Model Application Guide
1 HPA Axis (Equations hpa axis and HPA Axis Models)
Measurements required. (1) Salivary cortisol at waking, +30 min (cortisol awakening response, CAR), noon, and bedtime (why: four time points constrain circadian amplitude \(a_c\), phase \(\phi_c\), and mean cortisol level). (2) ACTH stimulation test (why: distinguishes central from adrenal dysfunction—the model predicts enhanced central feedback, not adrenal failure). (3) Dexamethasone suppression test (why: constrains the feedback sensitivity parameter \(n_F\)—enhanced suppression = higher \(n_F\)).
Worked example. Healthy HPA: \(a_c = 0.6\) (60% circadian modulation), \(n_F = 2.5\). The cortisol awakening response (CAR) is \(F_{\text{peak}} / F_{\text{waking}} \approx 1 + 0.6 = 1.6\) (60% rise). In ME/CFS with \(a_c = 0.3\) and \(n_F = 4.0\) (steeper feedback): CAR \(\approx 1.3\) (30% rise—blunted). Basal cortisol is reduced because the steeper feedback suppresses CRH at lower cortisol concentrations: healthy trough \(F \approx 2\) nmol/L, ME/CFS trough \(F \approx 1.2\) nmol/L. The ACTH response to CRH stimulation is preserved (the pituitary and adrenal are intact), but the hypothalamic drive is suppressed. This pattern matches Cleare’s findings (Cleare et al. 1999): low-normal cortisol with preserved adrenal reserve.
Inter-model dependencies. Inputs: stress signals (\(\sigma_{\text{stress}}\)) from external events and from pain model (Central Sensitization and Pain Amplification); pro-inflammatory cytokines modulate CRH release. Outputs: cortisol \(F\) inhibits pro-inflammatory cytokine production (Chapter Integrated Multi-System Models, Equation cortisol immune), modulates viral reactivation rate (Chapter Immune System Models), and influences circadian sleep drive.
Scope and rationale. The three-variable model (CRH, ACTH, cortisol) omits ultradian pulsatility (~90-minute pulses), mineralocorticoid vs. glucocorticoid receptor dynamics, and cortisol-binding globulin. These are acceptable omissions for capturing the ME/CFS HPA phenotype (blunted diurnal rhythm, enhanced feedback) at clinically relevant timescales (hours to days). Sub-hourly dynamics would require a pulsatile model with \(>\) 10 additional parameters.
Falsification criteria. The model predicts that ME/CFS HPA dysfunction is central (enhanced hypothalamic feedback) rather than adrenal (reduced cortisol synthesis). Falsified if: ACTH stimulation tests in ME/CFS cohorts show impaired adrenal cortisol production (reduced cortisol response to exogenous ACTH), indicating primary adrenal insufficiency rather than central dysregulation.
Subgroup qualifier: The central-feedback model applies to the functional HPA phenotype characteristic of established ME/CFS. It does not apply to, and is not falsified by, the post-viral pituitary-injury subgroup described in Chapter Endocrine and Metabolic Dysfunction (Maladaptive Chronic Inflammatory Signaling), in which SARS-CoV-1, SARS-CoV-2, or other coronaviruses directly damage pituitary corticotrophs via ACE2-mediated infection or autoimmune hypophysitis (Leow et al. 2005) (Carosi et al. 2024). In this subgroup, reduced ACTH output reflects structural pituitary damage—a mechanistically distinct pathway that coexists with but does not invalidate the central-feedback model for the broader ME/CFS population. Identifying this subgroup requires dynamic testing (ACTH stimulation test or ITT) rather than morning cortisol, and its prevalence in post-COVID ME/CFS cohorts remains to be established prospectively.
Clinical implications. Whom to treat: patients with documented blunted CAR (\(< 30%\) rise) and/or low 24-hour urinary free cortisol. How: the model predicts that low-dose hydrocortisone replacement (5–10 mg/day, physiological, not pharmacological) should partially restore the anti-inflammatory brake on the immune system. The model also predicts that stress management interventions reduce \(\sigma_{\text{stress}}(t)\), freeing cortisol capacity for immune regulation rather than stress response. The enhanced feedback sensitivity (\(n_F\) elevated) means that standard-dose glucocorticoids will produce exaggerated suppression—lower doses are predicted to be both safer and more effective in ME/CFS than in primary adrenal insufficiency.
