Neural Network Based Mortality Prediction in Covid-19 Dataset

The application of machine learning (ML) is widespread throughout the economy. ML models have long been used in several technological sectors to specify and rank adverse threat characteristics. Forecasting problems are commonly addressed using a variety of prediction approaches. This research examined a blood sample database to find significant predictive indicators of disease mortality to aid preventive planning and decision-making in healthcare systems. This makes several clinical and demographic assumptions to solve the problem of forecasting the mortality result (death) of COVID-19 patients. A sizable cohort of COVID-19 patients with labeled mortality outcomes makes up the dataset used in this investigation. The dataset is preprocessed into training and testing sets, handling missing values and normalizing features. Three, seven, and nine biomarkers are highly effective at predicting patient death ML and soft computing technologies were used to choose the best candidates for this more than 98.35% of the time. The COVID-19 virus, viewed as a severe threat to humans, can be predicted using ML and soft computing models, as this research illustrates. In this research, two machine learning (ML) standard models— Support Vector Machine (SVM), Neural Network Auto Regression technique (NNETAR), and Decision Tree (DT)—are employed, along with one soft computing model. Based on the results, the NNETAR model performs well compared to other models in use.