Revolutionizing Parkinson’s Disease Diagnosis: An Advanced Data Science and Machine Learning Architecture

The presence of speech impairment during the early stages of Parkinson’s disease has motivated several experts to try to predict whether a patient has the disease by applying various techniques based on Machine Learning. Currently, there is no adequate technique for that process. This paper proposes an analysis using three Learning Models (Support Vector Machine, Random Forest, and Multi-Layer Perceptron Neural Network with and without dimensionality reduction executing the Partial Least-Squares Discriminant Analysis). To classify healthy patients from diseased ones by applying quality measures to compare the results obtained with those acquired in related research. The dataset used also contains the Q-factor wavelet transform of each sample to increase the accuracy of the models (0 to negative and 1 to positive for this disease). This research gives way to future works, which will be in charge of improving the values achieved by optimizing the execution times using more advanced techniques.