Identificación de riesgo de progresión de la enfermedad de Parkinson utilizando Machine Learning

Background: Parkinson’s Disease (PD) affects the lives of patients in different ways, who initially do not know what the progression of the disease will be like in the coming years and the changes it will imply in their life plans. The objective of this study is to implement a Parkinson’s progression risk prediction model using autonomous learning techniques. Methodology: 11 variables are selected that are used in other related studies and are taken from clinical data of patients with PD captured through evaluations and questionnaires. Prediction models are implemented using Machine Learning techniques: Logistic Regression, Random Forest and Gradient Boosting Machine (GBM). Performances are evaluated through the area under the Receiver Operating Characteristic (ROC) curve. Results: A set of data with quality and according to the needs of the case study is obtained using the PPMI database with which a robust model of risk of PD progression is implemented over a period of 10 years, with an area under the curve (AUC) of 0.9, interpretable, reliable and with high usability for patients, caregivers and qualified personnel who will be able to make more accurate decisions from the data.