RESUMO
BACKGROUND: Social network interventions are an effective approach to promote physical activity. These interventions are traditionally designed using self-reported peer nomination network data to represent social connections. However, there is unexplored potential in communication data exchanged through web-based messaging apps or social platforms, given the availability of these data, the developments in artificial intelligence to analyze these data, and the shift of personal communication to the web sphere. The implications of using web-based versus offline social networks on the effectiveness of social network interventions remain largely unexplored. OBJECTIVE: This study aims to investigate the differences in the impact of social network interventions on physical activity levels (PALs) between networks derived from web-based communication and peer nomination data. METHODS: We used the data on sociometric questionnaires, messages from a web-based communication app, and PAL (number of steps per day) of 408 participants in 21 school classes. We applied social network analysis to identify influential peers and agent-based modeling to simulate the diffusion of PAL and explore the impact of social network interventions on PAL among adolescents in school classes. Influential peers (n=63) were selected based on centrality measures (ie, in-degree, closeness, and betweenness) to spread the intervention. They received health education, which increased their PAL by 17%. In sensitivity analyses, we tested the impact of a 5%, 10%, and 20% increase in PAL among influential peers. RESULTS: There was a 24%-27% overlap in selected influential peers between the 2 network representations. In general, the simulations showed that interventions could increase PAL by 5.0%-5.8% within 2 months. However, the predicted median impact on PAL was slightly higher in networks based on web-based communication data than peer nomination data for in-degree (5.7%, IQR 5.5%-6.1% vs 5.5%, IQR 5.2%-5.8%; P=.002), betweenness (5.6%, IQR 5.4%-5.9% vs 5.0%, IQR 4.7%-5.3%; P<.001), and closeness centrality (5.8%, IQR 5.6%-6.1% vs 5.3%, IQR 5.0%-5.6%; P<.001). A large variation in impact was observed between school classes (range 1.5%-17.5%). Lowering the effectiveness of health education from 17% to 5% would reduce the overall impact of the social network intervention by 3-fold in both networks. CONCLUSIONS: Our findings showed that network interventions based on web-based communication data could increase PAL. Web-based communication data may therefore be a valuable addition to peer nomination data for future social network intervention design. Artificial intelligence methods, including agent-based modeling, can help to design these network interventions and provide insights into the role of network characteristics in their effectiveness.
RESUMO
Bluetooth (BT) data has been extensively used for recognizing social patterns and inferring social networks, as BT is widely present in everyday technological devices. However, even though collecting BT data is subject to random noise and may result in substantial measurement errors, there is an absence of rigorous procedures for validating the quality of the inferred BT social networks. This paper presents a methodology for inferring and validating BT-based social networks based on parameter optimization algorithm and social network analysis (SNA). The algorithm performs edge inference in a brute-force search over a given BT data set, for deriving optimal BT social networks by validating them with predefined ground truth (GT) networks. The algorithm seeks to optimize a set of parameters, predefined considering some reliability challenges associated to the BT technology itself. The outcomes show that optimizing the parameters can reduce the number of BT data false positives or generate BT networks with the minimum amount of BT data observations. The subsequent SNA shows that the inferred BT social networks are unable to reproduce some network characteristics present in the corresponding GT networks. Finally, the generalizability of the proposed methodology is demonstrated by applying the algorithm on external BT data sets, while obtaining comparable results.
RESUMO
BACKGROUND: Social network interventions targeted at children and adolescents can have a substantial effect on their health behaviors, including physical activity. However, designing successful social network interventions is a considerable research challenge. In this study, we rely on social network analysis and agent-based simulations to better understand and capitalize on the complex interplay of social networks and health behaviors. More specifically, we investigate criteria for selecting influence agents that can be expected to produce the most successful social network health interventions. OBJECTIVE: The aim of this study was to test which selection criterion to determine influence agents in a social network intervention resulted in the biggest increase in physical activity in the social network. To test the differences among the selection criteria, a computational model was used to simulate different social network interventions and observe the intervention's effect on the physical activity of primary and secondary school children within their school classes. As a next step, this study relied on the outcomes of the simulated interventions to investigate whether social network interventions are more effective in some classes than others based on network characteristics. METHODS: We used a previously validated agent-based model to understand how physical activity spreads in social networks and who was influencing the spread of behavior. From the observed data of 460 participants collected in 26 school classes, we simulated multiple social network interventions with different selection criteria for the influence agents (ie, in-degree centrality, betweenness centrality, closeness centrality, and random influence agents) and a control condition (ie, no intervention). Subsequently, we investigated whether the detected variation of an intervention's success within school classes could be explained by structural characteristics of the social networks (ie, network density and network centralization). RESULTS: The 1-year simulations showed that social network interventions were more effective compared with the control condition (beta=.30; t100=3.23; P=.001). In addition, the social network interventions that used a measure of centrality to select influence agents outperformed the random influence agent intervention (beta=.46; t100=3.86; P<.001). Also, the closeness centrality condition outperformed the betweenness centrality condition (beta=.59; t100=2.02; P=.046). The anticipated interaction effects of the network characteristics were not observed. CONCLUSIONS: Social network intervention can be considered as a viable and promising intervention method to promote physical activity. We demonstrated the usefulness of applying social network analysis and agent-based modeling as part of the social network interventions' design process. We emphasize the importance of selecting the most successful influence agents and provide a better understanding of the role of network characteristics on the effectiveness of social network interventions.
