Skip to main content
Fig. 5 | BMC Medical Genomics

Fig. 5

From: Improving lung cancer risk stratification leveraging whole transcriptome RNA sequencing and machine learning across multiple cohorts

Fig. 5

Cross-validation performance for down-classification. The performance is evaluated on within-indication training samples with low/intermediate pre-test risk (N = 162) using 10 repeats of 5-fold CV. The original Gould model was used to score training samples. GLM(m) is a generalized linear regression model only containing main effects of clinical features and genomic features. GLM(i) includes main effects and interactions between clinical features and genomic features. GLM(m) and GLM(i) used the same set of input clinical features: age, gender, nodule size, pack-year, years since quitting smoking, specimen collection timing and genomic smoking index. “Ensemble” is the final GSC classifier

Back to article page