特定学习障碍（specific learning disorder, SLD）是儿童常见神经发育障碍，对其学业表现和生活质量产生显著影响。目前，SLD诊断主要依赖标准化测试和专业评估，过程复杂且耗时。多项研究表明，机器学习可通过分析测试分数、手写样本、眼动数据、神经影像数据及基因数据等多类数据，自动学习输入特征与输出标签的关联，实现高效预测，在SLD的早期筛查、辅助诊断及机制研究中展现出巨大潜力。该文综述了机器学习在SLD辅助诊断研究中的应用，并探讨其处理不同数据类型时的表现。

Specific learning disorder (SLD) is a common neurodevelopmental disorder in children that significantly affects academic performance and quality of life. At present, diagnosis mainly relies on standardized tests and professional evaluations, a process that is complex and time-consuming. Multiple studies have shown that machine learning can analyze diverse data, including test scores, handwriting samples, eye movement data, neuroimaging data, and genetic data, to automatically learn the relationships between input features and output labels and achieve efficient prediction. It shows great potential for early screening, auxiliary diagnosis, and research on underlying mechanisms in SLD. This article reviews the applications of machine learning in the auxiliary diagnosis of SLD and discusses its performance when handling different data types.