Skip to main content
Fig. 1 | Cancer Cell International

Fig. 1

From: Determining the prognosis of Lung cancer from mutated genes using a deep learning survival model: a large multi-center study

Fig. 1

Flowchart of the proposed deep learning survival (DLS) model to determine disease prognosis. The somatic mutational databases were derived from non-small cell lung cancer (NSCLC) samples. In the MSK-MET cohort (training), the selected genes were trained to predict overall survival (OS) using deep learning. After adjusting the training parameters, the DLS models were validated for OS in the MSK-MET (inter-validation), OncoSG, MSK-CSC, and TCGA-LUAD cohorts. The trained DLS model was fine-tuned and re-trained using the MIND cohort. The DLS model was validated in the MSKCC and POPLAR/OAK cohorts. The COX models were analyzed in all patients. The C-indices of the DLS model, COX model, and tumor-node-metastasis (TNM) staging were compared in patients treated without immunotherapy (IO) regarding OS. The C-indices of the DLS model, COX model, tumor mutational burden (TMB), and programmed death-ligand 1 (PD-L1) expression were also compared among patients treated with IO regarding progression-free survival (PFS).

Back to article page

Cancer Cell International

ISSN: 1475-2867

Contact us