Connective Tissue Subsystem Extensions
The integrated model treats connective tissue as a single uniform compartment, yet clinical observations in ME/CFS with hEDS show differential symptom patterns across tissue types (vascular fragility, cervical ligament instability, joint hypermobility, skin extensibility). A multi-compartment ECM model captures this heterogeneity by assigning tissue-specific turnover rates and coupling strengths to four key compartments: vascular (\(\text{ECM}_{\text{vas}}\)), cervical ligament (\(\text{ECM}_{\text{cerv}}\)), tendon (\(\text{ECM}_{\text{tend}}\)), and dermal (\(\text{ECM}_{\text{derm}}\)). Each compartment follows coupled dynamics:
\[ \frac{d [\text{ECM}_{i}]}{d t} = k_\text{ECM,synth}^i - k_\text{ECM,deg}^i \cdot [\text{MMP}_{i}] \cdot [\text{ECM}_{i}], \quad i \in {\text{vas}, \text{cerv}, \text{tend}, \text{derm}} \]
where synthesis rates \(k_\text{ECM,synth}^i\) and MMP-dependent degradation rates \(k_\text{ECM,deg}^i\) vary by tissue. Vascular ECM has the fastest turnover (hours to days) and is most sensitive to inflammatory MMP induction; cervical ligaments have the slowest turnover (months to years) and are most vulnerable to cumulative strain accumulation; tendons have intermediate turnover; dermal ECM is moderately fast with distinct collagen type ratios. The MMP activity \(\text{MMP}_{i}\) in each compartment responds to both systemic signals (IL-6, TNF-α from the cytokine network) and local stimuli (mechanical strain \(\sigma_i\), mast cell activation \(\text{MC}_{d}\)):
\[ \text{MMP}_{i} = \text{MMP}_{0} \cdot (1 + \alpha_\text{inf} \cdot [\text{IL-6}] + \alpha_i_\text{strain} \cdot \sigma_i + \alpha_i_\text{MC} \cdot \text{MC}_{d}) \]
The tissue-specific strain coefficients \(\alpha_i_\text{strain}\) reflect differing mechanotransduction sensitivity: cervical ligaments are highly strain-sensitive (high \(\alpha_\text{cerv}^\text{strain}\)) and accumulate MMP under minor orthostatic stress, explaining POTS exacerbation in hEDS patients; tendons respond to activity-related strain; vascular ECM responds to shear stress and pulsatile pressure. Mast cell coupling \(\alpha_i^\text{MC}\) varies by tissue density: gut mucosa has high mast cell density and high coupling, driving SIBO-mediated ECM degradation. The model predicts that symptom clusters correlate with dominant compartments: patients with dominant vascular ECM instability present with orthostatic intolerance and syncope; cervical ligament dominance presents with neck pain, headaches, and dysautonomia; tendon dominance presents with joint instability and dislocations; dermal dominance presents with skin extensibility and wound healing deficits. Treatment strategies can be compartment-targeted: strengthening cervical support (physical therapy for deep neck flexors) reduces \(\sigma_\text{cerv}\) and MMP activation in that compartment; beta-blockers reduce pulsatile pressure on vascular ECM; mast cell stabilizers reduce \(\alpha_i^\text{MC}\) across compartments.
Certainty: 0.45. Tissue-specific ECM turnover rates are well-established in connective tissue biology, and the differential clinical presentation in hEDS is documented. The novel application to ME/CFS symptom patterning requires validation that: (1) ME/CFS patients with different symptom clusters show corresponding differences in ECM turnover markers (e.g., collagen degradation products specific to each tissue type); (2) compartment-specific MMP activity can be measured non-invasively (e.g., imaging biomarkers for vascular ECM stiffness); and (3) interventions targeting specific compartments produce the predicted symptom-selective improvements. The cross-compartment coupling via systemic inflammation and mast cell activation suggests that effective treatment requires addressing both local strain drivers and systemic amplifiers.
