An ensemble learning approach for pneumonia detection using human chest X-ray images

The advancement in technology has transformed different aspects of our daily lives. The most profound is its influence on medicine. Whether diagnosis or treatment, advances in technology have improved the quality of medical care. This article proposes a model to classify pneumonia effectively with the help of three models: InceptionV3, VGG-16, and EfficientNet. After getting the results for each model, the accuracy and AUC score of each model are compared with the ensemble of the three models, which is the proposed model. With the help of the models, binary classification was performed to categorize the chest radiograph datasets into normal and pneumonic lung detection. The Inception model achieved a maximum accuracy of 84.13% and the VGG16 model obtained a maximum accuracy of 60.42%. The EfficientNet model provided a maximum accuracy of 86.38%, the highest of the three models implemented individually. The proposed model is the ensemble of these three models which yielded a better accuracy as compared to them which is 90.38% which makes the proposed model efficient than the three models.