北京、广州地区环境污染物暴露日均值及健康指标数据（2023-2024）

本数据集基于在北京、广州两地开展的定组研究设计，全面涵盖颗粒物、重金属污染物的个体暴露日均值、不同时间窗浓度均值，以及肺功能、FeNO、动态心电图等多项健康生理指标。数据采集于北京建筑大学、广州工业大学等现场，使用了包括便携式PM₂.₅监测仪、个体采样器及多种医疗仪器在内的专业设备。

数据处理与质控环节极为严格：滤膜采样前经高温预处理以去除背景干扰，并在恒温恒湿环境下使用高精度天平进行称重，确保了物理测量的准确性；所有监测数据均与气象参数及受试者活动日志进行时间同步，并对数据缺失值采用了一套严谨的分级填补算法进行处理，从而在源头上保障了数据的可靠性、完整性与一致性。

目前，该数据集已被成功应用于构建“暴露—活动—健康”时序关联模型，用以解析污染物混合暴露的联合健康效应与组分间的交互作用，为评估空气净化器、佩戴口罩等个人防护措施的有效性提供了直接科学证据，并支撑了基于机器学习算法的健康风险识别新方法开发。

展望未来，这批高质量数据在环境健康领域具有广阔前景，可用于深入建立特定毒性组分与心血管、呼吸系统早期损害之间的暴露—反应关系，精准识别敏感人群与关键毒性组分，从而为修订环境空气质量标准、实施区域性的精准污染防控策略提供坚实的决策依据。

This dataset is designed based on a panel study conducted in Beijing and Guangzhou, comprehensively covering 24-hour individual exposure averages of particulate matter and heavy metal pollutants, time-window-specific concentration means, as well as multiple health and physiological indicators including pulmonary function, fractional exhaled nitric oxide (FeNO), and ambulatory electrocardiography. Data were collected at sites such as Beijing University of Civil Engineering and Architecture and Guangzhou University of Technology, using professional equipment including portable PM₂.₅ monitors, personal samplers, and various medical instruments.

The data processing and quality control procedures are extremely rigorous: filter membranes undergo high-temperature pretreatment before sampling to eliminate background interference, and are weighed using a high-precision balance in a temperature- and humidity-controlled environment to ensure the accuracy of physical measurements; all monitoring data are time-synchronized with meteorological parameters and subject activity logs, and a rigorous hierarchical imputation algorithm is adopted to handle missing data, thus guaranteeing the reliability, completeness and consistency of the dataset at the source.

Currently, this dataset has been successfully applied to construct an "Exposure-Activity-Health" temporal association model, aiming to analyze the combined health effects of mixed pollutant exposure and the interactions between pollutant components, providing direct scientific evidence for evaluating the effectiveness of personal protective measures such as air purifiers and mask wearing, and supporting the development of novel machine learning-based health risk identification methods.

Looking ahead, this high-quality dataset has broad prospects in the field of environmental health. It can be used to further establish exposure-response relationships between specific toxic components and early cardiovascular and respiratory system damage, accurately identify sensitive populations and key toxic components, thereby providing solid decision-making basis for revising ambient air quality standards and implementing regional precise pollution prevention and control strategies.