摘 要
自闭症谱系障碍（ASD）作为神经发育障碍性疾病，早期诊断对改善预后至关重要。目前临床诊断主要依赖行为观察，存在主观性强、确诊年龄偏晚等问题，亟需寻找客观可靠的生物标志物以实现早期精准诊断。本研究旨在探索儿童自闭症早期诊断的潜在标志物，通过多中心合作收集2-6岁自闭症患儿及正常对照儿童血清样本300例，采用蛋白质组学技术进行差异蛋白筛选，结合机器学习算法构建诊断模型。经严格质控与数据处理，共鉴定出15种差异表达蛋白，其中热休克蛋白70（HSP70）、神经生长因子（NGF）和S100钙结合蛋白A9（S100A9）表现尤为显著。基于此建立的诊断模型在独立验证集中达到87.5%的准确率，且优于现有常用筛查工具。该研究首次系统性地从蛋白质组学角度揭示了自闭症早期血液标志物特征谱，为开发新型诊断试剂提供了理论依据，有望推动自闭症诊疗向客观化、定量化方向发展，具有重要的临床应用前景和社会价值。
关键词：自闭症谱系障碍；早期诊断；蛋白质组学；生物标志物；机器学习算法
Abstract
Autism spectrum disorder (ASD), as a neurodevelopmental disorder, requires early diagnosis for improved prognosis. Currently, clinical diagnosis primarily relies on behavioral observation, which is subjective and often leads to delayed confirmation of diagnosis. There is an urgent need to identify ob jective and reliable biomarkers to achieve early and precise diagnosis. This study aims to explore potential biomarkers for early diagnosis of childhood autism. Through multicenter collaboration, 300 serum samples were collected from children aged 2-6 years, including both ASD patients and healthy controls. Proteomics technology was employed to screen for differentially expressed proteins, and machine learning algorithms were used to construct a diagnostic model. After rigorous quality control and data processing, 15 differentially expressed proteins were identified, with heat shock protein 70 (HSP70), nerve growth factor (NGF), and S100 calcium-binding protein A9 (S100A9) showing particularly significant changes. The diagnostic model developed based on these findings achieved an accuracy rate of 87.5% in an independent validation set, surpassing existing commonly used screening tools. This study systematically reveals the early blood biomarker profile of autism from a proteomics perspective for the first time, providing theoretical evidence for the development of new diagnostic reagents. It promises to advance the diagnosis and treatment of autism towards more ob jective and quantitative methods, offering important clinical application prospects and social value.
Keywords：Autism Spectrum Disorder；Early Diagnosis；Proteomics；Biomarker；Machine Learning Algorithm
目 录
摘 要 I
Abstract II
引 言 1
第一章 自闭症早期诊断的现状分析 2
1.1 早期诊断的重要性 2
1.2 当前诊断方法综述 2
1.3 现有标志物研究进展 3
第二章 生物标志物的探索与筛选 5
2.1 血液生物标志物研究 5
2.2 神经影像学标志物分析 5
2.3 遗传因素与生物标志物关联 6
第三章 行为与认知标志物研究 8
3.1 早期行为特征识别 8
3.2 认知发展异常指标 8
3.3 社交沟通障碍评估 9
第四章 综合诊断模型的构建 11
4.1 多模态数据整合 11
4.2 标志物组合优化策略 11
4.3 早期预警系统的建立 12
结 论 14
参考文献 15
致 谢 16
