This study assessed the performance of modeling approaches to estimate personal exposure in Kenyan homes where cooking fuel combustion contributes substantially to household air pollution (HAP). We measured emissions (PM_2.5, black carbon, CO); household air pollution (PM_2.5, CO); personal exposure (PM_2.5, CO); stove use; and behavioral, socioeconomic, and household environmental characteristics (eg, ventilation and kitchen volume). We then applied various modeling approaches: a single-zone model; indirect exposure models, which combine person-location and area-level measurements; and predictive statistical models, including standard linear regression and ensemble machine learning approaches based on a set of predictors such as fuel type, room volume, and others. The single-zone model was reasonably well-correlated with measured kitchen concentrations of PM_2.5 (R² = 0.45) and CO (R² = 0.45), but lacked precision. The best performing regression model used a combination of survey-based data and physical measurements (R² = 0.76) and a root mean-squared error of 85 µg/m³, and the survey-only-based regression model was able to predict PM2.5 exposures with an R² of 0.51. Of the machine learning algorithms evaluated, extreme gradient boosting performed best, with an R² of 0.57 and RMSE of 98 µg/m³.

中文翻译：

---

估算个人家庭空气污染暴露的建模方法和性能：肯尼亚的案例研究

本研究评估了建模方法的性能，以估计肯尼亚家庭中的个人暴露，在这些家庭中，烹饪燃料燃烧对家庭空气污染 (HAP) 有重大贡献。我们测量了排放量（PM _2.5、黑碳、CO）；家庭空气污染（PM _2.5，CO）；个人暴露（PM _2.5, 一氧化碳); 炉灶使用；以及行为、社会经济和家庭环境特征（例如，通风和厨房容积）。然后我们应用了各种建模方法：单区域模型；间接暴露模型，结合了人员位置和区域级别的测量；和预测统计模型，包括标准线性回归和基于一组预测变量（如燃料类型、房间体积等）的集成机器学习方法。单区模型与测量的 PM _2.5 ( R ²  = 0.45) 和 CO ( R ²  = 0.45) 的厨房浓度相当相关，但缺乏精确度。性能最佳的回归模型结合了基于调查的数据和物理测量 ( R²  = 0.76) 和 85 µg/m ^{3 的}均方根误差，仅基于调查的回归模型能够以0.51的R ²预测 PM2.5 暴露。在评估的机器学习算法中，极端梯度提升表现最好，R ²为 0.57，RMSE 为 98 µg/m ³。