MPD-DF: Multimodal Phenotyping Dataset of Driving Fatigue -- The Raw Dataset and Questionnaire Information

Driving fatigue analysis using multimodal physiological signals has gained significant attention in human factor engineering. We established a publicly available Multimodal Phenotyping Dataset of Driving Fatigue (MPD-DF) from 50 participants using a standardized 2-hour driving simulation. This meticulously curated dataset incorporated multidimensional subjective and objective metrics for fatigue assessment, including multimodal physiological recordings (32-channel electroencephalogram, single-lead electrocardiogram, dual-channel electrooculogram, and thoracic respiratory effort signals). The supplementary metadata encompassed fatigue-associated questionnaire assessment results along with fatigue level annotations by an expert physician. The validity of the dataset was rigorously verified through multiple dimensions: (1) efficacy of fatigue induction, (2) statistical analysis of questionnaire results, (3) evaluation of signal quality, and (4) correlation between physiological signals and fatigue levels. This systematically validated dataset supports fatigue-related algorithm development, cross-dataset model validation, and investigation of fatigue mechanisms, thereby providing essential resources for transportation safety research.

基于多模态生理信号的驾驶疲劳分析在人因工程领域已受到广泛关注。本研究依托标准化2小时驾驶模拟范式，招募50名受试者，构建了公开可用的驾驶疲劳多模态表型数据集（Multimodal Phenotyping Dataset of Driving Fatigue，MPD-DF）。该数据集经过精心整理与筛选，涵盖了用于疲劳评估的多维度主客观指标，包含多模态生理信号记录：32通道脑电图（electroencephalogram）、单导联心电图（electrocardiogram）、双导联眼电图（electrooculogram）以及胸式呼吸努力信号（thoracic respiratory effort signals）。补充元数据包含与疲劳相关的问卷评估结果，以及专业医师标注的疲劳等级信息。本数据集的有效性通过多维度开展了严格验证：(1) 疲劳诱导效果验证；(2) 问卷结果统计分析；(3) 信号质量评估；(4) 生理信号与疲劳等级的相关性分析。这款经过系统验证的数据集可支撑疲劳相关算法开发、跨数据集模型验证以及疲劳机制研究，从而为交通安全领域的科研工作提供了关键资源。