Statistical modelling, dichotomization and non-linearity.

TanboÄŸa, Ibrahim Halil

Thank you for your 1 interest in our research. In our study 2 , we constructed a risk prediction model for Covid-19 related in-hospital mortality using one of the largest sample sizes. In relation to the raised questions, first, we would like to provide a brief overview of the basics of statistical modeling. Three main types of statistical modeling have been described, including descriptive, predictive, and explanatory. 3 Descriptive modeling is aimed at summarizing or representing the data structure in a compact manner. Therefore, effect estimates for each variable included in the model may be of interest in this type of modeling. However, in explanatory or causal modeling, the main focus is obtaining estimates of the causal effect of the exposure of interest after controlling for the potential confounders. Namely, point estimates for the confounders are not the output of interest in this type of modeling. Finally, predictive modeling aims to provide an estimate for the probability of the occurrence of an event using a set of predictors. In other words, the focus in predictive modeling is the performance of the model rather than the estimates for individual variable included in the model. This article is protected by copyright. All rights reserved.