Length of hospital stay model of COVID-19 patients with quantile Bayesian with penalty LASSO

Abstract

This study aims to identify the best model for the length of hospital stay of COVID-19 patients in West Sumatra Province using Bayesian LASSO quantile regression method and Bayesian Adaptive LASSO quantile regression method. The quantile analysis is employed in Bayesian concept to produce more effective and natural estimated values, especially for data with non-normal distribution. The combination of the LASSO method and Adaptive LASSO as a variable selection method was applied to obtain the best model and produce estimated values that are close to the actual values. A comparison of the estimated values ​​generated from the two methods was conducted using data from 1737 COVID-19 patients at M. Djamil General Hospital in Padang from March to December 2020. The result obtained is that the Bayesian Adaptive LASSO quantile regression method generally yields a shorter 95% confidence interval, with MAD (Median Absolute Deviation), MSE (Mean Squared Error), RMSE (Root of Mean Squared Error) values smaller than those produced by the Bayesian LASSO quantile regression method. The length of hospital stay of COVID-19 patients in West Sumatra is significantly influenced by age, the diagnosis of COVID-19 patients in the positive category, the patient's discharge status in the cured and death categories, and the number of comorbidities. Below the 0.50 quantile, the length of hospital stay for patients diagnosed with positive COVID-19 who were then declared cured is around three days and 4 hours longer than the length of stay for patients diagnosed with Person Under Supervision (PerUS). It is approximately 9 hours and 50 minutes longer than the length of stay of COVID-19 patients forced to go home. The length of stay of COVID-19 patients who died was around 16 hours 31 minutes shorter than the length of stay of COVID-19 patients who were forced to discharge from the hospital.