Design Model of Deep Stacking Network for Breast Cancer Prediction Using Microarray

Breast cancer diagnosis currently relies on clinical information, image radiology and histopathology. However molecular biology aspect needs to be considered for accurate diagnoses. Microarray technology allows the analysis of thousands of gene expression to be used as additional information for breast cancer diagnosis. This study aims to use microarray for breast cancer diagnosis by using machine learning. Machine learning is widely used for pattern analysis and can be used for microarray dataset, such as deep stacking network (DSN). Design of DSN is stacked each of base module which using a simple form of the multilayer perceptron. Using DSN is suitable for complex data like microarray dataset because it has a deep architecture (deep learning). Furthermore, DSN model does not use stochastic gradient descent which is difficult to be implemented on large scale of machine learning. In Indonesia, microarray technology is still not well known, therefore the current studies only use secondary data from cancer patients overseas. DSN which is a deep learning model is suitable to be used for microarray dataset that has a complex structure. Suggested for subsequent study using primary data from patient cancer in Indonesia so that the design model will be more suitable to be implemented for cancer patients in Indonesia.