Healthcare Embedded System for Predicting Parkinson's Disease Based on AI of Things

This study aims to build a healthcare-embedded system based on the Artificial Intelligence of Things (AIoT) to predict people affected by Parkinson's Disease (PD) via vocal disorders either locally or remotely. This system has been created based on several techniques of artificial intelligence such as dataset analysis, selection of classifiers and searching for the optimal hyperparameters that provide good performance without affecting the classifier's performance and the speed of classification. Also., this system is designed to be able to record audio, analyze sound signals and extract features based on several signal analysis libraries and artificial intelligence algorithms. We employed many intermediate libraries between the system, servers and the user to facilitate utilization of the system, to make it able to work anytime and anywhere in the world and to provide insight into the contribution of embedded systems based on AIoT for enhancing the quality of medical services. Furthermore, performance evaluation techniques have been used to estimate the accuracy., sensitivity., precision and area under the curve for several classifiers, and we selected the best classifiers to create an ensemble voting classifier that is the Kernel of our embedded system and it has an accuracy of 95.3%.