The HIF-1alpha-ECM coupled differential equations proposed in Chapter Energy Metabolism Models extend the energy metabolism model with hypoxia sensing and connective tissue degradation. This triadic coupling—HIF-1alpha induction by ROS, ECM degradation via HIF-1alpha-mediated MMP induction, and impaired energy production from reduced perfusion due to ECM degradation—creates a slow positive feedback loop operating on timescales of weeks to months. The loop architecture mirrors the energy–immune vicious loop (Section Energy–Immune Coupling) but with distinct timescales and tissue specificity: the energy–immune loop operates over hours to days and affects systemic inflammation, while the HIF-1alpha-ECM loop operates over weeks to months and produces structural tissue changes. This dual-loop structure predicts a “two-stage” progression model: early ME/CFS is dominated by the energy–immune loop (symptom flares, PEM, relapses), while late ME/CFS incorporates the HIF-1alpha-ECM loop (connective tissue degradation, vascular compliance changes, progressive symptom consolidation). The transition between stages depends on cumulative ROS exposure and ECM quality trajectories, which can be modeled by extending the damage accumulation equation (damage accumulation) to include an ECM damage component with slower repair kinetics. Interventions targeting HIF-1alpha stabilization (e.g., PHD activators) or ECM repair (specific collagen support) may be particularly relevant in late-stage disease where the structural loop is dominant, whereas early-stage disease may respond more to energy–immune cycle disruption (e.g., immunomodulation, mitochondrial support). This stage-specific treatment strategy emerges from the coupled multi-timescale model and is not apparent when considering either loop in isolation.
Certainty: 0.45. The HIF-1alpha-ECM coupling mechanism is biochemically validated in tendinopathy and other connective tissue disorders (Moschini, Mohanan, et al. 2026). The application to ME/CFS disease staging requires validation that: (1) ME/CFS patients show elevated HIF-1alpha and ECM degradation markers relative to healthy controls; (2) the temporal pattern of these markers (slow increase over months) differs from inflammatory markers (fast fluctuations over hours to days); and (3) clinical severity correlates more strongly with ECM markers in late-stage disease and with inflammatory markers in early-stage disease. If confirmed, the HIF-1alpha-ECM loop would explain why ME/CFS tends to worsen over time despite apparent symptom stability: slow structural degradation accumulates beneath the surface of fluctuating acute symptoms.
1 Neuroimmune Integration — GPCR Autoantibody and Vagal CAP Extensions
{{/* M29: CAP ODE vagus-immune axis (Tier 2, cert 0.35) */}}
Certainty: 0.35. Add a 4-variable CAP subsystem: V_eff(t) = vagal efferent activity, A_ch(t) = ACh at splenic T cell synapse, T_act(t) = activated splenic CD4+ T cell fraction, M_TNF(t) = macrophage TNF-α production. V_eff is driven by baroreflex output (existing S/V from ch52) and external inputs (taVNS, SPB). A_ch is a function of V_eff modulated by M2/M4 autoantibody blockade. T_act depends on A_ch via β2-AR, modulated by β2-AR autoantibodies. M_TNF is suppressed by T_act (the CAP efferent arm). Key parameter: α*β2 = 1/(1 + [AAb*β2]/K_d) reduces CAP gain. Above a critical β2-AR AAb threshold, the CAP is completely disabled — taVNS produces no TNF-α suppression. Below threshold, CAP gain increases with V_eff — taVNS is effective. (Blitshteyn, Doherty, and Steinman 2026)
Testable prediction. The relationship between β2-AR AAb titer and taVNS-induced TNF-α suppression is sigmoidal (not linear) with a critical threshold at ~2 × ULN. Above threshold: taVNS suppresses TNF-α by < 10%. Below threshold: by > 30%. Stratify ME/CFS patients by β2-AR AAb quartile and measure acute taVNS TNF-α suppression — the quartile × suppression interaction follows a sigmoid, not a linear gradient.
Existing model context. Extends ch53 vagal coupling (vagal coupling); connects to ch52 baroreflex model.
{{/* M30: GPCR binding-internalization ODE (Tier 3, cert 0.30) */}}
Certainty: 0.30. Extend the GPCR AAb ODE (ch53) with three-state receptor dynamics: free R(t), AAb-occupied R(t), internalized R_int(t). Signaling output S(t) depends on whether AAb is agonistic or antagonistic. Binding: d[AAb·R]/dt = k_on·AAb·R − k_off·[AAb·R] − k_int·[AAb·R]. Internalization rate k_int depends on β-arrestin bias (epitope-specific). The model captures the cell-based functional data from Fedorowski 2017/Kharraziha 2020: different autoantibody profiles produce different signaling/internalization ratios. (Fedorowski et al. 2017) (Kharraziha et al. 2020)
Testable prediction. Patients with internalization-biased autoantibodies (rapid k_int) have worse autonomic function (fewer functional receptors). Immunoadsorption removes AAb → R recovers on timescale of days (receptor recycling from intracellular pools). Recovery is faster for receptors with high recycling rates (β2-AR recycles quickly, M3 slowly). The model explains variable autonomic improvement after IA.
