Effective identification of Swine flu-H1N1 virus using machine learning algorithms

Abstract

The swine flu is a highly contagious virus and poses a significant threat to human health. The virus affects various physiological systems and presents severe symptoms that harm the affected and the community. Due to its high transmission rate and adverse health impacts, it is imperative that the authorities find ways to detect and predict its outbreak early. Therefore, machine learning presents an avenue to help identify the illness before it gets out of hand. Review of the use and ways through which machine learning can be used will lead to the ability of predicting the infection with precision and saving thousands of lives.  Current research work proposed the type of swine flu virus infecting a patient, using Machine Learning techniques. Developed in the prototype application are different ML techniques like Support Vector Machine, logistic regression, Decision Tree and Naive Bayes.  The process begins by pre-processing the data to clean the input from possibly noisy values, missing data, scale, and data reduction so that the input data will be prepared and ready for analysis. Subsequently, the pre-processed data is fed into the ML algorithms in order to elicit predictions regarding the type of swine flu virus from the patient's records.  Other measures adopted in this research, for example, accuracy, precision, recall, F1-score, were also used in the performance evaluation of the ML models. Such measures will be essential in reflecting the effectiveness of each single model in the right classification of the swine flu virus type.  The results show that the Support Vector Machine have outperformed the other ML models in all evaluation metrics, scoring an accuracy rate of 96.32%. This implies that the ensemble techniques are very effective in identifying the type of swine flu virus present in the patient to take more informed medical decisions and treatment strategies.