2 Autonomic Balance and Orthostatic Intolerance (Equations ans balance and baroreflex)
Measurements required. (1) Heart rate variability (HRV): time-domain (SDNN, RMSSD) and frequency-domain (LF/HF ratio) from 24-hour Holter or 5-minute resting ECG (why: constrains sympathetic/parasympathetic balance \(S/V\)). (2) Active standing test or tilt-table test: HR and MAP at 1, 3, 5, 10 minutes (why: constrains baroreflex gains \(G_S\), \(G_V\) and response time constants \(\tau_{\text{MAP}}\), \(\tau_{\text{HR}}\)). (3) Blood volume estimation: dye dilution or impedance cardiography (why: determines whether orthostatic intolerance is hypovolemic).
Worked example. POTS assessment: patient has resting HR = 72, standing HR at 10 min = 108 (\(\Delta\)HR = 36, meeting POTS criterion of \(\geq\) 30). Standing MAP is maintained at 82 mmHg (vs. 85 supine)—no orthostatic hypotension. Using the model with reduced blood volume (\(\Delta P_{\text{grav}}\) increased by 30%) and reduced \(G_V\) (parasympathetic gain halved): the model predicts the observed pattern—adequate MAP maintenance via excessive sympathetic compensation, at the cost of tachycardia and elevated cardiac energy demand. The energy cost of standing: the sympathetic overdrive increases cardiac oxygen consumption by ~15%, directly reducing the activity energy budget.
Inter-model dependencies. Inputs: histamine from mast cell model (vasodilation), cortisol from HPA (stress response modulation), pain from sensitization model (sympathetic drive). Outputs: heart rate and MAP to oxygen delivery equation (Chapter Integrated Multi-System Models); vagal tone to gut motility (Chapter Integrated Multi-System Models); sympathetic tone to energy demand.
Scope and rationale. The two-variable ANS model (\(S\), \(V\)) with baroreflex feedback captures the core hemodynamic physiology. It omits thermoregulatory reflexes, respiratory sinus arrhythmia, and detailed baroreceptor afferent processing. These could be added for specific applications (e.g., heat intolerance modeling, see ch10:Tonic Cutaneous Vasoconstriction Bias Explains Dual Heat+Cold Intolerance for a proposed ET-1 vasomotor mechanism and ch06:Spare Respiratory Capacity as Thermoregulatory Capacity Proxy in ME/CFS for spare respiratory capacity as thermoregulatory bottleneck) but are unnecessary for the primary targets: orthostatic HR/MAP dynamics and POTS subtyping.
Falsification criteria. The model predicts three distinct POTS parameter regimes (neuropathic, hypovolemic, hyperadrenergic) with different treatment responses. Falsified if: POTS subtype classification by tilt-table hemodynamics does not predict differential treatment response (e.g., if patients classified as hypovolemic by the model respond equally well to parasympathomimetics as to volume expansion).
Clinical implications. Whom to treat: all ME/CFS patients with orthostatic symptoms (the majority). How: the model provides mechanistic subtyping. Hypovolemic POTS (\(\Delta P_{\text{grav}}\) dominant): volume expansion (saline infusions, fludrocortisone, compression garments, increased salt/fluid intake). Neuropathic POTS (\(G_V\) reduced): parasympathomimetics (pyridostigmine (Raj et al. 2005)). Hyperadrenergic POTS (\(S\) elevated): sympatholytics (propranolol, clonidine). Mixed presentations: combination therapy targeting the dominant mechanism first.
3 Tryptophan–Kynurenine Axis (Equations tryptophan and ido regulation)
Measurements required. (1) Plasma tryptophan, kynurenine, and kynurenine/tryptophan (K/T) ratio (why: K/T ratio is a direct readout of IDO activity). (2) Serotonin (platelet-poor plasma or CSF) (why: validates the serotonin depletion prediction). (3) IFN-\(\gamma\) (why: the primary driver of IDO upregulation in the model). (4) Quinolinic acid and kynurenic acid (why: neurotoxic vs. neuroprotective branches of the kynurenine pathway).
