Wei-Zhi Nie

Hypergraph normal world modeling for normal-only logical visual anomaly detection

Nonlinear Functional Connectivity and ICA Reveal Default Mode Network Hyperconnectivity in Parkinson’s Disease: A Resting-State fMRI Study

It is demonstrated that nonlinear FC enhances sensitivity to PD-related network alterations, while spatial features highlight clinically relevant biomarkers and contributes to the methodological development of resting-state fMRI and its translational application to PD.

PGP-Clinical-TimeKAN: Prior-Guided Joint Probabilistic Forecasting of Clinical Trajectories

Joint trajectory forecasting provides an inspectable intermediate task, but accurate physiology forecasts alone do not ensure a calibrated event detector, and a trajectory-derived risk score remains weaker than a dedicated GRU-D classifier, which limits the present clinical claim.