A Multi-Modal Dataset of EEG Signals and M-CHAT Assessments for Autism Spectrum Disorder Detection

This dataset integrates electroencephalogram (EEG) recordings with behavioral screening outcomes from the Modified Checklist for Autism in Toddlers (M-CHAT), providing a multi-modal resource for studying early detection of Autism Spectrum Disorder (ASD). The dataset includes data from 28 children — 15 diagnosed with ASD and 13 typically developing controls — all assessed using the same experimental and behavioral protocols. EEG signals were recorded using the NeuroCONCISE FlexEEG 8-channel system following the international 10–20 electrode placement scheme. From the EEG data, four statistical features were extracted for each subject: skewness, kurtosis, coefficient of variation, and entropy, representing distributional, variability, and complexity-based signal characteristics. In parallel, M-CHAT scores were collected for each participant to provide behavioral assessment measures. The dataset also includes benchmark classification results obtained using multiple machine-learning algorithms (SVM, Neural Network, Logistic Regression, Random Forest, XGBoost) for reference and reproducibility. This resource enables researchers to explore neurophysiological and behavioral correlates of ASD, test feature-fusion approaches, and benchmark novel detection algorithms.本数据集整合了脑电图（electroencephalogram, EEG）记录与幼儿自闭症改良检查表（Modified Checklist for Autism in Toddlers, M-CHAT）的行为筛查结果，为自闭症谱系障碍（Autism Spectrum Disorder, ASD）的早期检测研究提供了多模态研究资源。本数据集涵盖28名儿童的相关数据，其中15名确诊为自闭症谱系障碍，13名为典型发育对照个体，所有受试者均采用统一的实验与行为评估方案。 脑电信号采用NeuroCONCISE FlexEEG 8通道系统采集，遵循国际10-20电极放置规范。针对每名受试者的脑电数据，共提取四类统计特征：偏度（skewness）、峰度（kurtosis）、变异系数（coefficient of variation）与熵（entropy），分别表征信号的分布特性、变异程度与复杂度特征。 同时，本数据集为每名受试者收集了幼儿自闭症改良检查表得分，作为行为评估指标。数据集还包含采用多种机器学习算法（支持向量机Support Vector Machine, SVM、神经网络Neural Network、逻辑回归Logistic Regression、随机森林Random Forest、XGBoost）得到的基准分类结果，以供研究参考与成果复现。 该数据集可为研究者开展自闭症谱系障碍的神经生理与行为关联分析、特征融合方法测试以及新型检测算法基准测评提供研究支撑。