Uncovering the Pre-Deterioration State during Disease Progression Based on Sample-Specific Causality Network Entropy...

Area of Science:

• Biomedical data analysis
• Complex systems biology
• Network medicine

Background:

• Complex diseases can exhibit sudden shifts, or tipping points, indicating critical states.
• Early detection of pre-deterioration states is vital for managing severe disease progression.
• Conventional statistical methods struggle with high-dimensional, limited-sample data common in complex diseases.

Purpose of the Study:

• To introduce a novel quantitative approach, sample-specific causality network entropy (SCNE), for detecting critical states in complex diseases.
• To address the limitations of existing methods in identifying pre-deterioration states within high-dimensional biological data.
• To capture dynamic alterations in molecular causal relations for pinpointing disease tipping points.

Main Methods:

• Inferred sample-specific causality networks for individual subjects.
• Quantified dynamic changes in molecular causal relationships.
• Applied SCNE to numerical simulations and real-world datasets (colorectal cancer, influenza, TCGA tumors).

Main Results:

• SCNE accurately identified critical points and pre-deterioration states in complex diseases.
• The approach demonstrated superior performance compared to six existing single-sample methods.
• Validated findings through simulations and diverse biological datasets, including single-cell and cancer genomics data.

Conclusions:

• SCNE is an effective tool for detecting critical states and pre-deterioration in complex diseases.
• The method offers a robust solution for analyzing high-dimensional biological data.
• Computational findings were further supported by the analysis of signaling biomarkers.