Resting-state voxel-wise dynamic effective connectivity predicts risky decision-making in patients with bipolar disorder type I

• Published in: Neuroscience Vol. 564; pp. 135 - 143
• Main Authors: Ji, Shanling, Zhang, Hongyong, Zhou, Cong, Liu, Xia, Liu, Chuanxin, Yu, Hao
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 09.01.2025
• Subjects: Adult; Bipolar Disorder - diagnostic imaging; Bipolar Disorder - physiopathology; Brain - diagnostic imaging; Brain - physiopathology; Connectome - methods; Decision Making - physiology; Dynamic effective connectivity; Female; Humans; Impulsive Behavior - physiology; Magnetic Resonance Imaging - methods; Male; Middle Aged; Neural Pathways - diagnostic imaging; Neural Pathways - physiopathology; Predictive modeling; Rest - physiology; Risk-Taking; Risky decision-making; Young Adult; Dynamic effective connectivity; Predictive modeling; Risky decision-making
• ISSN: 0306-4522; 1873-7544; 1873-7544
• DOI: 10.1016/j.neuroscience.2024.11.024

[Display omitted] •Dynamic effective connectivity-based prediction model is achieved using rs-fMRI.•Linear regression model predicts risky decision-making in patients with BD-I.•This model effectively predicts impulsivity in patients with BD-I. Patients with Bipolar Disorder type I (BD-I) exhibit maladaptive risky decision-making, which is related to impulsivity, suicide attempts, and aggressive behavior. Currently, there is a lack of effective predictive methods for early intervention in risky behaviors for patients with BD-I. This study aimed to predict risky behavior in patients with BD-I using resting-state functional magnetic resonance imaging (rs-fMRI). We included 48 patients with BD-I and 124 healthy controls (HC) and constructed voxel-wise functional connectivity (FC), dynamic FC (dFC), effective connectivity (EC), and dynamic EC (dEC) for each subject. The Balloon Analogue Risk Task (BART) was employed to measure the risky decision-making of all participants. We applied connectome-based predictive modeling (CPM) with five regression algorithms to predict risky behaviors as well as Barratt Impulsivity Scale (BIS) scores. Results showed that the BD-I had significantly lower risky adjusted pump scores compared to HC. The dEC-based linear regression-CPM model exhibited significant predictive ability for the adjusted pump scores in BD-I, while no significant predictive power was observed in HC. Furthermore, this model successfully predicted non-planning impulsiveness, motor impulsiveness, and BIS total score, but failed for attentional impulsiveness in BD-I. These findings provide a foundation for future work in predicting risky behaviors of psychiatric patients by using voxel-wise dEC underlying resting state.