Developmental origins of child neurodevelopment: Novel approaches to clinical risk prediction using an international metadataset.

全球约每10名5岁以下儿童中就有1名面临发育迟缓的风险。这对生命过程的影响是巨大的，在幼儿期出现认知、运动和语言延迟以及行为问题，伴随着较差的教育程度、较低的个人财务状况、较差的身体健康以及成年期药物滥用和精神疾病风险增加。有强有力的证据表明，从怀孕到孩子3岁生日这段时间是大脑发育的黄金机会之窗。与其他年龄较大的婴儿相比，出生后早期的脑刺激活动对儿童和成人的教育、健康和经济结果具有最大和最持久的有益影响。虽然在出生时和非常早期评估认知、语言和运动技能在技术上具有挑战性，但一些与胎儿、出生有关的因素和儿童因素(如胎儿生长、早产、儿童生长、健康和社会环境)在儿童早期与这些结果有关。然而，这些关联并不是普遍观察到的，不同国家的人口报告了不同的模式。此外，研究使用了不同的技术来评估儿童的发展，限制了它们之间的可比性。在国际人口中，了解与发育迟缓有关的风险因素是尽早确定有发育迟缓风险的儿童的关键。在神经发育评估不可行(如在新生儿或资源有限的情况下)或敏感性有限(如在婴儿期)的情况下，使用风险概况为确定处于风险中的儿童和适当确定干预目标提供了一种低成本的替代战略。然而，到目前为止，还没有这样的风险预测工具可以应用于非常早期的生活和国际人口中。为了解决这些问题，我的项目将汇集已经从9项研究收集的数据，以检查从受孕到孩子两岁生日这段时间内，生长、健康和环境影响之间的关系，对孩子两岁时的认知、运动、语言和行为结果的影响。综合起来，这一元数据集将代表来自13个国家(巴西、刚果民主共和国、芬兰、格林纳达、危地马拉、印度、意大利、肯尼亚、巴基斯坦、泰国、斯洛伐克、南非和英国)的8015名儿童。重要的是，在所有有贡献的研究中，根据一项标准化的发育测试(INTERNAL-NDA)来衡量两岁时的儿童发展。我将进行统计分析，以量化多种早期生活影响和两年后发育迟缓之间的关联，并研究这些因素如何相互作用来调整风险。我将应用统计学和机器学习的方法将风险因素分组，并使用这些方法来构建模型，以预测出生时和一岁时两岁时发育迟缓的风险。我将在来自印度迈索尔的500名儿童的独立数据集(不包括在元数据集中，在该数据集中的发育里程碑在两年后没有根据NDA进行评估)中评估该风险模型的性能。该项目将极大地惠及在高、中、低收入国家从事儿童健康工作的研究人员和临床医生，方法是：(1)扩大对与早期发育迟缓相关的风险因素的了解；(2)建立一种工具，用来预测早期发育迟缓的风险。及早识别有发育迟缓风险的儿童，以便在大脑发育的敏感窗口内有针对性地采取治疗策略，是防止儿童丧失发展潜力的关键。

Approximately 1 in 10 children under 5 years of age, globally, are at risk of delayed development. The lifecourse impact of this is substantial with cognitive, motor and language delays, and behavioural problems, during early childhood, associated with poorer educational attainment, lower personal finances, poorer physical health and increased risk of substance abuse and mental illness during adulthood. There is strong evidence that the period between conception and the child's 3rd birthday is 'a golden window of opportunity' for brain development. Brain stimulation activities during the early postnatal years, when compared with others administered at older ages, have been shown to have the greatest and most enduring beneficial impact on educational, health and economic outcomes through childhood and adult life. While it is technically challenging to assess cognitive, language and motor skills at birth and during very early life, a number of foetal, birth-related and childhood factors (such as foetal growth, prematurity, child growth, health, and social environment) have been associated with these outcomes during early childhood. However, these associations are not universally observed, and different patterns have been reported from populations in different countries. Moreover, studies have used different techniques to assess child development, limiting the comparability between them. Understanding, in international populations, the risk factors associated with development delay are key to identifying, as early as possible, children at risk of delay. Where neurodevelopmental assessments are not feasible (such as in the newborn or in resource-limited settings) or limited in sensitivity (such as during infancy), the use of risk profiles presents an alternative, low-cost strategy for the identification children at risk and for the appropriate targeting of interventions. However, as yet, there is no such risk-prediction tool for developmental delay that can be applied during very early life and across international populations.To address these issues, my project will bring together data already collected from 9 studies to examine, across high-, middle- and low-income countries, the associations between growth, health and environmental influences, in the period between conception and the child's second birthday, on cognitive, motor, language and behavioural outcomes at 2 years. Taken together, this metadataset will represent 8,015 children from 13 countries (Brazil, DRC, Finland, Grenada, Guatemala, India, Italy, Kenya, Pakistan, Thailand, Slovakia, South Africa and the UK). Importantly, child development at 2 years is measured, in all contributing studies, on one standardised development test (the INTER-NDA). I will undertake statistical analyses to quantify the association between multiple early life influences and developmental delay at 2 years and study how these factors interact with each other to modify risk. I will apply statistical and machine-learning approaches to group risk factors into clusters and use these to construct models to predict, at birth and at 1 year, the risk for developmental delay at 2 years. I will evaluate the performance of this risk-model in an independent dataset (not included in the metadataset, and in which developmental milestones were not assessed at 2 years on the INTER-NDA) of 500 children from the Mysore, India. This project will significantly benefit researchers and clinicians working in child health, across high-, middle- and low-income countries, by (1) extending the understanding of the risk factors associated with early developmental delay and (2) construction of a tool with which to predict, during early life, the risk of developmental delay. Identifying children at risk of developmental delay early, in order to target treatment strategies within sensitive window of brain development, is key to preventing the loss of developmental potential in children.