Abstract
As a highly sophisticated disease that humanity faces, cancer is known to be associated with dysregulation of cellular mechanisms in different levels, which demands novel paradigms to capture informative features from different omics modalities in an integrated way. Successful stratification of patients with respect to their molecular profiles is a key step in precision medicine and in tailoring personalized treatment for critically ill patients. In this article, we use an integrated deep belief network to differentiate high-risk cancer patients from the low-risk ones in terms of the overall survival. Our study analyzes RNA, miRNA, and methylation molecular data modalities from both labeled and unlabeled samples to predict cancer survival and subsequently to provide risk stratification. To assess the robustness of our novel integrative analytics, we utilize datasets of three cancer types with 836 patients and show that our approach outperforms the most successful supervised and semi-supervised classification techniques applied to the same cancer prediction problems. In addition, despite the preconception that deep learning techniques require large size datasets for proper training, we have illustrated that our model can achieve better results for moderately sized cancer datasets.
Highlights
Advances in big data and high-throughput technologies during the past decade have led to massive accumulation of high-dimensional omics data, which enables the data-driven prediction of disease prognosis using molecular profiles
We compared the performance of the proposed model with two baselines: 1) when we substitute the deep belief parts with the supervised support vector machine (SVM) classifiers and 2) when we use semi-supervised graph-based Laplacian SVMs as a surrogate method
Our results suggest that integration of latent features generated by deep belief networks from different modalities leads to improvements for majority of the cases
Summary
Advances in big data and high-throughput technologies during the past decade have led to massive accumulation of high-dimensional omics data, which enables the data-driven prediction of disease prognosis using molecular profiles. Fakoor et al (2020) used a stack of sparse autoencoders along with an augmenting dimensionality reduction step using PCA, to learn features from gene expression data that can help classify cancer types They developed three variants of their proposed paradigm and showed that they perform reasonably well across different datasets in some of their devised experiments, but not all. The addition of PCA to extract new features from randomly selected probes is a necessary step in their pipeline as the sparse stacked autoencoder is not enough by itself to predict informative features Their approach uses only a single data modality, i.e., gene expression data, for prediction of cancer type.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