Existing model context. Extends ch53 GPCR AAb ODE; requires receptor-specific kinetics.
Falsifiable prediction. RCT (n≥40, IA vs sham) will show COMPASS-31 between-group difference ≥15 points and receptor recovery time under 14 days for β2-AR. Falsified if COMPASS-31 difference under 10 points or receptor recovery ≥21 days.
Certainty: 0.25. The Kang et al.(Kang et al. 2026) data provide quantitative cytokine reduction time-courses and cognitive recovery rates suitable for ODE parameterisation. An exosome variable E_exo(t) crosses the BBB with rate constant k_BBB, reduces neuroinflammation I_n(t) with efficacy eta_hsp, and enhances glymphatic clearance G(t). The resulting 3-variable ODE system (E_exo, I_n, G) would predict: (a) the dose-frequency regimen achieving sustained neuroinflammation reduction with minimised fluctuation amplitude, (b) the therapeutic window before chronic inflammation alters BBB EV trafficking kinetics, and (c) the optimal cargo mRNA species for each patient subgroup based on their dominant neuroinflammatory cytokine profile. The model is parameterisable from Kang et al. cytokine time-course data but has not been implemented.
Certainty: 0.20. EV BBB crossing is regionally heterogeneous (hippocampus vs cortex vs brainstem) and temporally variable (inflammation flares) (Kang et al. 2026). A stochastic model with spatially varying rate constants k_BBB(x,t) would predict regional variance in exosome accumulation matching experimental CV > 0.4. This would inform dosing protocols but has no parameterisation data for ME/CFS-specific BBB heterogeneity.
Certainty: 0.25. A PK/PD model linking exosome plasma concentration C(t), CNS accumulation C_cns(t), and downstream cytokine modulation I(t) could predict optimal dosing schedules for sustained neuroinflammation suppression. Kang et al.(Kang et al. 2026) cytokine time-course data provide initial parameterisation. Predicted outcomes: thrice-weekly dosing maintains neuroinflammation suppression with under 15% fluctuation. Validatable in mouse studies; no human PK parameters exist.
2 Brain Clearance Architecture: Multi-Compartment Glymphatic Model
Certainty: 0.30 (model formalism; parameter values from mouse data).
The paper’s existing glymphatic framework treats clearance as a single-compartment process. Chayama et al. (2026) (Chayama et al. 2026) demonstrate that brain-derived proteins drain through anatomically distinct compartments (dorsal dura, dorsal skull, basal skull, cribriform plate, nasal cavity) with different kinetics (\(k_{\text{skull}} = -0.008\), \(k_{\text{dura}} = 0.127--0.261\)). A multi-compartment ODE model can capture this architecture:
State variables:
- \(Q_{\text{dorsal}}(t)\): waste in dorsal drainage compartment (cortex → dorsal dura/skull)
- \(Q_{\text{basal}}(t)\): waste in basal drainage compartment (striatum → basal skull/nasal)
- \(Q_{\text{skull}}(t)\): waste in skull marrow compartment (slow outflow)
- \(Q_{\text{nasal}}(t)\): waste in nasal/cribriform compartment (rapid turnover)
- \(Q_{\text{blood}}(t)\): waste leaked into bloodstream (inflammation-induced rerouting)
Governing equations: \[ (d Q_{\text{dorsal}})/(d t) = G_{\text{dorsal}} - k_{\text{dorsal}} dot Q_{\text{dorsal}} + k_{\text{inflam,dorsal}}(t) dot Q_{\text{blood}} \] \[ (d Q_{\text{basal}})/(d t) = G_{\text{basal}} - k_{\text{basal}} dot Q_{\text{basal}} + k_{\text{inflam,basal}}(t) dot Q_{\text{blood}} \] \[ (d Q_{\text{skull}})/(d t) = G_{\text{skull}} - k_{\text{skull}} dot Q_{\text{skull}} \] \[ (d Q_{\text{nasal}})/(d t) = G_{\text{nasal}} - k_{\text{nasal}} dot Q_{\text{nasal}} \] \[ (d Q_{\text{blood}})/(d t) = \sum_{\text{i}} k_{\text{inflam,i}}(t) dot Q_{\text{i}} - k_{\text{clear,blood}} dot Q_{\text{blood}} \]
ME/CFS parameter modifications:
- \(k_{\text{dorsal}}, k_{\text{basal}}, k_{\text{skull}}, k_{\text{nasal}}\): reduced from normal (glymphatic impairment)
- \(k_{\text{inflam,i}}(t)\): spikes during PEM (inflammation function of cytokine levels)
- \(G_i\): may be normal or elevated (mitochondrial dysfunction increases waste per ATP produced)
Testable predictions: 1. Regional symptom patterns (executive vs. memory vs. sensorimotor brain fog) correlate with compartment-specific \(k_i\) impairment 2. PEM severity correlates with peak \(k_{\text{inflam,i}}(t)\) magnitude 3. Treatment effect heterogeneity predicted by which compartment’s \(k_i\) improves (NE-targeted: dorsal/advective; AQP4-targeted: diffuse/slow; anti-inflammatory: reduces \(k_{\text{inflam}}\))
Limitations: Compartment kinetic constants extrapolated from mouse data. Human compartment boundaries may differ. Requires MR-AIV compartment-resolved velocity measurements for parameter estimation, which do not yet exist for ME/CFS.
