Kato Daigo, Okuno Akiko, Ishikawa Tetsuo, Itakura Shoji, Oguchi Shinji, Kasahara Yoshiyuki, Kanenishi Kenji, Kitadai Yuzo, Kimura Yoshitaka, Shimojo Naoki, Nakahara Kazushige, Hanai Akiko, Hamada Hiromichi, Mogami Haruta, Morokuma Seiichi, Sakurada Kazuhiro, Konishi Yukuo, Kawakami Eiryo
Department of Artificial Intelligence Medicine, Graduate School of Medicine, Chiba University, Chiba, Japan.
Department of Pediatrics, Graduate School of Medicine, Chiba University, Chiba, Japan.
JMIR Pediatr Parent. 2025 Apr 4;8:e58337. doi: 10.2196/58337.
The early identification of developmental concerns requires understanding individual differences that may represent early signs of neurodevelopmental conditions. However, few studies have longitudinally examined how child and maternal factors interact to shape these early developmental characteristics.
We aim to identify factors from the perinatal to infant periods associated with early developmental characteristics that may precede formal diagnoses and propose a method for evaluating individual differences in neurodevelopmental trajectories.
A prospective longitudinal observational study of 147 mother-child pairs was conducted from gestation to 12 months post partum. Assessments included prenatal questionnaires and blood collection, cord blood at delivery, and postpartum questionnaires at 1, 6, and 12 months. The Modified Checklist for Autism in Toddlers (M-CHAT) was used to evaluate developmental characteristics that might indicate early signs of atypical neurodevelopment. Polychoric or polyserial correlation coefficients assessed relationships between M-CHAT scores and longitudinal variables. L2-regularized logistic regression and Shapley Additive Explanations predicted M-CHAT scores and determined feature contributions.
Twenty-one factors (4 prenatal, 3 at birth, and 14 postnatal) showed significant associations with M-CHAT scores (adjusted P values<.05). The predictive accuracy for M-CHAT scores demonstrated reasonable predictive accuracy (area under the receiver operating characteristic curve=0.79). Key predictors included infant sleep status after 6 months (nighttime sleep duration, bedtime, and difficulties falling asleep), maternal Kessler Psychological Distress Scale scores, and Mother-to-Infant Bonding Scale scores after late gestation.
Maternal psychological distress, mother-infant bonding, and infant sleep patterns were identified as significant predictors of early developmental characteristics that may indicate emerging developmental concerns. This study advances our understanding of early developmental assessment by providing a novel approach to identifying and evaluating early indicators of atypical neurodevelopment.
早期识别发育问题需要了解可能代表神经发育状况早期迹象的个体差异。然而，很少有研究纵向考察儿童和母亲因素如何相互作用以塑造这些早期发育特征。
我们旨在确定从围产期到婴儿期与可能先于正式诊断的早期发育特征相关的因素，并提出一种评估神经发育轨迹个体差异的方法。
对147对母婴进行了一项从妊娠到产后12个月的前瞻性纵向观察研究。评估包括产前问卷和血液采集、分娩时的脐带血以及产后1个月、6个月和12个月的问卷。使用改良版幼儿自闭症检查表（M-CHAT）来评估可能表明非典型神经发育早期迹象的发育特征。多列或多系列相关系数评估了M-CHAT分数与纵向变量之间的关系。L2正则化逻辑回归和夏普利值加法解释预测了M-CHAT分数并确定了特征贡献。
21个因素（4个产前因素、3个出生时因素和14个产后因素）与M-CHAT分数显示出显著关联（校正P值<0.05）。M-CHAT分数的预测准确性显示出合理的预测准确性（受试者操作特征曲线下面积=0.79）。关键预测因素包括6个月后的婴儿睡眠状况（夜间睡眠时间、就寝时间和入睡困难）、母亲的凯斯勒心理困扰量表分数以及妊娠晚期后的母婴联结量表分数。
母亲心理困扰、母婴联结和婴儿睡眠模式被确定为可能表明新出现的发育问题的早期发育特征的重要预测因素。本研究通过提供一种识别和评估非典型神经发育早期指标的新方法，推进了我们对早期发育评估的理解。
