ABSTRACT
This chapter discusses how the methods of machine learning can be used to develop such a nonintrusive tool for risk stratification and screening recommendations. It describes artificial neural networks (ANNs), and discusses its application to a wider range of cancers. The chapter also describes traditional statistical and other machine learning methods in comparison with the ANN. An ANN is a machine learning method that relies on optimization theory in which inputs are fed-forward to produce predictions and then the error can be backpropagated to train the model. The standard method for the training of an ANN is via backpropagation of errors with a gradient-based optimizer, such as gradient descent. For an in-depth look at how the ANN performs, the chapter focuses on lung cancer. Multivariable logistic regression is designed to fit a function of n variables with n + 1 coefficients, with the extra value being an intercept term and equivalent to an ANN with all the hidden layers removed.
