A Machine Learning Approach for Prediction of Length of Stay for the Kid’s Inpatient Database

Due to a high mortality rate of children in the United States, there is an immediate need to analyze the pediatric patient care information. In this paper, we perform an exploratory and predictive analysis of the length of stay on kids' inpatient data. For this study, we used the Kids' 2016 inpatient database. In our study, we developed a prediction model using Random Forest Regression for the length of stay for inpatient kids. The accuracy of our model is 94.15%. We found that female children have longer length of stay as they are more at risk during hospitalization. Further, we found that the number of complicated births could lead to longer length of stay for unborn kids.