Fig. 7.

ML algorithms used in the last decade to build monotherapy response prediction. Earlier prediction models were likely developed mainly using classical ML algorithms. Later, the DL algorithms were used mostly to develop the models. The majority of the studies used multi-omics data (mutation, CNV, methylation, and gene expression) collected from large screening studies such as CCLE, GDSC, CTRP, etc. EN – elastic net, RF – random forest, NN – neural network, RR – ridge regression, BM-MKL – Bayesian multitask multi-kernel learning, SVM – support vector machine, LASSO - least absolute shrinkage and selection operator, CNN – convolutional neural network, DNN – deep neural network, AE – autoencoder, VAE – variational autoencoder, MF – matrix factorization, VNN – visual neural network, GCN – graph convolutional network.