Geriatric depression and anxiety disorders often manifest as neuropsychiatric symptoms among those with mild cognitive impairment. Both tend to co-occur, and overlap in symptomology and etiology. Such commonalities are likely to be reflected in the brain as common neural correlates. Using connectome-based predictive modeling (CPM), we examined the functional and structural connectomes predicting depression and anxiety symptoms, and subsequently the overlap and cross-syndrome generalization of the connectomes associated with either disorder. Ninety-one older adults completed self-reported measures of depression and anxiety, and underwent diffusion tensor imaging and resting-state functional magnetic resonance imaging. Functional connectivity (FC) and structural connectivity (SC) matrices were derived from these scans and, in various combinations, entered into CPM models to predict either type of symptoms. Leave-one-out cross-validation was performed. Predictive accuracy was assessed via the correlation between predicted and observed scores (ρ_{predicted-observed}). While FC or SC features alone significantly predicted either type of symptoms, these symptoms were best predicted by models that consisted of both FC and SC features (depression: ρ_{predicted-observed} = 0.497; anxiety: ρ_{predicted-observed} = 0.455). The features common to depression and anxiety were identified and entered into another model which was similarly accurate in predicting either type of symptoms. Moreover, cross-syndrome generalization was observed— the depression-associated features significantly predicted anxiety symptoms (ρ_{predicted-observed} = 0.403) and vice-versa (ρ_{predicted-observed} = 0.378). These FC and SC features are complementary biomarkers of geriatric depression and anxiety symptoms. Both types of symptoms are largely underpinned by common patterns of altered FC and SC, alluding to the transdiagnostic neurobiological susceptibility in both disorders.

老年抑郁症和焦虑症通常表现为轻度认知障碍患者的神经精神症状。两者往往同时发生，并在症状和病因学上重叠。这些共性很可能会在大脑中反映为常见的神经关联。使用基于连接组的预测模型 (CPM)，我们检查了预测抑郁和焦虑症状的功能和结构连接组，以及随后与这两种疾病相关的连接组的重叠和跨综合征泛化。91 名老年人完成了自我报告的抑郁和焦虑测量，并接受了弥散张量成像和静息状态功能磁共振成像。功能连通性 (FC) 和结构连通性 (SC) 矩阵是从这些扫描中得出的，并且以各种组合，输入 CPM 模型以预测任一类型的症状。进行了留一法交叉验证。通过预测得分和观察得分之间的相关性评估预测准确性（ρ_预测观察）。虽然 FC 或 SC 特征单独显着预测任一类型的症状，但这些症状最好由由 FC 和 SC 特征组成的模型预测（抑郁：ρ_{预测 - 观察} = 0.497；焦虑：ρ_{预测 - 观察} = 0.455）。识别出抑郁和焦虑的共同特征并将其输入另一个模型，该模型在预测任何一种症状方面同样准确。此外，观察到交叉综合征泛化——抑郁相关特征显着预测焦虑症状（ρ_{预测 - 观察} = 0.403），反之亦然（ρ_{预测 - 观察} = 0.378）。这些 FC 和 SC 特征是老年抑郁和焦虑症状的互补生物标志物。这两种类型的症状主要是由改变的 FC 和 SC 的共同模式所支撑，暗示这两种疾病的跨诊断神经生物学易感性。