Worked example. Healthy: \(v_{\text{IDO}} = v_{\text{IDO,basal}} = 0.5\) \(\mu\)M/h, \(v_{\text{TPH}} = 1.0\) \(\mu\)M/h. With dietary tryptophan input \(I_W = 2.0\) \(\mu\)M/h: steady-state \(W \approx 60\) \(\mu\)M, serotonin production = $ 1.0 /(K_{} + 60) $ \(\mu\)M/h. In ME/CFS with \([\text{IFN-} \gamma] = 20\) pg/mL (vs. 5 healthy): \(v_{\text{IDO}} = 0.5 + 2.0 \times 20^2/(15^2 + 20^2) = 0.5 + 1.28 = 1.78\) \(\mu\)M/h. Now IDO outcompetes TPH: tryptophan drops to ~35 \(\mu\)M, serotonin production falls to 0.54 \(\mu\)M/h (37% reduction), and kynurenine production increases to 1.1 \(\mu\)M/h (doubled). This quantifies the “serotonin steal” mechanism.
Inter-model dependencies. Inputs: IFN-\(\gamma\) from immune model drives IDO. Outputs: serotonin level affects mood, sleep (via sleep model), and pain modulation; kynurenine metabolites contribute to neurotoxicity and microglial activation.
Scope and rationale. The model tracks the branch point (IDO vs. TPH competition for tryptophan) but omits downstream kynurenine metabolism (quinolinic acid, kynurenic acid, picolinic acid). This branch-point focus captures the clinically relevant serotonin depletion mechanism. A full kynurenine pathway model would require tissue-specific enzyme data (brain vs. peripheral IDO) currently unavailable for ME/CFS.
Falsification criteria. The model predicts that the K/T ratio correlates with IFN-\(\gamma\) levels and that serotonin depletion is proportional to IDO upregulation. Falsified if: ME/CFS patients show serotonin depletion without elevated K/T ratio (indicating a serotonin synthesis or degradation defect unrelated to IDO), or if anti-inflammatory treatment that reduces IFN-\(\gamma\) does not normalize K/T ratio.
Clinical implications. Whom to treat: patients with elevated K/T ratio (measurable by standard metabolomics). How: the model predicts that SSRIs will have reduced efficacy in these patients because the synthesis bottleneck (not reuptake) limits serotonin availability. More effective approaches: (1) reduce IFN-\(\gamma\) (anti-inflammatory therapy, LDN) to de-repress TPH; (2) tryptophan supplementation (but the IDO trap hypothesis predicts this may paradoxically increase kynurenine production if IDO is in the high-activity state); (3) BH₄ supplementation (addresses TPH cofactor limitation, Tetrahydrobiopterin Competition Model). The model provides a specific negative prediction: SSRIs should fail preferentially in patients with high K/T ratios—a testable pharmacogenomic hypothesis.
4 BH₄ Competition (Equation bh4 dynamics)
Measurements required. (1) Serum BH₄ and BH₂ (why: the BH₄:BH₂ ratio indicates net oxidative stress on the cofactor pool). (2) Serum neopterin (why: marker of GTP cyclohydrolase activity, the rate-limiting BH₄ synthesis enzyme, upregulated by IFN-\(\gamma\)). (3) Concurrent serotonin, dopamine metabolites (HVA), and NO metabolites (nitrate/nitrite) (why: validates the three-way competition—all three should be reduced when BH₄ is limiting). (4) IFN-\(\gamma\) (why: driver of iNOS-mediated BH₄ consumption).
Worked example. Healthy: BH₄ pool = 10 nmol/L, consumed by TPH (2), TH (2), eNOS (1), iNOS (0.5) = 5.5 total, with synthesis rate matching consumption. In ME/CFS with IFN-\(\gamma\)-driven iNOS upregulation: iNOS consumption increases to 3.5 nmol/L/h. Total demand = 8.5 vs. supply of 5.5 (assuming compensatory GTPCH upregulation adds 1.5). Deficit = 1.5 nmol/L/h. The three hydroxylases compete: TPH drops from 2 to 1.3 (35% serotonin reduction), TH drops from 2 to 1.3 (35% dopamine/NE reduction), eNOS drops from 1 to 0.6 (40% NO reduction, with uncoupling producing superoxide instead). All three systems fail simultaneously from a single cofactor shortage.
Inter-model dependencies. Inputs: IFN-\(\gamma\) drives iNOS BH₄ consumption; oxidative stress converts BH₄ to BH₂. Outputs: serotonin and catecholamine levels to neurotransmitter models; NO to endothelial function (eNOS model, Coagulation and Microvascular Dynamics), cerebrovascular regulation (Cerebral Blood Flow Autoregulation), and coagulation balance.