Certainty: 0.25 (model formalism; spatial parameterization requires Toscano 2026 MR-AIV data).
Toscano et al. (2026) identified two glymphatic transport modes: fast advective perivascular flow (\(\sim\) 3 \(\mu\)m/s, LC-NE vasomotion-driven) and slow diffusive interstitial transport (\(\sim\) 0.1 \(\mu\)m/s, AQP4-mediated). Chayama et al. (Chayama et al. 2026) add compartmentalized exit routes. Combining these into a spatial advection-diffusion PDE:
\[ (\partial C_{\text{fast}})/(\partial t) + \nabla dot (v_{\text{fast}} dot C_{\text{fast}}) = -k_{\text{ex}} dot (C_{\text{fast}} - C_{\text{slow}}) + S_{\text{fast}} \] \[ (\partial C_{\text{slow}})/(\partial t) = \nabla dot (D_{\text{slow}} \nabla C_{\text{slow}}) + k_{\text{ex}} dot (C_{\text{fast}} - C_{\text{slow}}) - k_{\text{exit}} dot C_{\text{slow}} + S_{\text{slow}} \]
Where \(v_{\text{fast}}(x,t)\) is the advective velocity (NE oscillation amplitude-dependent), \(D_{\text{slow}}\) is the effective diffusion coefficient (AQP4-dependent), and \(k_{\text{exit}}\) is the compartment-specific exit rate from the slow compartment to border drainage routes (incorporating Yang/Chayama compartmental kinetics).
ME/CFS modifications:
- \(v_{\text{fast}}\) reduced (LC-NE oscillation impairment, Section Neuronal Protein Tracing Reveals Physiological Brain Clearance Architecture)
- \(D_{\text{slow}}\) reduced (AQP4 depolarization, Section Glymphatic Dysfunction and Brain Waste Accumulation)
- \(k_{\text{exit}}\) compartment-specific reduction depending on which border compartment is impaired
Predictions: 1. Advective-dominant impairment patients (low \(v_{\text{fast}}\), preserved \(D_{\text{slow}}\)) respond to NE-targeted treatments 2. Diffusive-dominant impairment patients (low \(D_{\text{slow}}\), preserved \(v_{\text{fast}}\)) respond to AQP4-targeted treatments 3. Dual impairment requires combination therapy
Limitations: Requires 3D MR-AIV velocity field data for spatial parameter estimation. PDE discretization introduces numerical errors. No ME/CFS spatial glymphatic data exist.
Certainty: 0.25 (model formalism; parameter estimation requires PEM blood NfL data that do not yet exist).
Chayama et al. (Chayama et al. 2026) demonstrate that acute inflammation (LPS) causes a switch in clearance routing from normal border drainage to bloodstream leakage. In ME/CFS, PEM involves a cytokine surge that may trigger this same routing disruption. A bifurcation model captures the threshold-dependent nature of this switch:
State variable: \(R_i(t)\) = routing state for compartment \(i\) (\(R_i = 0\): normal border drainage; \(R_i = 1\): bloodstream leakage) Driver: \(I(t)\) = systemic inflammation level (IL-6, TNF-alpha concentration)
\[ (d R_i)/(d t) = \alpha dot \sigma(I(t) - I_{\text{thresh,i}}) dot (1 - R_i) - \beta dot R_i \]
where \(\sigma(z) = 1/(1 + e^{-k_{\text{thresh}} dot z})\) is a sigmoid switching function, \(I_{\text{thresh,i}}\) is the inflammation threshold for routing disruption in compartment \(i\), \(\alpha\) is the activation rate, and \(\beta\) is the recovery rate.
