COVID-19 Symptom Analysis and Prediction Using Machine Learning Techniques

ABSTRACT

The globe is still in a state of panic as the epidemic of COVID-19 continues to spread. Vaccines have been introduced all across the world, but many people continue to be affected. As a result, knowing former patients’ medical information may benefit medics in their battle against the disease. Artificial Intelligence (AI) has developed as a groundbreaking tool with capabilities such as meteorology, forecasts in the medical sector, predictive analytics, and so on. One of the most prominent areas of AI is Machine Learning (ML) that has recently shown promising results in a variety of fields, including medicine and, most recently, COVID analysis. In this paper, we have performed two works with regard to COVID crisis. Firstly, we have conducted a study on the most impacting symptoms noticed in a person with COVID by applying a correlation analysis with Pearson and Spearman correlation coefficients. Second, the COVID dataset was analyzed using six machine learning methods for classification tasks Random Forest (RF), Gradient Boosting, Decision Trees (DTs), Naïve Bayes (NB), Bernoulli Naïve Bayes, and Support Vector Machine (SVM), with the most influential symptoms as inputs. The performance of these algorithms are measured using the metrics, namely accuracy, F1-score, precision score, and Area under the Receiver Operating Characteristic Curve (ROC-AUC) score. On evaluating the applied machine learning algorithms, it can be concluded that all the six algorithms were found to be efficient in distinguishing the positive and negative cases of COVID with promising values of the performance metrics. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.