A Learning Model for Acute Myeloid Leukemia Prediction Using Dense Polynomial Dimensionality-Based Predictor

Analysis of microarray data is extremely complex and considered as a hot topic in recent research. Acute Myeloid Leukemia (AML) prediction based on machine learning shows huge impact on prediction which automatically diagnoses the disease severity and any malfunctions. It is important to design the relevant classifier that processes the large data volume with large data size. Deep learning is an updated machine learning approach for mitigating these issues. It is easy to handle the huge volume of data because of the large number of hidden layers. The proposed classification methodology is used for understanding the training of the proposed Dense Polynomial Dimensionality based Predictor Model (). The hidden neuron numbers are large in a sufficient way where the proposed is elaborated to predict AML. AML and ALL samples are classified using five layers in the deep network model. The data is partitioned as 20% data and 80% data testing and training in the network. Compared with other classifiers, the satisfying outcome from the proposed is higher and fulfilling. The validation is done in three datasets: Kaggle, Gene expression and Bio GPS and it gives 96% accuracy, 94% precision, 96% recall, 96% F1-score, and 98% AUROC while executing with Kaggle; then, 95.50% accuracy, 94% precision, 95% recall, 96% F1-score, and 96% AUROC is achieved while executing with Gene expression and finally 98% accuracy, 94.5% precision, 98.5% recall, 96% F1-score, and 94% AUROC is achieved while executing with Bio GPS. Based on this analysis, it is proven that the model works well with the proposed and establishes a better trade-off.