Scope and rationale. The model tracks aggregate BH₄/BH₂ rather than tissue-specific pools. This is a significant simplification—brain, endothelial, and immune BH₄ pools may be independently regulated. The aggregate model captures the key insight (three-way competition from a shared cofactor) but cannot predict tissue-specific depletion patterns.
Falsification criteria. The model predicts that serotonin, catecholamine, and NO deficits correlate with each other and with inflammatory burden (IFN-\(\gamma\)). Falsified if: these deficits occur independently (e.g., serotonin depletion without catecholamine depletion in the same patient), indicating that the shared BH₄ bottleneck is not the dominant mechanism.
Clinical implications. Whom to treat: patients with low BH₄:BH₂ ratio and concurrent serotonin/catecholamine/NO deficits (the “multi-domain neurovascular signature”). How: (1) Sapropterin (exogenous BH₄) predicted to simultaneously improve mood, cognition, autonomic function, and vascular health; (2) anti-inflammatory therapy that reduces iNOS releases BH₄ for TPH/TH/eNOS without exogenous supplementation; (3) the model predicts that folinic acid (supports BH₂ \(->\) BH₄ recycling via DHFR) is a lower-cost alternative to sapropterin for mild BH₄ depletion. Key negative prediction: direct neurotransmitter replacement (SSRIs, methylphenidate) addresses symptoms but not the upstream bottleneck, predicting incomplete or diminishing response over time.
5 Sleep–Wake Cycle (Equations sleep wake)
Measurements required. (1) Polysomnography with sleep architecture analysis (why: slow-wave sleep percentage constrains \(r_{\text{decay}}\); sleep latency constrains \(\theta_{\text{on}}\)). (2) Actigraphy over 7–14 days (why: captures circadian rhythm regularity and sleep timing variability). (3) Melatonin onset time (dim-light melatonin onset, DLMO) (why: constrains circadian phase \(\phi_s\) and amplitude \(C_1\)).
Worked example. Healthy: \(r_{\text{build}} = 0.05\)/h, \(r_{\text{decay}} = 0.10\)/h, \(C_1 = 0.4\). After 16 h waking: \(S = S_{min} + (S_{max} - S_{min})(1 - e^{-0.05 \times 16}) = 0.55\). After 8 h sleep: \(S\) decays to $ 0.55 e^{-0.10 } = 0.25$ (below the circadian threshold, triggering waking). In ME/CFS with \(r_{\text{build}} = 0.08\)/h (faster pressure build due to energy deficit), \(r_{\text{decay}} = 0.06\)/h (impaired clearance), \(C_1 = 0.2\) (blunted circadian signal): after 16 h waking \(S = 0.72\) (high sleep pressure, falls asleep easily); after 8 h sleep \(S = 0.72 \times e^{-0.06 \times 8} = 0.44\) (residual sleep pressure, wakes unrefreshed). The patient experiences excessive daytime sleepiness (\(S = 0.44\) vs. 0.25 healthy upon waking) despite adequate sleep duration.
Inter-model dependencies. Inputs: ATP consumption rate affects adenosine accumulation (sleep pressure build rate); HPA circadian amplitude affects \(C_1\); pain from sensitization model disrupts sleep onset. Outputs: sleep quality modulates glymphatic clearance, cortisol rhythm, and next-day energy budget.
Scope and rationale. The Borbély two-process model is the standard quantitative framework for sleep regulation. It omits REM/NREM cycling, specific neurotransmitter dynamics (orexin, GABA), and sleep microarchitecture. These omissions are acceptable for predicting the ME/CFS sleep phenotype (unrefreshing sleep, excessive daytime sleepiness) but insufficient for modeling specific sleep disorders (narcolepsy, sleep apnea).
Falsification criteria. The model predicts that unrefreshing sleep in ME/CFS is driven by reduced \(r_{\text{decay}}\) (impaired sleep pressure clearance) and elevated \(r_{\text{build}}\) (faster accumulation). Falsified if: polysomnography shows normal slow-wave sleep quantity and quality in patients with severe unrefreshing sleep (would indicate that the subjective experience is driven by central perception rather than objective sleep physiology).
