Identification of Mild Hepatic Encephalopathy Based on Multi-level Functional Connectivity Hypernetwork.

Area of Science:

• Neuroscience
• Medical Imaging
• Network Science

Background:

• Early diagnosis of mild hepatic encephalopathy (MHE) is crucial for treatment and preventing progression.
• Existing brain hyper-connectivity network models for neurological disorders are limited by considering only low-level temporal synchronization.

Purpose of the Study:

• To propose a novel high-level hyper-connectivity network using resting-state functional magnetic resonance imaging (rs-fMRI) for improved diagnosis of MHE.
• To capture complex interactions among brain regions for enhanced diagnostic capabilities in neurological disorders.

Main Methods:

• Construction of multi-level high-level hyper-connectivity networks from rs-fMRI data.
• Extraction and combination of node hyperdegree, hyperedge global importance, and hyperedge dispersion features.
• Application of gradient boosting decision tree for feature selection and classification, validated using leave-one-out cross-validation.

Main Results:

• The proposed method demonstrated considerable performance in diagnosing MHE.
• Validation on an independent Autism Spectrum Disorder (ASD) dataset confirmed the model's generalization ability.
• Identification of key brain regions and hyperedge features provided interpretability for the diagnostic model.

Conclusions:

• The novel high-level hyper-connectivity network is effective for diagnosing MHE.
• The method exhibits strong generalization capabilities, applicable to other neurological disorders.
• The approach offers valuable insights into brain network alterations associated with MHE.