Some Generalized Linear Models to Explore the Effective Medical Derivatives: Biostatistical Analyses to Ensure Better Services for the Hospital Patients in the Post-COVID Era

• Activity Number: 282 - Sampling and Ensembling in Statistical Computing
• Type: Contributed
• Date/Time: Tuesday, August 9, 2022 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Computing
• Author(s): Mian Arif Shams Adnan and Silvia Irin Sharna*
• Companies: Bowling Green State University and Bowling Green State University
• Keywords: Cox Proportional Hazard Model; Logistic Regression; Matel Haenszel Test; ROC Curve; Poisson Regression; Weibull Regression; Logistic Regression; Matel Haenszel Test; ROC Curve; Poisson Regression; Weibull Regression

Abstract

COVID-19 makes a drastic requirement of knowing the proper medications, treatments, risk factors, benevolent factors for the admitted patients in hospitals. A plethora of patients, deficit of doctors, huge uses of the medical devices/accessories, serge time of life and death events made the hospital and pharmaceutical systems cumbersome. Several survival analyses like survival rates’ comparisons, Logistic Regression, Weibull Regression, Poisson Regression models and Cox Proportional Hazard Models have been found to be appropriate survival models for the said data for the COVID and non—COVID patients in hospitals. Receiver’s Operating Characteristic (ROC) curve was found to observe how better was the suggestive predictive logistic model. The optimum classification or cutoff point has been found to ensure the accuracy more than 90 %. Several odds ratios for the presence and absence of the stratum variable, odds ratio for the collapsed variable and Mantel Haenszel test statistic were obtained. Sensitivity, Specificity, False Positive Rate, False Negative Rate, Positive Predictive and Negative Predictive rates were found to observe the performance of the predictive logistic model.