ME/CFS implications:
- \(I_{\text{thresh,i}}\) may be reduced in ME/CFS (sensitized BBB, chronic neuroinflammation)
- \(\beta\) may be reduced (slow recovery, consistent with 70x structural recovery timescale)
- Hysteresis possible: \(R_i\) may remain elevated after \(I(t)\) returns to baseline
- Repeated PEM episodes may cause chronic \(R_i\) elevation (structural remodeling of clearance routes)
Falsifiable prediction: Single PEM episode causes transient routing disruption (\(R_i \rightarrow 1\)) with recovery time tau = 1/beta. Repeated PEM reduces \(I_{\text{thresh}}\) (sensitization). Anti-inflammatory pretreatment raises \(I_{\text{thresh}}\), preventing routing disruption.
Limitations: LPS threshold data from acute mouse model; human PEM cytokine magnitudes may be orders of magnitude below the LPS threshold. Bifurcation structure is speculative. No human PEM routing disruption data exist.
3 Brain Clearance Architecture
Certainty: 0.30 (model formalism; parameter values from mouse data).
The paper’s existing glymphatic framework treats clearance as a single-compartment process. Chayama et al. (2026) (Chayama et al. 2026) demonstrate that brain-derived proteins drain through anatomically distinct compartments (dorsal dura, dorsal skull, basal skull, cribriform plate, nasal cavity) with different kinetics (\(k_{\text{skull}} = -0.008\), \(k_{\text{dura}} = 0.127--0.261\)). A multi-compartment ODE model can capture this architecture:
State variables:
- \(Q_{\text{dorsal}}(t)\): waste in dorsal drainage compartment (cortex → dorsal dura/skull)
- \(Q_{\text{basal}}(t)\): waste in basal drainage compartment (striatum → basal skull/nasal)
- \(Q_{\text{skull}}(t)\): waste in skull marrow compartment (slow outflow)
- \(Q_{\text{nasal}}(t)\): waste in nasal/cribriform compartment (rapid turnover)
- \(Q_{\text{blood}}(t)\): waste leaked into bloodstream (inflammation-induced rerouting)
Governing equations: \[ (d Q_{\text{dorsal}})/(d t) = G_{\text{dorsal}} - k_{\text{dorsal}} dot Q_{\text{dorsal}} + k_{\text{inflam,dorsal}}(t) dot Q_{\text{blood}} \] \[ (d Q_{\text{basal}})/(d t) = G_{\text{basal}} - k_{\text{basal}} dot Q_{\text{basal}} + k_{\text{inflam,basal}}(t) dot Q_{\text{blood}} \] \[ (d Q_{\text{skull}})/(d t) = G_{\text{skull}} - k_{\text{skull}} dot Q_{\text{skull}} \] \[ (d Q_{\text{nasal}})/(d t) = G_{\text{nasal}} - k_{\text{nasal}} dot Q_{\text{nasal}} \] \[ (d Q_{\text{blood}})/(d t) = \sum_{\text{i}} k_{\text{inflam,i}}(t) dot Q_{\text{i}} - k_{\text{clear,blood}} dot Q_{\text{blood}} \]
ME/CFS parameter modifications:
- \(k_{\text{dorsal}}, k_{\text{basal}}, k_{\text{skull}}, k_{\text{nasal}}\): reduced from normal (glymphatic impairment)
- \(k_{\text{inflam,i}}(t)\): spikes during PEM (inflammation function of cytokine levels)
- \(G_i\): may be normal or elevated (mitochondrial dysfunction increases waste per ATP produced)
Testable predictions: 1. Regional symptom patterns (executive vs. memory vs. sensorimotor brain fog) correlate with compartment-specific \(k_i\) impairment 2. PEM severity correlates with peak \(k_{\text{inflam,i}}(t)\) magnitude 3. Treatment effect heterogeneity predicted by which compartment’s \(k_i\) improves (NE-targeted: dorsal/advective; AQP4-targeted: diffuse/slow; anti-inflammatory: reduces \(k_{\text{inflam}}\))
Limitations: Compartment kinetic constants extrapolated from mouse data. Human compartment boundaries may differ. Requires MR-AIV compartment-resolved velocity measurements for parameter estimation, which do not yet exist for ME/CFS.
Certainty: 0.25 (model formalism; spatial parameterization requires Toscano 2026 MR-AIV data).
