Sexuaren eragina Parkinsonen gaixotasunaren lehenengo faseetan, adierazpen ez-motorretan eta ikasketa automatikoan oinarrituta

Abstract

According to clinical publications, non-motor symptoms of Parkinson’s disease (PD) present unequal evolutions for different sexes, but we did not find works based on machine learning that analyze whether it can affect the early stages of the disease. In this study we have analyzed the influence of sex on the classification of patients with PD and controls using machine learning algorithms with non-motor symptoms. For this purpose, the PPMI database and three machine learning algorithms (XGBoost, MLP and SVM) were used, achieving an accuracy rate of over 80%. Then, using SHAP values, it was concluded that sex has little influence on classification. Finally, an attempt has been made to classify sex from the patient type information. The results show that in the database we have used, non-motor symptoms do not vary significantly according to sex.