Recommender system is an information filtering tool used to alleviate information overload for users on the web. Collaborative filtering recommends items to users based on their historical rating information. There are two approaches: memory-based, which usually provides inaccurate but explainable recommendations; and model-based, whose recommendations are more precise but hard to understand. Here we propose a Bayesian model that not only provides us with recommendations as good as matrix factorization models, but these predictions can also be explained. The model is based on both user-based and item-based collaborative filtering approaches, which recommends items by using similar users' and items' information, respectively. Experiments carried out using four datasets present good results compared to several state-ofthe-art baselines, achieving the best performance using the Normalized Discounted Cumulative Gain (nDCG) quality measure and also improving the prediction's accuracy in some datasets.
In recent years, the use of recommender systems has become popular on the web. To improve recommendation performance, usage, and scalability, the research has evolved by producing several generations of recommender systems. There is much literature about it, although most proposals focus on traditional methods’ theories and applications. Recently, knowledge graph-based recommendations have attracted attention in academia and the industry because they can alleviate information sparsity and performance problems. We found only two studies that analyze the recommendation system’s role over graphs, but they focus on specific recommendation methods. This survey attempts to cover a broader analysis from a set of selected papers. In summary, the contributions of this paper are as follows: (1) we explore traditional and more recent developments of filtering methods for a recommender system, (2) we identify and analyze proposals related to knowledge graph-based recommender systems, (3) we present the most relevant contributions using an application domain, and (4) we outline future directions of research in the domain of recommender systems. As the main survey result, we found that the use of knowledge graphs for recommendations is an efficient way to leverage and connect a user’s and an item’s knowledge, thus providing more precise results for users.
Background
Diabetes is a public health problem worldwide. Although diabetes is a chronic and incurable disease, measures and treatments can be taken to control it and keep the patient stable. Diabetes has been the subject of extensive research, ranging from disease prevention to the use of technologies for its diagnosis and control. Health institutions obtain information required for the diagnosis of diabetes through various tests, and appropriate treatment is provided according to the diagnosis. These institutions have databases with large volumes of information that can be analyzed and used in different applications such as pattern discovery and outcome prediction, which can help health personnel in making decisions about treatments or determining the appropriate prescriptions for diabetes management.
Objective
The aim of this study was to develop a drug recommendation system for patients with diabetes based on collaborative filtering and clustering techniques as a complement to the treatments given by the treating doctor.
Methods
The data set used contains information from patients with diabetes available in the University of California Irvine Machine Learning Repository. Data mining techniques were applied for processing and analysis of the data set. Unsupervised learning techniques were used for dimensionality reduction and patient clustering. Drug predictions were obtained with a user-based collaborative filtering approach, which enabled creating a patient profile that can be compared with the profiles of other patients with similar characteristics. Finally, recommendations were made considering the identified patient groups. The performance of the system was evaluated using metrics to assess the quality of the groups and the quality of the predictions and recommendations.
Results
Principal component analysis to reduce the dimensionality of the data showed that eight components best explained the variability of the data. We identified six groups of patients using the clustering algorithm, which were evenly distributed. These groups were identified based on the available information of patients with diabetes, and then the variation between groups was examined to predict a suitable medication for a target patient. The recommender system achieved good results in the quality of predictions with a mean squared error metric of 0.51 and accuracy in the quality of recommendations of 0.61, which is acceptable.
Conclusions
This work presents a recommendation system that suggests medications according to drug information and the characteristics of patients with diabetes. Some aspects related to this disease were analyzed based on the data set used from patients with diabetes. The experimental results with clustering and prediction techniques were found to be acceptable for the recommendation process. This system can provide a novel perspective for health institutions that require technologies to support health care personnel in the management of diabetes treatment and control.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.