Date Published: June 6, 2019
Publisher: Public Library of Science
Author(s): Suwen Lin, Louis Faust, Pablo Robles-Granda, Tomasz Kajdanowicz, Nitesh V. Chawla, Rodrigo Huerta-Quintanilla.
Social networks influence health-related behavior, such as obesity and smoking. While researchers have studied social networks as a driver for diffusion of influences and behavior, it is less understood how the structure or topology of the network, in itself, impacts an individual’s health behavior and wellness state. In this paper, we investigate whether the structure or topology of a social network offers additional insight and predictability on an individual’s health and wellness. We develop a method called the Network-Driven health predictor (NetCARE) that leverages features representative of social network structure. Using a large longitudinal data set of students enrolled in the NetHealth study at the University of Notre Dame, we show that the NetCARE method improves the overall prediction performance over the baseline models—that use demographics and physical attributes—by 38%, 65%, 55%, and 54% for the wellness states—stress, happiness, positive attitude, and self-assessed health—considered in this paper.
Social networks play an important role in the diffusion of behavior, attitudes, tastes, and beliefs. Several studies have shown that such characteristics leverage the existing social connections and ties for diffusion. This phenomenon is demonstrative of the similarity or homophily between the nodes in the network (ego and alter, for example) and also of the social influences that affect people. Some examples of this diffusion process include: the spread mechanism of diverse health conditions over social networks—such as obesity  and smoking , the effect of social network on personal psychological traits—such as affection  and happiness , and the spread of health behavior through social networks . People’s interactions through social networks or social media platforms have also been used to discover aspects of emotions experienced by individuals , mental illness [7, 8], and activity patterns . Different social network types, such as friendship or non-friendship networks, can also provide insights about mental health in adults .
Based on the discussed methods and framework, we performed two sets of experiments. First, we evaluated the interactions among the variables associated with social network structure and those related to health behavior. The objective of this analysis was to validate whether their interactions were meaningful. Second, we used our framework to predict various wellness states. We compared the performance of our framework with two baselines. One of them applies our framework to either health-behavior data or network features in isolation and the other one comes from the random generation. The objective was to verify our hypothesis that combining network effects and self-similarity would lead to better predictions.
The main contributions of this paper can be summarized as follows: