 Non-small cell lung cancer, NSLC, patients can have varying clinical courses and outcomes, even within the same tumor stage. This study explores the use of deep learning applications in medical imaging to automatically quantify radiographic characteristics and potentially improve patient stratification. The study used an integrative analysis on seven independent datasets across five institutions totaling 1,194 NSLC patients. Using external validation in computed tomography, CT, data, the study identified prognostic signatures using a 3D convolutional neural network, CNN, for patients treated with radiotherapy and surgery. The CNN predictions were significantly associated with two-year overall survival from the start of respective treatment for radiotherapy and surgery patients. The CNN was also able to significantly stratify patients into low and high mortality risk groups in both the radiotherapy and surgery datasets. Additionally, the CNN outperformed random forest models built on clinical parameters and demonstrated high robustness against test retest and inter-reader variations. To gain a better understanding of the characteristics captured by the CNN, the study identified regions with the most contribution towards predictions and highlighted the importance of tumor surrounding tissue and patient stratification. The study presents preliminary findings on the biological basis of the captured phenotypes as being linked to cell cycle and transcriptional processes. However, the study has limitations, including the retrospective nature of the research and the opaque black box nature of deep learning networks. This article was authored by Ahmed Hosni, Chintan Parmer, Thibault Peacorer, and others. We are article.tv, links in the description below.