GC-SMS

GC-SMS applies genetic algorithm–based modeling to identify site-specific prognostic biomarkers and improve survival prediction for cardia and non-cardia gastric cancers.

Key Features:

• Genetic algorithm-based methods: Implements a Genetic Algorithm-based Support Vector Machine (GA-SVM) and GA-based Cox regression (GA-Cox) to analyze TCGA transcriptomic data for site-specific biomarker discovery.
• Prognostic biomarker identification: Identified 10 prognostic biomarkers for cardia gastric cancer and 13 for non-cardia gastric cancer used as combined predictors in survival models.
• Model evaluation metrics: Evaluates predictive performance using area under time-dependent receiver operating characteristic (ROC) curve (AUC) and concordance index (C-index); AUC improved from 0.720 to 0.899 for cardia cancer (P = 8.75E-08) and from 0.798 to 0.994 for non-cardia cancer (P = 7.11E-16), with C-index values of 0.816 (cardia) and 0.812 (non-cardia).

Scientific Applications:

• Site-specific prognosis: Provides distinct prognostic models for cardia versus non-cardia gastric cancers based on transcriptomic biomarkers.
• Personalized treatment stratification: Enables biomarker-driven risk stratification to inform individualized therapeutic decision-making for gastric cancer patients.

Methodology:

Analysis of TCGA transcriptomic data using Genetic Algorithm-based Support Vector Machine (GA-SVM) and GA-based Cox regression (GA-Cox) with evaluation by time-dependent ROC AUC and concordance index (C-index).