Early and Rapid COVID-19 Diagnosis Using a Symptom-Based Machine Learning Model

Abstract

The COVID-19 pandemic has resulted in a significant global health crisis, claiming over 6.3 million lives. Rapid and accurate detection of COVID-19 symptoms is essential for effective public health responses. This study utilizes machine learning algorithms to enhance the speed and accuracy of COVID-19 diagnosis based on symptom data. By employing the Spearman feature selection algorithm, we identified the most predictive features, thereby improving model performance and reducing the number of features required. The decision tree algorithm proved to be the most effective, achieving an accuracy of 98.57%, perfect sensitivity of 1, and high specificity of 0.97. Our results indicate that combining various symptoms with AI-based machine learning techniques can accurately detect COVID-19 patients. These findings surpass previous studies, demonstrating superior performance across multiple evaluations. The integration of feature selection with advanced machine learning models offers a practical and efficient tool for early COVID-19 diagnosis, improving patient management and public health responses. This approach holds significant promise for enhancing pandemic management and healthcare delivery.