Vulture-Based AdaBoost-Feedforward Neural Frame Work for COVID-19 Prediction and Severity Analysis System

Abstract

In today's scenario, many scientists and medical researchers have been involved in deep research for discovering the desired medicine to reduce the spread of COVID-19 disease. However, still, it is not the end. Hence, predicting the COVID possibility in an early stage is the most required matter to reduce the death risks. Therefore, many researchers have focused on designing an early prediction mechanism in the basis of deep learning (DL), machine learning (Ml), etc., on detecting the COVID virus and severity in the human body in an earlier stage. However, the complexity of X-ray images has made it difficult to attain the finest prediction accuracy. Hence, the present research work has aimed to develop a novel Vulture Based Adaboost-Feedforward Neural (VbAFN) scheme to forecast the COVID-19 severity early. Here, the chest X-ray images were employed to identify the COVID risk feature in humans. The preprocessing function is done in the initial phase; the error-free data is imported to the classification layer for the feature extraction and segmentation process. This investigation aims to track and segment the affected parts from the trained X-ray images by the vulture fitness and to segment them with a good exactness rate. Subsequently, the designed model has gained a better segmentation accuracy of 99.9% and a lower error rate of 0.0145, which is better than other compared models. Hence, this proposed model in medical applications will offer the finest results. Graphical abstract

Development of a Machine Learning Model for Predicting 28-Day Mortality of Septic Patients With Atrial Fibrillation

“…The AdaBoost had been widely used in clinical research in recent years. It exhibited great performance in the detection of many diseases, such as obstructive sleep apnea, early detection of sepsis, or COVID-19 (40,41). One of the merits of AdaBoost is that it has a high degree of precision and takes into account the weight of each classifier.…”

“…To date, ML models have shown superior performance in predicting prognosis, and these models can help make decisions about interventions and drug use (39). Like decision trees, AdaBoost improves prediction accuracy by assigning different weights to each weak classifier and combining sets of weak classifiers into a single strong one (40). The AdaBoost had been widely used in clinical research in recent years.…”

Predicting ICU Admission for COVID-19 Patients in Saudi Arabia: A Comparative Study of AdaBoost and Bagging Methods

“…The best average accuracy result for AdaBoost-CNN model was 94.5%, wheres it was 89% for the AdaBoost-ResNet-152 model. Mary et al [16] aimed to predict COVID-19 severity by identifying and classifying COVID-19 cells in a chest X-ray dataset. The used dataset contains 10,000 images of chest X-rays, as well as CSV files, which were located at the Kaggle site.…”

“…The dataset comprised 639 records, with 44 features that 15 Reverse Transcription Polymerase Chain Reaction 16 Site: https://hesn.moh.gov.sa/webportal/ included clinical and demographic information about symptomatic and asymptomatic patients. There are three types of variables in the dataset: Boolean (15), categorical (15), and numerical (18).…”