Diagnosis of Neurodegenerative Diseases by Gait Analysis using Triblock CNN and Deep RQA Techniques

Abstract

Neurodegenerative diseases are one of the most concerning medical disorders worldwide. Millions of global deaths and suffering could be averted through early diagnosis of neurodegenerative diseases necessitating the crucial need for technological innovations. Therefore, novel techniques are required urgently for this international humanitarian cause potential to saving numerous lives every day. With this in mind, a new technique based on Deep learning and Machine learning models has been presented. Neurodegenerative diseases (NDD) are caused by brain dysfunctions, triggering severe walking abnormalities. Human gait analysis distinguishes people using their walking style, which indicates the unique walking pattern of an individual. Personal examinations are vital in individual disease identifications, however intelligence-based systems are proficient in precise diagnosis of distinct Neurodegenerative Diseases than conventional approaches. The study data comprises gait patterns of Parkinson’s Disease, Huntington’s Disease, Amyotrophic Lateral Sclerosis and healthy control subjects. The proposed system recognizes gait patterns with utmost precision, distinguishing NDD using a remarkable gait analysis method. Hence, to extract optimal gait features ensuring accurate disease identification, this study proposes Triblock Convolutional Neural Network architecture and compact Deep Recurrence Quantification Analysis (TBCNN_DRQA) techniques. The Machine learning classifier helps in the categorization and disease diagnosis using feature vectors of TBCNN_DRQA. The TBCNN_DRQA technique with random forest classifier, achieves in identifying NDD with 99.96% higher accuracy. The results certain the precision and reliability of the proposed technique, thereby effectuating advanced identification of disease types thereby aiding doctors to start effective early treatment and rehabilitation.