Network-based machine learning to identify biomarkers for systemic lupus erythematosus.

Area of Science:

• Biomarker discovery
• Network biology
• Machine learning in medicine

Background:

• Systemic Lupus Erythematosus (SLE) presents diagnostic and therapeutic challenges due to its complexity.
• Conventional differential gene expression (DGE) analyses for SLE biomarkers have high false-positive rates and limited clinical utility.
• A gap exists in identifying actionable targets for SLE precision medicine.

Purpose of the Study:

• To develop a novel framework for robust biomarker discovery in SLE.
• To overcome limitations of traditional DGE analyses in identifying clinically relevant SLE markers.
• To enable precision medicine approaches for SLE through improved diagnostics and targeted therapies.

Main Methods:

• Developed NetSLE, a network-based machine learning framework integrating SLE prior knowledge and biological networks.
• Filtered false positives from differentially expressed genes (DEGs) to identify key biomarkers.
• Validated biomarker performance in predicting disease activity and stratifying patients.

Main Results:

• NetSLE identified a 150-gene biomarker panel, significantly outperforming conventional markers and full transcriptomes.
• Biomarkers accurately predicted SLE disease activity across independent patient cohorts.
• Identified drug repurposing candidates and enabled patient stratification into distinct immunological subtypes (AS1 and AS2).

Conclusions:

• NetSLE provides a translatable method for overcoming traditional biomarker discovery limitations in SLE.
• The 150-gene panel serves as a practical tool for enhancing diagnostic precision.
• The framework advances personalized medicine by guiding targeted treatments for SLE.