GenCPM: a toolbox for generalized connectome-based predictive modeling

Understanding brain-behavior relationships and predicting cognitive and clinical outcomes from neuromarkers are central tasks in neuroscience. Connectome-based Predictive Modeling (CPM) has been widely adopted to predict behavioral traits from brain connectivity data; however, existing implementations are largely restricted to continuous outcomes, often overlook essential non-imaging covariates, and are difficult to apply in clinical or disease cohort settings. To address these limitations, we present GenCPM, a generalized CPM framework implemented in open-source R software. GenCPM extends traditional CPM by supporting binary, categorical, and time-to-event outcomes and allows the integration of covariates such as demographic and genetic information, thereby improving predictive accuracy and interpretability. To handle high-dimensional data, GenCPM incorporates marginal screening and regularized regression techniques, including LASSO, ridge, and elastic net, for efficient selection of informative brain connections. We demonstrate the utility of GenCPM through analyses of the Anti-Amyloid Treatment in Asymptomatic Alzheimer's Disease (A4) Study and the Alzheimer's Disease Neuroimaging Initiative (ADNI), showing enhanced predictive performance and improved signal attribution compared to standard methods. GenCPM offers a flexible, scalable, and interpretable solution for predictive modeling in brain connectivity research, supporting broader applications in cognitive and clinical neuroscience.