Clinical implications. Whom to treat: all ME/CFS patients with unrefreshing sleep (the vast majority). How: the model predicts that sleep hygiene (regular schedule to preserve circadian amplitude \(C_1\)) is a necessary but insufficient intervention. Improving mitochondrial function (reducing adenosine accumulation rate, i.e., \(r_{\text{build}}\)) is predicted to improve sleep quality without sleep-targeted medication—a testable prediction. For patients with severely blunted \(C_1\): melatonin at DLMO-appropriate timing, bright light therapy in the morning. The model predicts that sedative medications (benzodiazepines, Z-drugs) increase total sleep time but do not address \(r_{\text{decay}}\), explaining why they rarely improve the “unrefreshing” quality.
6 Central Sensitization (Equations central sensitization and sfn dynamics)
Measurements required. (1) Quantitative sensory testing (QST): pressure pain threshold, temporal summation, conditioned pain modulation (why: temporal summation directly measures wind-up/\(k_{\text{wind}}\); conditioned pain modulation measures descending inhibition/\(k_{\text{resolve}}\)). (2) Skin biopsy with intraepidermal nerve fiber density (IENFD) count (why: directly measures \(\mathcal{F}\), small fiber density). (3) \(\beta\)-endorphin levels (why: constrains endogenous opioid modulation of \(k_{\text{resolve}}\)).
Worked example. A patient with moderate central sensitization (\(\mathcal{S} = 0.6\) on a 0–1 scale) and mild small fiber neuropathy (\(\mathcal{F}/\mathcal{F}_0 = 0.7\), 30% fiber loss). Pain amplification factor: \((1 + \alpha_{\text{sens}} \times 0.6) = 1.9\) (using \(\alpha_{\text{sens}} = 1.5\)). Nociceptive input increased by fiber loss: \(P_{\text{noci}} = P_0 \times (1 + 1.5 \times 0.3) = 1.45 P_0\). Combined pain score: $ 1.9 = 2.76 $ the pain expected from the inflammatory input alone. The bidirectional feedback predicts that reducing \(\mathcal{S}\) by 50% (e.g., via LDN) reduces pain to $ 1.45 = 2.10 $—a 24% improvement—and secondarily improves sleep by 10–15% and reduces sympathetic tone by 8–12% through the pain \(->\) ANS \(->\) sleep pathway.
Inter-model dependencies. Inputs: pro-inflammatory cytokines and microglial activation (drive spinal glial activation \(\mu_{1,\text{spinal}}\)); ROS and autoantibodies (drive small fiber degeneration); endorphins (modulate resolution rate). Outputs: pain feeds back to sympathetic tone (ANS model), sleep disruption (sleep model), and cognitive energy demand.
Scope and rationale. The base model tracks an aggregate sensitization state \(\mathcal{S}\) rather than specific dorsal horn circuits. The multi-compartment extension (Equations nerve sheath inflammation and total noci compartments) decomposes \(P_{\text{noci}}\) into four tissue-specific generators—SFN, nerve sheath, periarticular, and muscular—enabling prediction of pain distribution and temporal profile, which the aggregate model alone cannot.
Falsification criteria. The model predicts that pain reduction (lowering \(\mathcal{S}\)) should improve autonomic function and sleep quality through the feedback pathway, independent of direct sleep or autonomic interventions. Falsified if: effective analgesic treatment in ME/CFS patients (confirmed by QST normalization) does not improve HRV or sleep efficiency—would indicate that the pain \(->\) ANS \(->\) sleep coupling is weaker than modeled. The multi-compartment extension predicts that peripheral nerve blocks should transiently reduce pain in targeted regions even in centrally sensitized patients. Falsified if: diagnostic nerve blocks show no analgesic effect in patients with high \(\mathcal{N}_s\) or \(P_{\text{peri}}\)—would indicate that peripheral generators contribute less than modeled.
Clinical implications. Whom to treat: patients with elevated temporal summation on QST (indicating high \(k_{\text{wind}}\)) or reduced IENFD (\(\mathcal{F}/\mathcal{F}_0 < 0.8\)). How: (1) LDN operates through both immune modulation and endorphin upregulation—the model predicts analgesic onset (days) precedes anti-inflammatory onset (weeks); (2) for structurally maintained sensitization (high \(\mathcal{S}\) with low \(k_{\text{resolve}}/k_{\text{wind}}\)): neuromodulatory approaches (low-dose ketamine as NMDAR antagonist, transcranial stimulation); (3) the model predicts that treating the upstream inflammatory drive (reducing \(\mu_{1,\text{spinal}}\) and \(P_{\text{noci}}\)) is more effective than treating sensitization directly, because it addresses both the input and the feedback amplification simultaneously.
