Hysteresis and the Intervention Window

Hysteresis occurs when the path from health to disease differs from the path from disease to health. In the ME/CFS model, hysteresis arises from two sources: (1) the saddle-node bifurcation structure of the energy–immune system (structural hysteresis), and (2) the epigenetic consolidation of disease parameters (parametric hysteresis, Section Extended Subsystem Couplings).

1 Structural Hysteresis

In a bistable system, the disease attractor appears at a critical parameter value \(\theta_\text{onset}\) (saddle-node bifurcation) but does not disappear until a different value \(\theta_\text{recovery} != \theta_\text{onset}\). The hysteresis width \(\Delta \theta = |\theta_\text{recovery} - \theta_\text{onset}|\) quantifies how much further the system must be pushed for recovery compared to onset. For the energy–immune subsystem, the model predicts:

\[ \Delta \alpha_\text{CI} = \alpha_\text{CI}^\text{recovery} - \alpha_\text{CI}^\text{onset} > 0 \tag{1}\]

meaning that Complex I activity must be restored to a level above the value at which disease was triggered, not merely to the trigger value. This asymmetry is invisible without the bifurcation diagram: verbal reasoning about positive feedback loops identifies that the disease is self-sustaining but cannot quantify how much intervention overshoot is required for recovery. The model computes \(\Delta \alpha_\text{CI}\) as a function of the feedback strengths, providing a quantitative recovery target.

2 Parametric Hysteresis and the Intervention Window

Epigenetic consolidation (Section Extended Subsystem Couplings) adds a second layer of hysteresis that increases over time. The methylation index \(\mathcal{M}\) evolves toward values that stabilize the disease attractor, progressively widening the hysteresis loop. The direction of pathogenic epigenetic change — gain vs loss — depends on the loci involved and the balance between DNMT3A/B-mediated hypermethylation and DNMT3B-redistribution-mediated hypomethylation at ProB repeats. The unified vector model in Chapter Formal Causal Hierarchy Analysis (Per-Locus Dynamics: Vector Model for Bidirectional Methylation) captures both directions; the scalar \(\mathcal{M}\) used here is a compressed summary (see Speculation Consolidation as Loss of Methylation for locus-class details). This predicts an intervention window: the period after disease onset during which the epigenetic contribution to hysteresis is still small, and recovery requires only modest intervention above the structural hysteresis threshold.

The model estimates the intervention window duration as the time required for \(\mathcal{M}\) to reach half its equilibrium disease value:

\[ \tau_\text{window} \approx \frac{ln 2}{k_\text{DNMT} \cdot overline(C)_\text{pro}} \tag{2}\]

where \(overline(C)_\text{pro}\) is the mean pro-inflammatory cytokine level in the disease state. For typical parameter values, \(\tau_\text{window} ~ 3\)–$ 12$ months, consistent with clinical observations that early intervention (within the first year of ME/CFS) is associated with better outcomes. After the intervention window closes, recovery requires both sufficient perturbation to cross the structural separatrix and sustained intervention to reverse epigenetic changes—a process operating on the slow timescale \(\tau_\text{epi}\).

This intervention window concept is a unique product of the coupled dynamical model: it emerges from the interaction of two timescales (fast state dynamics and slow epigenetic dynamics) that cannot be analyzed without formal mathematical treatment. It provides a quantitative basis for the clinical urgency of early diagnosis and treatment in ME/CFS.