This study broadly examines how the hierarchical organization of brain dynamics differs in individuals with schizophrenia. To investigate this, the researchers applied a thermodynamic framework based on the fluctuation-dissipation theorem (FDT) to resting-state fMRI data from patients and healthy controls, and then used whole-brain modeling and machine-learning classifiers to analyze perturbability patterns. The results showed that individuals with schizophrenia demonstrated significantly greater deviations from FDT, indicating increased non-equilibrium and heightened hierarchical organization across multiple brain networks. These deviations were not only widespread but also correlated with symptom severity, particularly negative symptoms and formal thought disorder. Furthermore, features derived from FDT deviation allowed a support vector machine classifier to distinguish patients from controls with about 82.5% accuracy, outperforming traditional connectivity-based measures. In their discussion, the authors suggest that these hierarchical disruptions may reflect altered predictive-coding processes in schizophrenia and argue that such model-based measures could serve as promising biomarkers for psychiatric disorders.