A novel approach based on CatBoost and explainable artificial intelligence for diagnosis of COVID-19 cases using patients' symptoms

Abstract

The COVID-19 virus, which was discovered in December 2019 in the city of Wuhan, China and quickly spread throughout the world, continues to be an important threat to the health of the world. Despite all the strategies used to deal with the spread of COVID-19, more contrivances are still needed to deal with its consequences. In this research, the clinical characteristics of people have been used as input data to diagnose a person with COVID-19, which is the result of collecting information from similar studies. Also, various algorithms including support vector machine, logistic regression, k nearest neighbor (k=9), simple bayes, random forest, LightGBM, XgBoost and CatBoost have been used, among which the CatBoost algorithm, with a sensitivity of 97.97%, accuracy 97.72% and 96.96% accuracy showed the best results. In this algorithm, the trial and error method has been used to adjust hyperparameters as accurately as possible to achieve the desired results, and SHAP is used to interpret the results and determine the impact of features on the output.

Keywords

• catboost algorithm
• corona virus
• deep neural network
• covid-19 disease
• machine learning
• shap