Federated learning (FL), which allows for cooperative model training across decentralized data sources while protecting data privacy, has become a ground-breaking method for machine learning. FL is being utilized extensively across a number of fields, including cloud computing, blockchain, IoT, wireless networks, and most notably in medical field. Everything is data, and maintaining data privacy is another critical consideration. No option to obtain any information from healthcare providers, as no healthcare organization is willing to disclose the personal information of its patients. FL fulfills an enormous role in this instance. Due to the effectiveness of the federal learning privacy protection framework, it is now exerting a significant influence on the healthcare sector. Traditional machine learning approaches do not protect data secrecy; however, data privacy and security are ensured by federated learning mechanisms. This paper includes a comprehensive case study on the “Covid Data” dataset to predict Covid-19. Here, four ML classification models such as AdaBoost, Linear Discriminant Analysis, Extra Tree Classifier, and Decision Tree are employed to compute the best model’s accuracy along with performance metrics like precision, recall, and F1-score and we also used 10 cv cross-validation score. To elevate model performance, two feature selection methods, Mutual Information and Chi2, are used. The results enhance our comprehension of FL’s function in developing decentralized machine learning models and their consequences for real-world applications.