A Contextual Wellness Recommender System

Since their introduction, personalized recommendation systems have experienced remarkable evolution, aimingto provide recommendations corresponding to the needs and preferences of users, particularly in the field of healthcare. However, these systems have encountered limitations, particularly regarding the dynamic nature of users health needs. Contextual recommender systems take this dynamic into account and use users contextual data to generate personalized recommendations tailored to their needs. In this context ,Internet of Things (IoT) technologies assume a pivotal role, enabling remote monitoring of various health aspects, facilitating proactive healthcare interventions, and crafting personalized treatment plans. The data collected by IoT devices serves as a valuable resource, enhancing the effectiveness of personalized recommendations.To address this challenge, we propose a novel approach named “Contextual Wellness Recommender System” ,a methodology that fully exploits the data collected by several connected health devices. Our methodology relies on the use of advanced machine learning and data analysis techniques to intelligently integrate contextual information into the recommendation process.Using machine learning algorithms, we will train two models to recognize patterns and correlations in IoT sensor data and other contextual factors. The used data includes users physical activity,demographics, lifestyle habits and vital signs. By taking into account these multiple dimensions of data, our models will be able to generate personalized recommendations that will allow users to proactively take care of their health.With an accuracy of over 90\% on Model\_A and more than 80\% on Model\_B on both training and validation data,our proposed approach stands out for its use of several and diverse connected health devices to generaterecommendations. This innovative approach increases efficiency, personalization and adaptability by adapting to different individuals health conditions and fully exploiting their contextual data.