7 Neuroimmune Model Extensions
{{/* M31: Multi-scale neuroimmune DAG (Tier 1, cert 0.40) */}}
Certainty: 0.40. The Blitshteyn 2026 neuroimmune framework (Blitshteyn, Doherty, and Steinman 2026) proposes shared pathophysiology across POTS, ME/CFS, and Long COVID. A multi-scale DAG formalizes this into a testable network with four layers: (1) Genetic/Trigger layer: HLA-DRB115:01, post-infectious trigger (SARS-CoV-2, EBV), molecular mimicry epitope; (2) Autoimmune layer: GPCR autoantibody production (β2, M2, M4, α1), intrathecal synthesis, ganglionic AChR antibodies; (3) CNS layer: brainstem neuroinflammation (TSPO-PET+ dorsolateral medulla), baroreflex gain resetting via NTS GPCR internalization, cholinergic anti-inflammatory pathway blockade, neurovascular coupling failure; (4) Clinical output layer*: orthostatic HR increment, CBF, fatigue, PEM, pain, cognitive dysfunction.
Canonical paths. Path A: GPCR AAb → baroreflex resetting → orthostatic tachycardia. Path B: GPCR AAb → CAP blockade → systemic inflammation + vagal withdrawal → fatigue. Path C: Intrathecal AAb → brainstem neuroinflammation → central sensitization → PEM amplification.
Testable prediction. DAG centrality analysis identifies GPCR autoantibody production and brainstem neuroinflammation as the two highest-betweenness nodes — interventions targeting either node produce the largest downstream effects. Combining both (immunoadsorption + taVNS) should produce supra-additive effects, testable in a 2×2 factorial trial.
Existing model context. Extends ch53 DAG with neuroimmune nodes; connects to ch53 integrated model.
Falsifiable prediction. DAG centrality will show GPCR AAb and brainstem neuroinflammation as top-3 by betweenness, each with ≥1.5× interaction with ≥2 other nodes. Falsified if centrality analysis identifies different top nodes or no node has ≥1.5× interaction count.
{{/* M28: Intrathecal AAb compartment ODE (Tier 2, cert 0.35) */}}
Certainty: 0.35. The existing GPCR AAb ODE (ch53) treats autoantibodies as a single plasma compartment. Adding a CSF compartment with intrathecal production captures the possibility of CNS-restricted autoantibody synthesis (from brainstem B cell aggregates, Germinal Center-Like B Cell Aggregates in Dorsolateral Medulla Driving Intrathecal GPCR Autoantibody Synthesis). Variables: A_p(t) = plasma titer, A_csf(t) = CSF titer. Kinetic terms: peripheral production by plasma cells, CNS crossing (k_brain_in, modulated by BBB permeability P_BBB), intrathecal production (k_intrathecal, active only if CNS B cells present), and CSF clearance. The model predicts: early disease: A_csf ~ k_brain_in·A_p (peripheral source dominates); established disease: intrathecal production dominates, A_csf/A_p ratio increases. This explains discordant IA response (rapid peripheral improvement despite slow CNS recovery — CSF autoantibodies clear slowly). (Blitshteyn, Doherty, and Steinman 2026)
Testable prediction. Fitted to paired serum+CSF autoantibody data (n=20, 3 timepoints over 6 months), the model shows: A_csf/A_p > 0.05 in ≥ 30% of ME/CFS patients; intrathecal production parameter k_intrathecal correlates with TSPO-PET signal in dorsolateral medulla (r > 0.5). IA reduces A_p by ≥ 70% but A_csf by ≤ 30% within 2 weeks.
Existing model context. Extends ch53 GPCR AAb ODE; requires BBB permeability variable from ch53.
