Physical fitness recommender framework for thyroid patients using restricted boltzmann machines

Abstract

These days, people can easily acquire the information from online sources. Individuals are generally using recommendation services before buying products considering the availability of online. Recommendation systems propose the relevant services or products to users. But sometimes people face issues while retrieving health related information from the recommender systems. A focus on keeping people healthy is one way to address the serious societal concern of healthcare domain. A health-based physical recommender system suggests workout plans for users using their activity level and health condition. A personalized approach is the most effective solution for the fitnessbased recommender framework based on user's desired characteristics. This article presents a personalized fitness recommender system for thyroid patients. The proposed fitness recommender model integrates the user's data like personal and health profile, preferences, calorie intake, and activity level. The proposed hybrid model is built using Restricted Boltzmann Machines (RBM) integrating content based and matrix factorization techniques. The results of experiments prove that the proposed hybrid model outperforms than content based, pure RBM and matrix factorization recommendation techniques. The current proposal achieves the personalization approach by incorporating user's thyroid health condition and exercise preferences in recommendation process. The recommended result of hybrid RBM method is revised based on user's new preferences.