Risk Prediction in Patients With Heart Failure With Preserved Ejection Fraction Using Gene Expression Data and...

Area of Science:

• Genomics
• Cardiovascular Medicine
• Computational Biology

Background:

• Heart failure with preserved ejection fraction (HFpEF) presents significant challenges due to high mortality, heterogeneity, and poor prognosis.
• Genomic data analysis offers a promising avenue for stratifying HFpEF patients into distinct risk categories for targeted therapies.

Purpose of the Study:

• To develop and evaluate machine learning models for predicting HFpEF patient risk and identifying subgroups with high mortality risk.
• To identify novel therapeutic targets for HFpEF by analyzing gene expression data.

Main Methods:

• Six machine learning models (GA-KPLS, LASSO, random forest, ridge regression, SVM, logistic regression) were applied to gene expression data from 149 HFpEF patients.
• Patient outcomes were assessed based on 3-year survival, classifying them into good-outcome and poor-outcome groups.
• Model performance was rigorously evaluated using established criteria.

Main Results:

• The Genetic Algorithm-Kernel Partial Least Squares (GA-KPLS) model demonstrated superior performance in predicting patient risk.
• 116 differentially expressed genes (DEGs) were identified between good and poor outcome groups, suggesting potential therapeutic targets.
• Enrichment analysis revealed DEGs are associated with Gene Ontology terms and Kyoto Encyclopedia of Genes and Genomes pathways relevant to HFpEF.

Conclusions:

• The GA-KPLS model provides a powerful tool for stratifying 3-year mortality risk in HFpEF patients.
• Identification of DEGs offers new avenues for developing precision treatments for HFpEF.
• Genomic-driven risk stratification can significantly aid in managing HFpEF patients.