The SARS-CoV2 virus, which causes COVID-19 (coronavirus disease) has become a pandemic and has expanded all over the world. Because of increasing number of cases day by day, it takes time to interpret the data thus the limitations in terms of both treatment and findings are emerged. Due to such limitations, the need for clinical decisions making system with predictive algorithms has arisen. Predictive algorithms could potentially ease the strain on healthcare systems by identifying the diseases. In this study, we design clinical predictive models that estimate, using artificial intelligence and data, which patients are susceptible to receive a COVID-19 disease. To evaluate the predictive performance of our models, accuracy, AUROC, and scores calculated. From 12,727 individuals, models were tested with basic information (sex, age) and the patient's type of case, which is the combination of their symptoms, their travel during the last 14 days, their contact with an infected person or their participation in a festival requiring a gathering. We used 5 machine learning algorithms (LR, SVM, k-NN, RF, XGBoost) and 1 deep learning algorithm (ANN). Our models were validated with train-test split approach. The experimental results indicate that our predictive models identify patients that have COVID-19 disease at an accuracy of 73% and AUC of 69%. It is observed that predictive models trained on patients' basic information and type of case could be used to predict COVID-19 infection in Senegal and can be helpful for medical experts to optimize the resources efficiently.
Keywords: Artificial Intelligence; Coronavirus; Deep Learning; Machine Learning; SARS-CoV2.
© 2022 The Author(s). Published by Elsevier B.V.