Weight Loss Prediction Using Social Network Analysis (BOOHEE)
In recent years, there have been many social networking applications aimed at helping users lose weight and providing outlets to interact with others also trying to manage their weight. The social networking aspect of these applications allows users to connect with other users’ profiles. As the users track their individual progress towards their weight loss goals, they can share their achievements and view the progress of users they are connected with on the platform. Our group conducted research into the possibility of the social aspects of weight loss applications creating contagion in weight loss. While numerous studies have shown how obesity and weight gain can spread from social connections, we wanted to explore how a weight loss application could have the opposite effect. We used data from the BOOHEE weight management operator and compared user interactions through the NetworkX library in Python. We used feature extraction to merge social activities between users to ultimately create graph networks for testing and evaluating our hypothesis. Applying graph convolutional network algorithms and training them with five different batches of dataset values allowed us to receive a maximum weight loss prediction of 71.8% accuracy. These findings allowed us to confidently show that weight loss applications like BOOHEE and the social interactions that take place on the platform can impact a user’s weight loss.