Proposing a new approach based on convolutional neural networks and random forest for the diagnosis of Parkinson's disease from speech signals

•A new architecture was proposed for the detection of Parkinson's disease from speech signals with the name SkipConNet + RF based on Convolutional Neural Networks (CNNs) and Random Forest (RF).•In this study, in which two different data sets were used, 99.11% accuracy was obtained with the proposed method.•With the SkipConNet + RF method, an improvement between 3% and 17.19% was achieved in the performance of the RF algorithm.