BALLADEER ADHD

Attention Deficit Hyperactivity Disorder (ADHD) is a prevalent neurodevelopmental disorder affecting children and adolescents, characterized by inattention, hyperactivity, and impulsivity. Current diagnostic methods primarily rely on subjective clinical evaluations, which are prone to bias. Advances in neurophysiological assessment, particularly through electroencephalography (EEG), eye tracking, and electrodermal activity (EDA), offer promising avenues for objective diagnosis and monitoring of ADHD. However, the lack of large, publicly available multimodal datasets has hindered progress in machine/deep learning-based ADHD classification and biomarker discovery. To address this gap, we present the BALLADEER ADHD Dataset, a comprehensive, multimodal dataset integrating EEG, eye tracking, and physiological signals from children and adolescents with ADHD and neurotypical controls. The dataset was collected by using a controlled experimental protocol involving cognitive and attentional tasks, including Attention Slackline, Attention Robots, and CogniFit. These tasks were designed to elicit neurophysiological responses associated with attentional control, response inhibition, and cognitive flexibility—key domains affected in ADHD. This dataset enables the development and validation of machine/deep learning models for ADHD classification and fosters research into digital biomarkers for objective assessment. Additionally, it supports cross-modal analyses, exploring the interplay between EEG dynamics, eye movements, and autonomic nervous system activity in attentional performance. We detail our methodology for data acquisition, preprocessing, and quality assurance, highlighting potential applications in clinical and research settings. By making this dataset publicly available, we aim to enhance transparency, reproducibility, and innovation in computational neuroscience and ADHD research.

注意缺陷多动障碍（Attention Deficit Hyperactivity Disorder, ADHD）是一种高发的神经发育障碍，累及儿童与青少年群体，核心临床表现为注意力涣散、多动及冲动症状。当前主流的ADHD诊断方法主要依赖主观临床评估，易引入评估偏倚。近年来神经生理学评估技术取得显著进展，尤其是脑电图（electroencephalography, EEG）、眼动追踪与皮肤电活动（electrodermal activity, EDA）等技术，为ADHD的客观诊断与动态监测提供了极具应用前景的路径。然而，目前缺乏大规模公开的多模态数据集，这一短板极大阻碍了基于机器学习/深度学习的ADHD分类研究与生物标志物挖掘工作。为填补这一研究空白，本研究发布BALLADEER ADHD数据集：这是一套综合多模态信息的数据集，整合了ADHD患者与神经典型对照个体的儿童及青少年群体的脑电图、眼动追踪数据与生理信号。该数据集通过标准化实验范式采集，包含多项认知与注意任务，具体为注意力走绳（Attention Slackline）、注意力机器人（Attention Robots）以及CogniFit任务。这些任务的设计初衷是诱发出与注意控制、反应抑制及认知灵活性相关的神经生理反应——而这三类认知功能正是ADHD患者受损的关键领域。本数据集可用于开发并验证基于机器学习/深度学习的ADHD分类模型，同时推动用于ADHD客观评估的数字生物标志物相关研究。此外，该数据集还支持跨模态分析，用于探究脑电图动态变化、眼动特征与自主神经系统活动在注意表现中的相互作用机制。本研究详细阐述了该数据集的采集、预处理与质量控制流程，并重点说明了其在临床与科研场景中的潜在应用价值。通过将该数据集公开共享，本研究旨在提升计算神经科学与ADHD研究领域的透明度、可重复性与创新水平。

搜集汇总

背景与挑战

背景概述

BALLADEER ADHD数据集是一个多模态数据集，整合了脑电图（EEG）、眼动追踪和生理信号，用于儿童和青少年注意力缺陷多动障碍（ADHD）的客观诊断与研究。该数据集通过认知任务收集ADHD患者和神经典型对照组的数据，支持机器学习模型开发和跨模态分析，旨在推动ADHD数字生物标志物发现和计算神经科学进步。