Efficient Machine Learning and Factional Calculus Based Mathematical Model for Early COVID Prediction

Abstract

Diseases are increasing with exponential rate worldwide. Its detection is challenging task due to unavailability of the experts. Machine learning models provide automated mechanism to detect diseases once trained. It has been used to predict and detect many diseases such as cancer, heart attack, liver infections, kidney infections. The new coronavirus has become one of the deadliest diseases. Its case escalated in unexpected ways. In the literature, many machine learning models such as Extreme Gradient Boosting (XGBoosting), Support Vector Machine (SVM), regression, and Logistic regression have been used. It has been observed that these models can predict COVID cases early but are unable to find the peak point and deadline of the disease. Hence, mathematical models have been designed to early predict and find peak point and dead-line in disease prediction. These mathematical models use integral calculus-based Ordinary Differential Equations (ODEs) to predict COVID cases. Governments are dependent on these models’ pre- diction for early preparation of hospitalization, medicines, and many more. Hence, higher prediction accuracy is required. It has been found in the literature that fractional calculus-based models are more accurate in disease prediction and detection. Fractional models provides to choose order of derivative with fractional value due to which information processing capability increases. In the present work, mathematical model using fractional calculus has been devised for prediction of COVID cases. In the model, quarantine, symptomatic and asymptomatic cases have been incorporated for accurate prediction. It is found that the proposed fractional model not only predicts COVID cases more accurately but also gives peak point and dead-line of the disease.