Model Coverage Justification
Part V develops models spanning energy metabolism (Chapter Energy Metabolism Models), immune dynamics (Chapter Immune System Models), neuroendocrine regulation (Chapter Neuroendocrine and Autonomic Models), integrated multi-system coupling (Chapter Integrated Multi-System Models), temporal evolution (Chapter Temporal Evolution and Disease Trajectories), and predictive applications (Chapter Predictive Applications and Clinical Translation). This section justifies why these specific models were selected and acknowledges what remains unmodeled.
1 Coverage of Part II Pathophysiology
The models were chosen to formalize the mechanisms discussed in Part II. Table Model Coverage Justification maps each pathophysiological domain from Part II to its corresponding model(s) in Part V.
| Part II Domain | Part V Model(s) | Chapter |
|---|---|---|
| Energy metabolism (Ch.~Energy Metabolism and Mitochondrial Function) | Glycolysis, Krebs, ETC, ROS, PEM, mitochondrial QC, lactate, carnitine | Ch.~Energy Metabolism Models |
| Immune dysfunction (Ch.~Immune System Dysfunction) | NK cell, cytokine network, T/B cell, viral reactivation, mast cell, coagulation | Ch.~Immune System Models |
| Neurological dysfunction (Ch. 8) | Central sensitization, small fiber neuropathy, CBF autoregulation | Ch.~Neuroendocrine and Autonomic Models |
| Neuroendocrine (Ch. 9) | HPA axis, tryptophan–kynurenine, BH4, catecholamines | Ch.~Neuroendocrine and Autonomic Models |
| Cardiovascular (Ch.~Cardiovascular Dysfunction) | Autonomic balance, OI/POTS, endothelial dysfunction | Chs.~Neuroendocrine and Autonomic Models, Immune System Models |
| Gut microbiome (Ch.~Gastrointestinal and Microbiome Dysfunction) | GI motility, SIBO dynamics, gut–brain–immune axis | Ch.~Integrated Multi-System Models |
| Sleep dysfunction (Ch. 12) | Sleep–wake cycle, circadian model | Ch.~Neuroendocrine and Autonomic Models |
| Integrative models (Ch.~Integrative Models and Multi-System Pathophysiology) | 64-variable whole-body model, bistability, bifurcation | Ch.~Integrated Multi-System Models |
| Cross-disease (Ch. 14) | EDS coupling, mast cell triangle | Ch.~Integrated Multi-System Models |
2 Selection Rationale
Three criteria guided model selection:
- Mechanistic specificity: Only domains where the underlying biochemistry is sufficiently characterized to write rate equations were modeled. Domains with purely phenomenological descriptions (e.g., “fatigue” as a subjective experience) appear as output variables rather than as dynamical subsystems.
- Clinical relevance: Models prioritize mechanisms that are either currently targetable by interventions or that generate measurable biomarkers. The energy metabolism models were developed in detail because mitochondrial function is both measurable (Seahorse assay, CPET) and targetable (CoQ10, D-ribose, pacing). Conversely, intracellular signaling cascades (NF-\(\kappa\)B, JAK-STAT) were omitted despite their mechanistic importance because they add model complexity without generating distinct clinical predictions at the whole-patient level.
- Inter-system coupling: Models were selected to capture the bidirectional feedback loops that distinguish ME/CFS from single-organ diseases. Every model in Part V participates in at least one cross-system coupling (catalogued in Chapter~Integrated Multi-System Models), ensuring that the model set captures the emergent properties that verbal reasoning cannot derive.
3 What Is Not Modeled
Despite the breadth of Part V, several domains discussed in Part II lack formal mathematical treatment:
- Detailed intracellular signaling: NF-\(\kappa\)B, JAK-STAT, mTOR, AMPK pathways are mentioned in Part II but are not modeled as separate ODE systems. Their net effects are absorbed into aggregate parameters (e.g., cytokine production rates, metabolic switching thresholds). Explicit signaling models would be needed for drug target identification at the molecular level.
- MicroRNA regulatory networks: Chapter 7 discusses miRNA dysregulation in ME/CFS (Cheema et al. 2023), but miRNA–mRNA interaction networks are not formalized. These networks involve combinatorial regulation (one miRNA suppresses hundreds of targets) that requires Boolean or constraint-based modeling rather than ODE approaches.
- Tissue-specific heterogeneity: All models treat each compartment (plasma, CNS, muscle) as homogeneous. Spatial heterogeneity within tissues—e.g., regional differences in brain perfusion, patchy muscle fiber dysfunction—is not captured. Partial differential equation (PDE) or agent-based models would be required.
- Psychological and social factors: Deconditioning, mood disturbance, social isolation, and their bidirectional interactions with physiology are acknowledged in Parts II–III but are not modeled. These factors are difficult to formalize with rate equations and are better addressed through behavioral models.
- Pharmacokinetics: Drug absorption, distribution, metabolism, and elimination are not modeled. Treatment effects are represented as parameter modifications (e.g., CoQ10 increases \(V_\text{max}^\text{ETC}\)) rather than through explicit PK/PD models. Integrating PK models would enable dosing optimization but requires drug-specific data beyond the scope of this work.
These omissions are deliberate trade-offs: each would add substantial model complexity while current data are insufficient to constrain the additional parameters. The model set in Part V is designed to be the minimum framework that captures the multi-system feedback architecture of ME/CFS while remaining parameterizable from existing data. Extensions to the omitted domains are identified as future work in the respective Model Application Guide sections of Chapters Energy Metabolism Models through Temporal Evolution and Disease Trajectories.