id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
cord-193856-6vs16mq3	Zhou, Tongxin	Spoiled for Choice? Personalized Recommendation for Healthcare Decisions: A Multi-Armed Bandit Approach	2020-09-13		.txt	text/plain	12295	647	41	The first component is a deep-learning-based feature engineering procedure, which is designed to learn crucial recommendation contexts in regard to users' sequential health histories, health-management experiences, preferences, and intrinsic attributes of healthcare interventions. Our evaluation results suggest that each of our proposed model components is effective and that our recommendation framework significantly outperforms a wide range of benchmark models, including UCB, e -greedy, and state-of-the-art conventional recommendation systems, such as context-aware collaborative filtering (CACF), probabilistic matrix factorization (PMF), and content-based filtering (CB). These research gaps motivate us to propose an online-learning scheme, i.e., multi-armed bandit (MAB), to address the dynamics and diversity in individuals' health behaviors to improve healthcare recommendations. To better adapt an MAB to the healthcare recommendation setting, we then further enhance our framework by synthesizing two model components, that is, deep-learning-based feature engineering and a diversity constraint. To improve the characterization of individuals' health-management contexts and enhance recommendation personalization, we design a deep-learning model to construct user embeddings.	./cache/cord-193856-6vs16mq3.txt	./txt/cord-193856-6vs16mq3.txt