Machine learning to predict impulse control disorders in Parkinson's disease

Impulse control disorders are a class of psychiatric disorders characterized by impulsivity. These disorders are common during the course of Parkinson's disease, decrease the quality of life of subjects, and increase caregiver burden. Being able to predict which individuals are at higher risk of developing these disorders and when is of high importance. The objective of this thesis is to study impulse control disorders in Parkinson's disease from the statistical and machine learning points of view, and can be divided into two parts. The first part consists in investigating the predictive performance of the altogether factors associated with these disorders in the literature. The second part consists in studying the association and the usefulness of other factors, in particular genetic data, to improve the predictive performance. In the first chapter, we present Parkinson's disease and impulse control disorders, review the literature on impulse control disorders in Parkinson's disease, introduce the main concepts of machine learning, and describe the databases from which we obtained data and the software used to analyze these data. In the second chapter, we investigate the predictive performance of several machine learning algorithms using features that have been associated with impulse control disorders in Parkinson's disease. In the third chapter, we investigate the association between impulse control disorders in Parkinson's disease and genetic risk scores for a broad range of phenotypes. In the last chapter, we investigate different approaches to integrate static data in recurrent neural networks and evaluate their predictive performance in the use case of predicting impulse control disorders in Parkinson's disease, with genetic data used as static data. Across these works, we highlight the importance of using machine learning algorithms, cross-validation and replication cohorts to unbiasedly estimate the predictive power of known and putative risk factors of impulse control disorders in Parkinson's disease.