Development of Healthcare Model Using AR-DVAE with Mayfly-MLP-BPN for Parkinson's Disease Detection

Research on reliable biomarkers that permit the creation of medical-decision support technologies is currently a major focus in healthcare biometrics. These gadgets aid in the analysis and nursing of conditions like Parkinson's disease (PD). If PD patients' vocal alterations could be detected, this would allow for early diagnosis and treatment, thereby preventing the onset of physically disabling symptoms. More recent research has centred on the suggestion of new automatic and objective tools to aid in diagnosis and severity assessment. In this study, we provide a dual approach to detecting PD. The first section extracts features mechanically with the help of an AR-DVAE model (autoregressive deep variational autoencoder). Lastly, the Mayfly Algorithm is used to maximise the bias weight in a modified multilayer perceptron utilising a backpropagation network (MLP-BPN). To conduct this research, we used the publicly available PD Using Digitized Graphics Tablet dataset. In comparison to other models, the suggested model performed better across the board, as shown by the validation analysis.