Virtual Support Groups for Perceived Belonging in Cancer Survivor Networks

Abstract

The transition from active oncology treatment to long term survivorship is often characterized by a profound reduction in clinical contact, leaving many cancer survivors to navigate complex psychosocial challenges in relative isolation. Virtual support networks have emerged as vital platforms for peer connection, yet the dynamics of how these digital interactions translate into a sustained sense of perceived belonging remain poorly understood. This study employs a prospective cohort design tracking 450 cancer survivors over a six month period on a specialized online support platform. By integrating behavioral tracking, natural language processing of user generated text, and temporal network analysis, we construct a predictive modeling framework to estimate individual levels of perceived belonging. Our predictive models, particularly those leveraging gradient boosted decision trees, achieve high accuracy in forecasting belonging trajectories. The findings highlight the critical role of reciprocal interaction patterns, first person plural pronoun usage, and emotional stability in posts as early indicators of strong peer integration. This research provides a quantitative foundation for designing proactive digital oncology interventions that can automatically identify isolated survivors and tailor support mechanisms to enhance their psychological well-being.