Results from the connectome-based predictive modeling failed to reveal a model that can predict longitudinal standard/extreme bingeing from changes in connectivity edges with cross-validation. Several factors may contribute to this finding. Firstly, a recent meta-analysis showed that individual connectivity edge has low test-retest reliability (Noble et al., 2019). Secondly, the strength of univariate association between whole-brain connectome and a behavioral measure of individual differences (bingeing frequency), is often small and might require a larger sample to detect (Marek et al., 2020). Thirdly, although some participants in this sample reported high levels of standard/extreme bingeing, on average, participants had relatively low bingeing frequency (median standard bingeing: sBinge=2.33 days/month, eBinge=2.71 days/month; median extreme bingeing: sBinge=0.35 days/month, eBinge=0.96 days/month), which might explain the weak associations between bingeing and change in connectivity.
