Prediction of household dust mite concentration based on machine learning algorithm

Abstract

Household dust mites (HDMs) are the important allergens causing allergic diseases in children. A predictive model can help us understand the concentration of HDMs in different areas of China to better prevent and control this kind of allergen. This study used 454 household inspection samples in childrens’ room obtained from China, Children, Homes, Health (CCHH) phase 2 study, conducted during 2013-2014. Spearman correlation and multiple logistic regression were used to explore the influencing factors of HDMs concentrations, by comprehensively considering residents’ lifestyle, building characteristics, environmental exposure, especially dampness-related exposures. This study used the Gradient Boosting Decision Tree(GBDT) algorithm to build the prediction model. The data from CCHH were used to established the prediction model. It was found that there were some differences in the influencing factors between two types of HDMs. The concentration of HDMs were found a significant correlation (p<0. 05)with the number of indoor moisture indicators. 17 influencing factors of HDMs concentrations from four aspects were finally established in this study. The training model of GBDT has a reasonable accuracy(R2>0. 9). This paper provides a reference for predicting the HDMs concentrations in children's bedrooms and the influence of the influencing factors.