The When to Worry about Language Study (W2W-L): Joint consideration of developmental patterning and neural substrates for enhancing earlyidentification of language impairment

项目摘要 初级语言障碍（PLI）在生命早期开始，影响6-8%的儿童。语言干预 越早给药效果越好。然而，语言习得的规范性差异， 幼儿期（此处为24-36个月）导致高误报率，妨碍准确识别 在学龄前晚期之前的PLI。这项研究提出了一种新颖的，有理论基础的， 神经发育框架，旨在产生一个敏感和具体的模型，幼儿PLI 风险在这种对发育敏感的转化方法中引入的创新包括：（1） 发育前兆模型使用最先进的方法来表征多种特征和生长 幼儿出现的语言模式的模式，在一个大的社区样本;（2）将EEG/ERP 语言和事务同步的神经生物标志物到PLI预测模型;和（3）考虑 紧急心理健康风险。精神健康风险是通过多方法测量易怒， 具有发展意义的风险标记，用于将常见问题内化和外化 PLI相关。拟议中的何时担心语言学习（W2 W-L）将利用 该团队现有资助的对350名婴儿（50%易怒，50%不易怒）的研究（R 01 MH 107652，Wakschlag， PI），并通过招募200名晚说话幼儿的新子样本来丰富它。这将产生一个巨大的， 550名24个月奥尔兹的不同样本，随访至54个月大（此时PLI可以可靠地评估）。的 关键的预测因素将是通过语言技能，语言处理， 和相应的神经生物标志物。中心结局是原发性语言障碍（PLI）状态， 学龄前，通过临床金标准措施评估。关键的风险修饰语是远端和近端 交易性语言环境的特征，以及易怒的纵向模式。 具体目标：目标1。具体说明语言技能、处理、神经生物标志物和 他们的增长到早期PLI预测。假设：1a.语言技能，处理和神经生物标志物将 每一个都对PLI预测有递增的贡献。1b.考虑纵向模式将增强预测。 目标二。确定事务语言环境的远端风险和近端保护功能， 为PLI的个体差异提供最大的解释力。假设2：家族史和贫穷 父母的语言能力将增加PLI的风险，父母输入的特征，以及行为和神经 同步将降低PLI风险。目标3：检查幼儿易怒，近端语言的相互影响 环境和PLI途径上的紧急语言模式。假设3：指定这些的模型 随着时间的推移，相互影响将使PLI预测变得更加尖锐，超出其个人解释的方差 影响。目标4。评价早期PLI风险识别的临床算法的可行性。我们将使用 机器学习方法，以生成基于目标1-3的敏感/特异性、可行的临床模型。

PROJECT SUMMARY Primary language impairment (PLI) begins early in life and affects 6-8% of children. Language intervention is maximally effective the earlier it is delivered. However, normative variation in language acquisition across toddlerhood (here, 24-36 months) contributes to a high rate of false positives, impeding accurate identification of PLI prior to late preschool age. The proposed study introduces a novel, theoretically- grounded, neurodevelopmental framework designed to generate a sensitive and specific model of toddler PLI risk. Innovations introduced in this developmentally-sensitive, translational approach include: (1) a developmental precursor model using state-of-the-art methods to characterize multiple features and growth patterns of toddler emergent language patterns, within a large community sample; (2) incorporating EEG/ERP neural biomarkers of language and transactional synchrony into PLI predictive models; and (3) considering emergent mental health risk. Mental health risk is captured via multi-method measures of irritability, a developmentally meaningful marker of risk for internalizing and externalizing problems that are common correlates of PLI. The proposed When to Worry about Language Study (W2W-L) will capitalize on the team's existed funded study of 350 infants (50% irritable and 50% non irritable) (R01MH107652, Wakschlag, PI) and enrich it via recruitment of a new sub-sample of 200 late talking toddlers. This will yield a large and diverse sample of 550 24 month olds, followed to age 54 months (when PLI can be reliably evaluated). The key predictor will be toddler emergent language patterns measured via language skill, language processing, and corollary neural biomarkers. The central outcome is primary language impairment (PLI) status at preschool age, assessed via clinical gold standard measures. Key risk modifiers are distal and proximal features of the transactional language environment, and longitudinal patterns of irritability. SPECIFIC AIMS: Aim 1. Specify the contribution of language skills, processing, neural biomarkers, and their growth to early PLI prediction. Hypotheses: 1a. Language skills, processing, and neural biomarkers will each contribute incrementally to PLI prediction. 1b. Considering longitudinal patterns will enhance prediction. Aim 2. Identify the distal risk- and proximal protective- features of the transactional language environment that provide greatest explanatory power for individual differences in PLI. Hypothesis 2: Family history and poor parental language ability will increase PLI risk, and features of parental input, and behavioral and neural synchrony will decrease PLI risk. Aim 3. Examine the mutual influences of toddler irritability, proximal language environment, and emergent language patterns on PLI pathways. Hypothesis 3: A model specifying these reciprocal influences over time will sharpen PLI prediction beyond variance explained by their individual influences. Aim 4. Evaluate feasibility of a clinical algorithm for earlier PLI risk identification. We will use machine learning approaches to generate a sensitive/specific, feasible clinical model building on Aims 1-3.