and effective solution for dysarthria detection.
• This advanced deep learning approach offers significant potential for improving early diagnosis and management of neurological conditions.
• The study highlights the capability of hybrid deep learning models in addressing complex speech disorder detection challenges.