An emperor penguin optimizer application for medical diagnostics

Extreme gradient boosting (XGBoost) is a broadly adopted machine learning approach often applied to classification problems. It outperforms many contemporary methods with high classification accuracy and admirable performance. However it presets a large number of parameters that require adequate adjustment in order to ensure proper functionality. To address this, a swarm intelligence (SI) algorithm is applied to optimize parameter selection. The Emperor penguin optimizer (EPO) is a well known nature-inspired algorithm. It mathematically models huddling behaviors that allow emperor penguins to survive the harsh arctic conditions and excels in tackling problems concerning optimization. The proposed model combines the EPO algorithm with the XGBoost methodology adapted to hyperparameter optimization and feature selection. The performance of the resulting EPO-XGBoost algorithm is compared with other contemporary algorithms, and results indicate it outperforms them in terms of classification accuracy and performance when tested on datasets concerning well-known real world medical diagnosis problems of breast cancer and diabetes.