| # Clinical Stability Topology | |
| The Clarus benchmark represents different layers of clinical stability. | |
| Each dataset corresponds to a different control system. | |
| --- | |
| # Stability Layers | |
| Physiological Layers | |
| Circulatory regulation | |
| Microcirculatory perfusion | |
| Cellular energy metabolism | |
| Respiratory control | |
| Endocrine regulation | |
| Thermoregulation | |
| Coagulation balance | |
| Clinical Dynamics | |
| Sepsis progression | |
| Intervention delay | |
| Organ coupling cascades | |
| Recovery windows | |
| Ventilation–perfusion mismatch | |
| Hemorrhage compensation | |
| Operational Layer | |
| Hospital system capacity | |
| Clinical workflow stability | |
| --- | |
| # Benchmark Goal | |
| Models must infer latent instability from interacting proxy signals across these layers. |