Predicting mood disorders in adolescence is a challenge that motivates research to identify neurocognitive predictors of symptom expression and clinical profiles. This study used machine learning to test whether neurocognitive variables predicted future manic or anhedonic symptoms in two adolescent samples risk-enriched for lifetime mood disorders (Sample 1, n = 73, ages = 13-25, M [SD] = 19.22 [2.49] years, 68% lifetime mood disorder) or familial mood disorders (Sample 2, n = 154, ages = 13-21, M [SD] = 16.46 [1.95] years, 62% first-degree family history of mood disorder). Participants completed cognitive testing and functional magnetic resonance imaging at baseline, for behavioral and neural measures of reward processing and executive functioning. Next, participants completed a daily diary procedure for 8-16 weeks. Penalized mixed-effects models identified neurocognitive predictors of future mood symptoms and stress-reactive changes in mood symptoms. Results included the following. In both samples, adolescents showing ventral corticostriatal reward hyposensitivity and lower reward performance reported more severe stress-reactive anhedonia. Poorer executive functioning behavior was associated with heightened anhedonia overall in Sample 1, but lower stress-reactive anhedonia in both samples. In Sample 1, adolescents showing ventral corticostriatal reward hypersensitivity and poorer executive functioning reported more severe stress-reactive manic symptoms. Clustering analyses identified, and replicated, five neurocognitive subgroups. Adolescents characterized by neural or behavioral reward hyposensitivities together with average-to-poor executive functioning reported unipolar symptom profiles. Adolescents showing neural reward hypersensitivity together with poor behavioral executive functioning reported a bipolar symptom profile (Sample 1 only). Together, neurocognitive phenotypes may hold value for predicting symptom expression and profiles of mood pathology. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

中文翻译：

---

神经认知风险表型预测青春期情绪症状。


预测青春期情绪障碍是一项挑战，它促使研究人员确定症状表达和临床特征的神经认知预测因素。这项研究使用机器学习来测试神经认知变量是否可以预测两个终生情绪障碍风险较高的青少年样本中未来的躁狂或快感缺失症状（样本 1，n = 73，年龄 = 13-25，M [SD] = 19.22 [2.49]年，68% 终生情绪障碍）或家族情绪障碍（样本 2，n = 154，年龄 = 13-21，M [SD] = 16.46 [1.95] 年，62% 一级情绪障碍家族史）。参与者在基线时完成了认知测试和功能磁共振成像，用于奖励处理和执行功能的行为和神经测量。接下来，参与者完成了为期 8-16 周的每日日记程序。惩罚混合效应模型确定了未来情绪症状的神经认知预测因子和情绪症状的压力反应变化。结果如下。在这两个样本中，表现出腹侧皮质纹状体奖赏敏感性低下和奖赏表现较低的青少年报告了更严重的应激反应性快感缺乏。较差的执行功能行为与样本 1 中的总体快感缺失增加有关，但两个样本中的应激反应性快感缺失均较低。在样本 1 中，表现出腹侧皮质纹状体奖赏过敏和执行功能较差的青少年报告了更严重的应激反应性躁狂症状。聚类分析识别并复制了五个神经认知亚组。以神经或行为奖赏敏感性低下以及执行功能一般至较差为特征的青少年报告了单相症状。 表现出神经奖赏过敏以及行为执行功能不佳的青少年报告了双相情感障碍症状（仅限样本 1）。总之，神经认知表型可能对预测症状表达和情绪病理学具有价值。 （PsycInfo 数据库记录 (c) 2023 APA，保留所有权利）。