Dithiothreitol oxidative potential (OP) of PM2.5 in Beijing: quantitative contribution of metals to OP and its prediction based on machine learning models

Atmospheric fine particulate matter (PM_2.5) has been acknowledged to exert adverse health effects through reactive oxygen species (ROS) initiated oxidative stress. Oxidative potential (OP) is a chemical parameter that reflects the abilities of PM to generate ROS and deplete antioxidants as well. Dithiothreitol (DTT) method was used in this study to measure the OP activity (described as OP^DTT). Some redox metals have been demonstrated to possess strong OP^DTT activities. However, it is challenging to estimate the contribution of each metal in PM_2.5 to OP^DTT by chemical method. In this study, the quantitative contribution and ranking of 22 kinds of metals/metalloids to OP^DTT were proposed by means of machine leaning models based on hundreds of Beijing PM_2.5 samples. Among several combined models, the combined multiple linear regression (modified version) and weighted least square model emerged as the most effective. It was revealed that not only well-known metals such as Mn and Cu were important, but also some less-recognized metals like Tl made significant contributions. The modeled results presented in this paper encompass the contributions of most metals to OP^DTT and are conducive to a comprehensive understanding of the health effects of PM_2.5.