Machine Learning Approaches for Survival Prediction in Covid-19 Patients: A Comparative Analysis

The COVID-19 pandemic has underscored the significance of accurately predicting patient survival in order to promptly administer efficient medical care. The augmentation of biological and healthcare service data volume has been found to enhance the precision of disease prognosis, survival prediction, and other clinical assessments. Numerous biological characteristics serve as the underlying factors contributing to the etiology of numerous diseases. Therefore, it is imperative to have precise medical data that possesses appropriate characteristics in order to facilitate an analysis that exhibits exceptional clinical accuracy. In order to effectively analyze data, it is imperative to employ a machine learning model that is both exact and accurate in predicting sickness or survival outcomes. An expeditious and accurate assessment of the disease's magnitude is crucial during a particular phase of a pandemic, such as the Covid-19 outbreak. The primary aim of this research is to employ machine learning methodologies in order to forecast the survival outcomes of individuals diagnosed with COVID-19. This will be accomplished by using a publicly available dataset comprising various medical attributes pertaining to 383,499 COVID-19 patients, which was collected and made accessible by the Directorate General of Epidemiology, Secretariat of Health in Mexico. Various machine learning techniques, including Regression methods, Artificial Neural Networks, Random Forest Classifier, Support Vector Machine, AdaBoost, and XGBoost, are employed on the dataset that has undergone diverse preprocessing procedures. The experimental findings demonstrated that the system yielded numerous advantages in comparison to previous efforts in the same field.