Toscano et al. (2026) identified two glymphatic transport modes: fast advective perivascular flow (\(\sim\) 3 \(\mu\)m/s, LC-NE vasomotion-driven) and slow diffusive interstitial transport (\(\sim\) 0.1 \(\mu\)m/s, AQP4-mediated). Chayama et al. (Chayama et al. 2026) add compartmentalized exit routes. Combining these into a spatial advection-diffusion PDE:
\[ (\partial C_{\text{fast}})/(\partial t) + \nabla dot (v_{\text{fast}} dot C_{\text{fast}}) = -k_{\text{ex}} dot (C_{\text{fast}} - C_{\text{slow}}) + S_{\text{fast}} \] \[ (\partial C_{\text{slow}})/(\partial t) = \nabla dot (D_{\text{slow}} \nabla C_{\text{slow}}) + k_{\text{ex}} dot (C_{\text{fast}} - C_{\text{slow}}) - k_{\text{exit}} dot C_{\text{slow}} + S_{\text{slow}} \]
Where \(v_{\text{fast}}(x,t)\) is the advective velocity (NE oscillation amplitude-dependent), \(D_{\text{slow}}\) is the effective diffusion coefficient (AQP4-dependent), and \(k_{\text{exit}}\) is the compartment-specific exit rate from the slow compartment to border drainage routes (incorporating Yang/Chayama compartmental kinetics).
ME/CFS modifications:
- \(v_{\text{fast}}\) reduced (LC-NE oscillation impairment, Section Neuronal Protein Tracing Reveals Physiological Brain Clearance Architecture)
- \(D_{\text{slow}}\) reduced (AQP4 depolarization, Section Glymphatic Dysfunction and Brain Waste Accumulation)
- \(k_{\text{exit}}\) compartment-specific reduction depending on which border compartment is impaired
Predictions: 1. Advective-dominant impairment patients (low \(v_{\text{fast}}\), preserved \(D_{\text{slow}}\)) respond to NE-targeted treatments 2. Diffusive-dominant impairment patients (low \(D_{\text{slow}}\), preserved \(v_{\text{fast}}\)) respond to AQP4-targeted treatments 3. Dual impairment requires combination therapy
Limitations: Requires 3D MR-AIV velocity field data for spatial parameter estimation. PDE discretization introduces numerical errors. No ME/CFS spatial glymphatic data exist.
Certainty: 0.25 (model formalism; parameter estimation requires PEM blood NfL data that do not yet exist).
Chayama et al. (Chayama et al. 2026) demonstrate that acute inflammation (LPS) causes a switch in clearance routing from normal border drainage to bloodstream leakage. In ME/CFS, PEM involves a cytokine surge that may trigger this same routing disruption. A bifurcation model captures the threshold-dependent nature of this switch:
State variable: \(R_i(t)\) = routing state for compartment \(i\) (\(R_i = 0\): normal border drainage; \(R_i = 1\): bloodstream leakage) Driver: \(I(t)\) = systemic inflammation level (IL-6, TNF-alpha concentration)
\[ (d R_i)/(d t) = \alpha dot \sigma(I(t) - I_{\text{thresh,i}}) dot (1 - R_i) - \beta dot R_i \]
where \(\sigma(z) = 1/(1 + e^{-k_{\text{thresh}} dot z})\) is a sigmoid switching function, \(I_{\text{thresh,i}}\) is the inflammation threshold for routing disruption in compartment \(i\), \(\alpha\) is the activation rate, and \(\beta\) is the recovery rate.
ME/CFS implications:
- \(I_{\text{thresh,i}}\) may be reduced in ME/CFS (sensitized BBB, chronic neuroinflammation)
- \(\beta\) may be reduced (slow recovery, consistent with 70x structural recovery timescale)
- Hysteresis possible: \(R_i\) may remain elevated after \(I(t)\) returns to baseline
- Repeated PEM episodes may cause chronic \(R_i\) elevation (structural remodeling of clearance routes)
Falsifiable prediction: Single PEM episode causes transient routing disruption (\(R_i \rightarrow 1\)) with recovery time tau = 1/beta. Repeated PEM reduces \(I_{\text{thresh}}\) (sensitization). Anti-inflammatory pretreatment raises \(I_{\text{thresh}}\), preventing routing disruption.
Limitations: LPS threshold data from acute mouse model; human PEM cytokine magnitudes may be orders of magnitude below the LPS threshold. Bifurcation structure is speculative. No human PEM routing disruption data exist.