From Manic Symptoms to Bipolar Disorder: Neural-behavioral Markers Using Two Analytic Models

项目总结 双相情感障碍(BD)是一种毁灭性的神经精神疾病，影响2%-5%的年轻人并导致发病率， 功能障碍和自杀。无躁狂发作的前驱躁狂症状通常出现在 BD I型和II型(BD-I/II)会发生，但只有不到60%的有躁狂症状的年轻人会发展为BD-I/II。 诊断和疾病进展的不确定性导致可能有害的干预和7-10年 推迟适当的治疗。因此，当务之急是将转换为BD-I/II的风险的客观生物标记物 在发病高峰期之前在青年中识别和测试。考虑到神经测量的结构和 与情绪和奖励处理相关的功能，结合临床和行为测量， 可以改善对青少年精神结果的预测，该项目将调查 住院期间病情最严重的青少年住院，旨在建立BD的预测模型。我们的目标是使用两个 不同的分析模型来检验我们的假设。首先，建立具有机器学习功能的一般线性模型(GLM) 第二，全脑ML模式识别模型。 我们将首先在与情绪和奖励相关的回路中识别BD-I/II的神经和行为标记物 正在处理。我们假设前额叶、杏仁核和纹状体区域的活动和连接性降低 表现为睡眠较少、活动量较少、情绪和认知能力较差的行为测量将区分BD- I/II来自临床配对的无躁狂症且健康的年轻人。接下来，我们将使用ML识别一个完整的脑神经 BD-I/II分类器与临床匹配的非躁狂症住院患者的比较。目标2是在两年后确定 并量化预测转换为BD-I/II的神经和行为指标，并测试个人 在一组独立的高症状风险青少年中进行转换。目标3是识别大脑行为 应用程序开发协会。培训样本包括13-17岁的青春期中期/之后的青少年 从全国唯一的BD青少年和普通青少年专科住院病房招募 70例特征良好的青少年BD-I/II患者，与70例住院患者作为临床配对组 没有狂热的青春。测试样本是由180名没有躁狂症状的青少年组成的独立小组 BD-I/II。招募60名健康对照。该项目包括情绪和奖励处理的神经功能 和结构，临床和行为测量，包括睡眠和活动，活动记录，计算机化 在住院期间和出院后两周内进行认知测量和自我报告。为期两年 后续，临床评估将确认诊断。这是第一次采用多模式评估的研究。 结合多模式成像方法对行为和情绪症状进行综合评估 疾病特异性异常和BD-I/II的预测。这项研究的发现可能识别生物学和 青少年转化为BD-I/II的行为标记物，并可能有助于疾病特异性风险的形成 计算器，用于移动应用的低成本生物传感器，以及新的干预目标。

PROJECT SUMMARY Bipolar disorder (BD) is a devastating neuropsychiatric illness that affects 2-5% of youth and causes morbidity, functional impairment, and suicide. Prodromal manic symptoms without manic episodes usually emerge before BD types I and II (BD-I/II) develop, but less than 60% of youth with manic symptoms will develop BD-I/II. The uncertainty of diagnosis and illness progression results in potentially detrimental interventions and 7-10 years delay appropriate treatments. It is thus imperative that objective biomarkers of risk for conversion to BD-I/II are identified and tested in youth before the peak onset of illness. Given that neural measures of structure and function associated with emotion and reward processing, in combination with clinical and behavior measures, can improve prediction of psychiatric outcomes in youth, this project will investigate brain-behavior relations in the most severely ill youth during inpatient stays and aims to build a predictive model of BD. We aim to use two distinct analytic models to test our hypotheses. First a general linear model (GLM) with a machine learning (ML) model of regularized regression with cross validation and second a whole brain ML pattern recognition model. We will first identify neural and behavioral markers of BD-I/II in circuitry associated with emotion and reward processing. We hypothesize that decreased activity and connectivity in prefrontal, amygdala, and striatal regions and behavioral measures showing less sleep, lower activity, and poorer mood and cognition will distinguish BD- I/II from clinically matched youth without mania and healthy. Next, we will identify using ML a whole brain neural classifier of BD-I/II relative to clinically matched inpatients without mania. Aim 2 is to, after two years, identify and quantify the neural and behavioral measures that predict conversion to BD-I/II, and to test individual conversion in an independent group of high symptomatic risk adolescents. Aim 3 is to identify brain-behavior associations for app development. Training samples include mid-/post- pubertal adolescents aged 13-17 years recruited from the nation’s only specialized inpatient unit for adolescents with BD and the general adolescent unit at our hospital; 70 well-characterized adolescents with BD-I/II, a clinically matched group of 70 inpatient youth without mania. Testing sample is an independent group of 180 adolescents with manic symptoms without BD-I/II. 60 healthy controls will be recruited. The project includes emotion and reward processing neural function and structure, clinical and behavioral measures including sleep and activity with actigraphy, computerized cognitive measures, and self-reports during inpatient evaluation and for two weeks post discharge. At two-year follow up, clinical assessments will confirm diagnoses. This is the first study to employ a multimodal assessment of behavior and mood symptoms combined with multimodal imaging methods to comprehensively assess disease-specific abnormalities and prediction of BD-I/II. Findings from this study may identify biological and behavioral markers of conversion to BD-I/II in adolescents and may contribute developing disease-specific risk calculators, low-cost biosensors for mobile applications, and novel targets of intervention.