Mapping Intersubject Variability in Functional Connectivity in Gray and White Matter to Predict Suicide Risk Among
Wanying Jing1, Lei Yang2, Yuting Guo3
1Department of Radiology, Yantai Yuhuangding Hospital, Qingdao University, Yantai, Shandong, China (W.J., K.C., F.D., Y.S., H.X., H.M., T.C., N.M.).
Rationale And Objectives:
Suicide is recognized as the most severe complication of major depressive disorder (MDD), and predicting suicidal behavior is a crucial step in efforts to reduce suicide rates. Previous studies have shown that intersubject variability in functional connectivity (IVFC) of gray matter (GM) and white matter (WM) in patients with MDD is associated with suicide rates. However, the predictive efficacy of GM-WM IVFC in assessing suicide risk among individuals with MDD has not yet been systematically investigated.
Materials And Methods:
This study included 178 patients diagnosed with MDD (18 with suicidal ideation [SI], 23 with suicide attempt [SA], 51 non-suicidal [NS]) and 173 healthy controls (HCs). We calculated IVFC values and filtered using the two-sample t-test and Least Absolute Shrinkage and Selection Operator (LASSO) method. Subsequently, GM IVFC and WM IVFC values were fused, followed using support vector machine (SVM) and eXtreme Gradient Boosting (XGBoost) modeling separately to predict SI and SA categories. The differences in GM and WM among the three groups (SI, SA, NS) in IVFC were examined using analysis of variance, followed by post-hoc multiple comparisons.
Results:
Our research has shown that IVFC has excellent predictive capabilities for SI (AUC: SVM = 0.951, XGBoost = 0.917) and SA (AUC: SVM = 0.956, XGBoost = 0.959). GM-WM IVFC exhibited the highest predictive accuracy. Significant differences in IVFC were revealed among the SI, SA, and NS groups.
Conclusion:
Our research demonstrated that the GM-WM IVFC has substantial predictive power for suicide risk among individuals with MDD.
