Advancing Parkinson's Disease Detection and Classification Through Integrated Deep Learning Models

Parkinson's disease (PD) is a neurodegenerative condition that severely impacts motor and non-motor functions, affecting millions worldwide. Early and accurate detection is crucial for effective intervention and improved patient outcomes. This study introduces a novel method for PD recognition and classification that integrates advanced Deep Learning (DL) techniques, specifically Convolutional Gated Recurrent Units (CGRUs), with Support Vector Classification (SVC) algorithms. The CGRUs are employed to analyse complex speech patterns, capturing the temporal interdependencies that reflect Parkinsonian gait characteristics. This comprehensive approach involves curating a diverse array of speech patterns to enhance the diagnostic process. Additionally, the SVC algorithm refines classification accuracy by differentiating between PD and non-PD instances based on features extracted by the CGRU. Our approach, utilizing 16 distinct DL and SVC models, aims to advance PD diagnosis through precise and early detection, demonstrating significant improvements in diagnostic accuracy and classification efficacy.