Whole-Body Systems Model
1 Model Architecture
The whole-body model couples the four subsystem models through the interaction variables identified above. The complete state vector comprises:
- Energy metabolism (8 variables): \([\text{ATP}]\), \([\text{ADP}]\), \([\text{NADH}]\), \([\text{NAD}^+]\), \([\text{Pyruvate}]\), \(\Delta \Psi\), \([\text{ROS}]\), \(D\)
- Immune system (12 variables): \(N_r\), \(N_a\), \(N_e\), \(T_n\), \(T_e\), \(T_m\), \(T_\text{ex}\), \(B_a\), \(P\), \([\text{Ab}]\), \(M_a\), \(V\) (viral load)
- Cytokines (6 variables): \([\text{IL-1}\beta]\), \([\text{IL-6}]\), \([\text{TNF-}\alpha]\), \([\text{IFN-}\gamma]\), \([\text{IL-10}]\), \([\text{TGF-}\beta]\)
- Neuroendocrine (8 variables): \(H\), \(A\), \(F\), \(W\), \([5 \text{HT}]\), \(K\), \([\text{DA}]\), \([\text{NE}]\)
- Autonomic/cardiovascular (4 variables): \(S\), \(V_\text{vagal}\), MAP, HR
- Sleep (2 variables): \(S_\text{sleep}\), \(C\)
- Gut (5 variables): \([\text{butyrate}]\), \([\text{LPS}]\), \(\mathcal{G}\) (motility index), \(B_\text{SI}\) (SIBO load), \(\eta\) (absorption efficiency)
The total system comprises approximately 45 state variables governed by coupled ODEs. Note that several symbols are reused across subsystems (\(S\) for sympathetic tone and sleep pressure; \(V\) for vagal tone and viral load); the subscripted forms (\(S_\text{sleep}\), \(V_\text{vagal}\)) are used in the integrated model to resolve ambiguity. While large, this system is modest compared to genome-scale metabolic models (thousands of variables) and computationally tractable with standard numerical solvers.
2 Healthy Baseline and ME/CFS Disease State
The healthy state is defined by parameter values producing physiological steady-state values for all state variables (normal ATP levels, low inflammatory markers, normal cortisol rhythm, balanced autonomic tone). The ME/CFS disease state is obtained by modifying the subset of parameters identified in the individual chapters as disease-relevant (e.g., \(\alpha_\text{CI}\), \(k_\text{exh}\), \(n_F\)). The integrated model predicts emergent properties of the disease state that are not apparent in subsystem models: for instance, the energy–immune vicious cycle amplifies modest parameter changes into substantial functional impairment, and the neuroimmune coupling propagates peripheral immune activation into CNS symptoms.
3 Perturbation Responses
The whole-body model is validated by simulating responses to standardized perturbations and comparing with clinical observations:
- Exercise challenge: A pulse increase in \(J_\text{demand}\) produces transient ATP depletion. In the healthy model, all variables return to baseline within hours. In the ME/CFS model, the perturbation triggers the PEM cascade (Post-Exertional Malaise Modeling), with secondary immune activation and delayed symptom worsening—reproducing the two-day CPET findings (Keller et al. 2024).
- Infection challenge: A pulse increase in viral load \(V\) activates the immune response. In the healthy model, the infection is cleared and the immune system returns to baseline. In the ME/CFS model, the immune response is sluggish (due to NK cell exhaustion and energy limitation), clearing the infection more slowly and producing a larger and more prolonged inflammatory response—consistent with the clinical observation that infections cause prolonged relapses in ME/CFS patients.
- Stress challenge: A pulse increase in \(\sigma_\text{stress}\) activates the HPA axis. In the healthy model, cortisol rises appropriately and returns to baseline. In the ME/CFS model, the cortisol response is blunted, providing insufficient anti-inflammatory feedback, and the stress-induced sympathetic activation exacerbates autonomic symptoms.