Optimized ensemble machine learning cancer classification system for clinical decision-making

Optimized ensemble machine learning cancer classification system for clinical decision-making

연구 분야: Artificial Intelligence

학회: Neural Computing and Applications

Biomedical and healthcare informatics research is accelerating at an unprecedented rate due to the growth and accumulation of large volumes of biological and clinical data. There are new opportunities to use big data to uncover new insights and develop creative approaches to improve the quality of cancer treatment. The microarray gene expression profile is used to efficiently and accurately classify cancer cells for clinical decision-making. In this study, a cancer classification system using an optimized ensemble machine learning approach which is based on artificial bee colony (ABC) optimization and an ensemble of machine learning classifiers is proposed as a result of this effort to distinguish between acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML) utilizing microarray gene expression patterns. The relevant cancer features are optimally selected using the ABC technique to improve the performance of the proposed ensemble learning-based classification system. Moreover, the gene expression dataset is balanced using an upsampling technique and an equal number of ALL and AML records have been used for experimentation. The proposed methodology’s accuracy and other performance metrics are looked at, and the suggested model is contrasted to several base machine learning algorithms based on performance criteria to show how useful it is. Furthermore, to demonstrate the importance of the suggested strategy, receiver operating characteristics (ROC) analysis has been performed, and it is seen that the area under the ROC curve of the proposed approach is higher compared to the existing approaches.