Reverse Cascade Recovery Prediction
If disease onset follows a characteristic cascade through the timescale hierarchy (fast variables perturbed first, slow variables consolidating later), then recovery under effective treatment should follow the reverse temporal order: the fastest-responding variables normalize first, and the slowest-responding variables normalize last.
The predicted recovery sequence, derived directly from The Disease State ODE System:
\[ \text{Ca}^{2+} \text{signaling} &\rightarrow \text{ATP/Complex I} \rightarrow \text{Safe mode} \rightarrow \text{Immune markers} \\ &\rightarrow \text{Autoantibodies} \rightarrow \text{Epigenetic marks} \]
with each step requiring the timescale \(\tau\) of the corresponding variable for substantial normalization. This sequence assumes that the net effect of epigeneic changes is reversible by passive demethylation — i.e., the dominant pathogenic methylation changes are gains that fade after disease-state signals are removed. The vector model (Per-Locus Dynamics: Vector Model for Bidirectional Methylation) complicates this: loci above \(m_i^\text{crit}\) may passively revert, but loci below \(m_i^\text{crit}\) (ProB repeats after DNMT3B redistribution) require active, targeted remethylation on a slower timescale. Epigenetic normalization may therefore need to occur earlier in the recovery sequence for the loss-dominant loci (remethylation at ProB repeats reduces the inflammatory signal \(C_\text{pro}\)) and later for the gain-dominant loci (passive demethylation after inflammation resolves). The simple “epigenetic marks come last” prediction is locus-dependent.
This prediction is non-trivial. Nonlinear dynamics can in principle disrupt simple timescale ordering—if the system has strong coupling between fast and slow variables, recovery order might not follow the onset order in reverse. The prediction is therefore a test of whether the timescale separation is clean enough to dominate over nonlinear coupling effects.
In patients responding to effective treatment, biomarker normalization follows the reverse of the onset cascade: calcium signaling markers normalize within days, energy markers within weeks, safe mode indicators within months, immune markers over months, autoantibodies over months to a year, and epigenetic marks over years. Deviation from this predicted order indicates the treatment is acting on a non-primary target or that a load-bearing lock is blocking progression.
Certainty: 0.35. The prediction follows logically from the ODE timescale hierarchy, but has never been observed or tested in ME/CFS. Nonlinear dynamics could disrupt the simple reversal ordering, and real biological systems may not respect the clean timescale separation assumed in the model.
Testable predictions:
- In patients responding to daratumumab (targeting autoantibodies), the temporal sequence of improvement should be: autoantibody decline (weeks) → immune marker normalization (months) → energy improvement (months) → symptom resolution (months), with epigenetic normalization lagging by years.
- Recovery “stalls” (plateau without further improvement) occur when the next-in-sequence parameter is a load-bearing lock that is not being addressed by the current treatment.
- Deviation from predicted order—e.g., energy markers improving before immune markers—indicates the treatment acts directly on the energy subsystem rather than through immune normalization.
Limitations: Testing requires longitudinal multi-biomarker monitoring during treatment, which is expensive and not standard in current ME/CFS trials. The prediction assumes relatively clean timescale separation; if the ODE system has strong fast–slow coupling, the simple reversal may not hold. The prediction assumes that the net effect of pathogenic methylation is gain (reversible by passive demethylation after inflammation resolves); loci below \(m_i^\text{crit}\) (ProB repeats after DNMT3B redistribution) require active remethylation on a different timescale and may reverse earlier or later depending on the locus class (see Per-Locus Dynamics: Vector Model for Bidirectional Methylation). The Rekeland et al. cyclophosphamide trial (Rekeland et al. 2020) tracked some longitudinal biomarkers but not with the temporal resolution needed to test recovery ordering.