Model development for better estimation of temperature and humidity effect on mortality

Many studies estimate the temperature’s effect on mortality with the daily mean temperature and the relative humidity. However, some studies considered the temperature percentile instead of temperature, and absolute humidity or dew point temperature instead of the relative humidity in the study. Therefore, we compared 6 combined models including daily mean temperature/temperature percentile and the relative humidity/absolute humidity/dew point temperature. We set the model using GAM with the temperature variable, the humidity variable, time trends and day of week. We compared AICs of the 6 models for finding out the best fitted model. After comparing 6 models, it is illustrated that model (T, DPT) had the minimum AIC value among the 6 models among 8 cites. Model (T, AH) had the minimum AIC among the 6 models among 3 cities and model (T, RH) and model (TP, AH) among 2 cities respectively. When comparing the models using the same temperature variables but the different humidity variables, the model employing the dew point temperature had the smallest AIC among the 3 models in most cities. Comparing the models using the same humidity variables but the different temperature variables, the models employing the temperature variables had the smaller AIC than the models employing temperature percentile in most cases.