Genome–wide association studies (GWAS) have revealed a plethora of putative susceptibility genes for Alzheimer’s disease (AD), with the sole exception of APOE gene unequivocally validated in independent study. Considering that the etiology of complex diseases like AD could depend on functional multiple genes interaction network, here we proposed an alternative GWAS analysis strategy based on (i) multivariate methods and on a (ii) telescope approach, in order to guarantee the identification of correlated variables, and reveal their connections at three biological connected levels. Specifically as multivariate methods, we employed two machine learning algorithms and a genetic association test and we considered SNPs, Genes and Pathways features in the analysis of two public GWAS dataset (ADNI-1 and ADNI-2). For each dataset and for each feature we addressed two binary classifications tasks: cases vs. controls and the low vs. high risk of developing AD considering the allelic status of APOEe4. This complex strategy allowed the identification of SNPs, genes and pathways lists statistically robust and meaningful from the biological viewpoint. Among the results, we confirm the involvement of TOMM40 gene in AD and we propose GRM7 as a novel gene significantly associated with AD.
Personalization of playlists is a common feature in music streaming services, but conventional techniques, such as collaborative filtering, rely on explicit assumptions regarding content quality to learn how to make recommendations. Such assumptions often result in misalignment between offline model objectives and online user satisfaction metrics. In this paper, we present a reinforcement learning framework that solves for such limitations by directly optimizing for user satisfaction metrics via the use of a simulated playlist-generation environment. Using this simulator we develop and train a modified Deep Q-Network, the action head DQN (AH-DQN), in a manner that addresses the challenges imposed by the large state and action space of our RL formulation. The resulting policy is capable of making recommendations from large and dynamic sets of candidate items with the expectation of maximizing consumption metrics. We analyze and evaluate agents offline via simulations that use environment models trained on both public and proprietary streaming datasets. We show how these agents lead to better user-satisfaction metrics compared to baseline methods during online A/B tests. Finally, we demonstrate that performance assessments produced from our simulator are strongly correlated with observed online metric results.