Performance Evaluation of Three Machine Learning Models for Clinical Diagnosis of Malaria Fever

ABSTRACT

Malaria disease is mostly transmitted by the bite of Anopheles mosquito. It is caused by Plasmodium parasites and is a fatal tropical disease that claimed lives in thousands. The disease is considered as a serious health problem across the globe. In this paper, the clinical diagnosis of malaria fever using machine learning technique is used to detect malaria infection in the University community. Machine learning (ML) algorithm makes effort to forecast the illness of a patient confirmed from their symptoms. ML technique is germane in taking decisions and predictions on diverse models of classification, regression, clustering and association on medical application, design and diagnosis. Multilayer perception, Random forest classifier and Gradient boosting are foremost ML methods used to resolve the medical diagnosis problem. These methods are used to analyze malaria infection based on clinical and laboratory symptoms with appropriate data and therefore present a more efficient result. The performance evaluation metrics — Accuracy (ACC), Mean Absolute Error (MAE), Root Mean Square Error (RMSE)— were used to measure the performance of the three models. The results confirmed that the Gradient boosting model performed more accurately (in the prediction) than the other two algorithms.

INDEX TERMS Artificial intelligence, clinical diagnosis, decision support system, machine learning