8 Nerve Sheath and Tissue Pain Compartments (Equations nerve sheath inflammation and total noci compartments)
Measurements required. (1) Tryptase and histamine (plasma or 24-hour urine) (why: constrains mast cell activation input \(H_{\text{mc}}\) for both nerve sheath and periarticular compartments). (2) Nerve conduction studies and neuromuscular ultrasound (why: identifies nerve trunk inflammation \(\mathcal{N}_s\) vs. distal SFN \(\mathcal{F}\); enlarged nerves on ultrasound suggest endoneurial edema). (3) Venous lactate at rest and post-exertion (why: constrains \([\text{Lac}]_m\) in the muscular compartment). (4) Serum NGF (why: constrains the TRPV1 upregulation term \(\alpha_{\text{NGF}}\) in periarticular input). (5) Postural pain variation diary (why: endoneurial ischemia worsens upright; if pain increases with standing and improves supine, the nerve sheath compartment with its \(O_{2,\text{endo}}\)/MAP coupling is likely dominant).
Worked example. A patient with MCAS overlap (\(H_{\text{mc}} = 3 \times\) normal), moderate oxidative stress (\([\text{ROS}] = 1.5 \times\) normal), mild SFN (\(\mathcal{F}/\mathcal{F}_0 = 0.8\)), and central sensitization (\(\mathcal{S} = 0.5\)). Using \(\rho_{\text{mc,joint}} = 20\) (joint mast cell density ratio):
- SFN input: \(P_0 \times (1 + 1.5 \times 0.2) = 1.3 P_0\)
- Nerve sheath: \(\gamma_{\text{sheath}} \cdot \mathcal{N}_s \approx 0.4 P_0\) (moderate, driven by ROS + mast cells)
- Periarticular: \(\rho_{\text{mc,joint}} \cdot H_{\text{mc}} \cdot (1 + \alpha_{\text{NGF}} \cdot [\text{NGF}]) \approx 0.8 P_0\) (high, driven by MCAS)
- Muscular: $ 0.3 P_0$ at rest (mild metabolic stress)
- Total \(P_{\text{noci,total}} = (1.3 + 0.4 + 0.8 + 0.3) P_0 = 2.8 P_0\)
- After central sensitization amplification: \((1 + 1.5 \times 0.5) \times 2.8 = 4.9 P_0\)
Treatment with mast cell stabilizer (reducing \(H_{\text{mc}}\) by 60%): periarticular drops to $ 0.32 P_0$, nerve sheath drops to $ 0.28 P_0$. New total: \((1.3 + 0.28 + 0.32 + 0.3) \times 1.75 = 3.85 P_0\)—a 21% pain reduction from mast cell stabilization alone, targeting the dominant compartments without directly addressing sensitization.
Inter-model dependencies. Inputs: ROS from oxidative stress model; mast cell activation \(H_{\text{mc}}\) from immune model; NO from BH₄ model (endoneurial perfusion); MAP from ANS model (postural perfusion); lactate and ATP from energy metabolism model (muscular compartment); joint stability from hEDS model (periarticular compartment). Outputs: total \(P_{\text{noci,total}}\) feeds into central sensitization wind-up (Equation central sensitization) and thence to the pain \(->\) ANS \(->\) sleep feedback cascade.
Scope and rationale. The four-compartment decomposition captures the clinically relevant distinction between pain types (nerve trunk vs. joint vs. muscle vs. distal neuropathic) while remaining tractable. It omits visceral pain (relevant to IBS comorbidity) and headache-specific mechanisms (trigeminovascular system). Each compartment uses at most two state variables, keeping the total model dimensionality manageable.
Falsification criteria. The model predicts that pain temporal profile correlates with compartment dominance: (1) fast fluctuation with mast cell episodes \(->\) periarticular/nerve sheath dominant; (2) 48 h post-exertional decay \(->\) muscular dominant; (3) slow progressive worsening \(->\) SFN dominant. Falsified if: longitudinal pain phenotyping shows no correlation between temporal dynamics and the biomarkers predicted to drive each compartment (e.g., if mast cell markers do not predict periarticular pain fluctuation).
Clinical implications. Whom to treat: all ME/CFS patients with significant pain. How: (1) Pain diary analysis (temporal pattern, distribution, postural variation) identifies likely dominant compartment(s); (2) MCAS-dominant patients (periarticular + nerve sheath): mast cell stabilizers (cromolyn, ketotifen), PEA, antihistamines; (3) metabolic-dominant patients (muscular): strict pacing, CoQ10 + NADH (improve \(dot(V) O_{2,\text{musc}}\)); (4) SFN-dominant patients: neuroprotection (alpha-lipoic acid), IVIG if autoimmune markers present; (5) all patients benefit from central sensitization reduction (LDN, NMDA antagonists) because \(\mathcal{S}\) acts as a gain multiplier on all compartments.