Enhanced Efficient Diagnosis System for Parkinson's Disease using Hybrid Deep Neural Network

a neurological condition caused by the loss of dopamine-producing cells, “Parkinson's disease (PD)” “Parkinson's disease (PD)” transcends people over sixty years old. Current analytical methods, for instance, the combined “Unified Parkinson's disease rating Scale (UPDRS), magnetic resonance imaging (MRI),” and genetic testing, are emotional, asset-escalated, and wasteful. This paper emphasizes the significance of finding innovative ways to complete tasks. Through several deep learning algorithms, this study aims to measure the severity of “Parkinson's disease (PD)” using various patient condition data, including voice, imaging data from UCI, PPMI, and MRI scans, genetic information, and clinical evaluations. We at first look at the patient data using several data processing techniques to achieve this goal. The attributes are therefore entered into a -stage evaluation system. Using the “Oppositional Aquila Optimizer-based feature selection with Deep belief network for Parkinson's disease Detection (OAOFS-DBNPDD)” technique, the primary stage is to determine the most enlightening patient highlights. The next stage is adding the chosen highlights to the imaginary “artificial ecosystem Optimizer (AEO)” algorithm. This approach increases the hyper parameters of the “Deep belief network (DBN)” model, which has lately shown use for speculation. We also presented a few fundamental findings on our suggested arrangement and “deep learning (DL)” measurements, including the prevailing heartiness of a transformer-based organization to depth compared to a conventional MLP organization. We looked at the outcomes of the “OAOFSDBNPDD” in comparison to conventional machine learning and deep learning models, including “DCAE-MLP,” SVM, “Random forest (RF),” “DCNN-Destroyed,” and more models. Subjective, costly, and ineffective are PD diagnostics such UPDRS, MRI, and genetic testing. Many machine learning techniques do not understand and combine biological data kinds. A novel two-stage system—OAOFS-DBNPDD— combines deep learning with effective feature selection to increase diagnostic accuracy. A significant void in scalable, accurate, and automated PD detection is filled by hybrid optimization methods using multimodal data—speech, imaging, genetics, and clinical assessments. In view of the results of the model, our suggested “OAOFS-DBNPDD” exhibits a high accuracy of 98.29%.