Graph Convolutional Networks based Non-Small Cell Lung Cancer Identification using RNA-seq Data from Blood Samples

A methodology for the identification of non-small cell lung cancer from blood samples, combining feature selection methods followed by Graph Convolutional Networks (GCN) with Genetic Algorithm (GA) optimization is presented. The methodology was tested on RNA-seq data from the GSE207586 dataset. The evaluation results showed that mixing the top 100 features from different feature selections and modeling with GCN offered the highest identification performance among all evaluated setups.Clinical Relevance: Ability to diagnose and correctly classify lung cancer subtypes non-invasively have been elusive up until recently. However, current solutions often rely on combination of imaging (chest X-ray, low dose CT scan) with blood sample and clinical data. While this could push the boundary in the non-invasive screening, diagnosis, and classification of lung cancers, it would be challenging to deploy in the primary care or in remote secondary care settings. This current research has demonstrated the combination of gene feature selection techniques and graph convolution network approaches could lead to the detection and subclassification of lung cancers with a very high diagnostic accuracy just from a single blood draw.