How Infection Itself Fuels New Waves: A Hidden Feedback in Epidemics

1.5 million active infections aren’t just a public health crisis—they’re a distributed bioreactor generating mutations at scale. Each viral replication inside a host is a roll of the dice, and when millions play, the odds shift. A new mathematical model reveals that this isn’t just background noise: it’s a central driver of epidemic dynamics. When infection levels rise, they don’t merely spread—they rewrite the rules of immunity itself, accelerating their own return.

This feedback loop, formalized in the SIRCm model by Ramírez-Gutiérrez et al. (2026), shows how high prevalence fuels immune escape, destabilizing what should be a stable endemic state and triggering self-sustained waves of disease—even in the absence of seasonal changes or behavioral shifts. The implication? Recurrent outbreaks may not require external triggers. They can emerge from within the system.

The Science

Epidemiologists have long used compartmental models to simulate disease spread. The classic SIR model divides populations into Susceptible, Infected, and Recovered groups, assuming recovery confers lasting immunity. But reality is messier. Immunity wanes. Pathogens evolve. And when they do, previously protected individuals become vulnerable again—not because time passed, but because the virus changed.

To capture this, researchers developed multi-strain models that track antigenic drift over time. These are powerful but complex, often requiring assumptions about cross-immunity and strain replacement. The SIRCm model takes a different path: instead of resolving individual variants, it embeds the effect of mutation into a single-strain framework through a feedback mechanism.

It builds on the SIRC model (Casagrandi et al., 2006), which adds a “cross-immune” compartment (C) between full recovery (R) and full susceptibility (S). Individuals move from R → C → S as immunity wanes gradually. This captures partial protection and delayed re-infection.

SIRCm goes further. It introduces two mutation feedback parameters—${\epsilon }_1$ and ${\epsilon }_2$—that amplify the rates at which immunity fails. Specifically:

• ${\epsilon }_1$ scales the rate at which recovered individuals (R) lose immunity due to viral mutation,
• ${\epsilon }_2$ scales the rate at which cross-immune individuals (C) revert to susceptibility.

Crucially, these rates aren't constant. They depend on current infection levels through a function $g(x)$, which represents the opportunity for mutation. The authors explore two forms:

1. Transmission-driven: $g(x)=SI$, meaning mutation chances spike during new infections.
2. Prevalence-driven: $g(x)=I$, meaning mutation scales with the total number of infected individuals, regardless of transmission intensity.

In both cases, more infection means faster erosion of population immunity—not through biological decay, but through viral evolution. This creates a nonlinear feedback: higher prevalence → more mutations → more immune escape → more susceptible hosts → higher prevalence.

The model is described by four coupled differential equations tracking flows between compartments (see

for structure). Despite its simplicity, it exhibits rich dynamics, including oscillations, bistability, and bifurcations—all arising from internal feedback rather than external forcing.

What They Found

The first major result is existential: for any positive feedback strength (${\epsilon }_1,{\epsilon }_2\ge 0$), and whenever ${\mathcal{R}}_0>1$, an endemic equilibrium exists. Using persistence theory—a branch of dynamical systems that studies long-term survival of processes—the authors prove that infection cannot die out once established, provided the basic reproduction number exceeds one.

But existence doesn’t imply stability. And here lies the surprise.

In the standard SIRC model (where ${\epsilon }_1={\epsilon }_2=0$), the endemic equilibrium is stable. Once reached, prevalence settles into a steady state. Not so in SIRCm.

Using numerical continuation methods, the team mapped the stability of the endemic equilibrium across parameter space. They found that for intermediate values of mutation feedback, the equilibrium becomes unstable via a Hopf bifurcation—a mathematical tipping point where steady states give way to sustained oscillations.

Endemic Prevalence vs. Mutation Feedback Strength

Label	Value
0.0	0.0032
0.5	0.018
1.0	0.052
1.5	0.11
2.0	0.21
2.5	0.38
3.0	0.57
3.5	0.72

Data from Figure 2 (transmission-driven model). Parameters: \(\mu=0.02\), \(\alpha=365/3\), \(\delta=1.6^{-1}\), \(\sigma=0.07874\), \(\gamma=0.35\), \(\beta_0\) varied.

illustrates how endemic prevalence $I^{\ast }$ climbs as feedback strength increases along the diagonal ${\tilde{\epsilon }}_1={\tilde{\epsilon }}_2=\tilde{\epsilon }$. At low $\tilde{\epsilon }$, the system behaves like SIRC, with low endemic prevalence. At high $\tilde{\epsilon }$, it converges to an SIS-like limit where nearly everyone cycles rapidly through infection and re-infection.

But in between, something remarkable happens: the smooth transition breaks down. Instead of settling into a new steady state, the system begins to oscillate.

Stability Regimes Across Feedback Parameters

Label	Value
0.0	0
0.5	0.1
1.0	0.4
1.5	0.9
2.0	1.6
2.5	2
3.0	2.5
3.5	2.8

Hopf bifurcation boundary from Figure 4 (transmission-driven model). Inside the dashed curve: unstable endemic equilibrium; outside: stable.

shows the region in (${\tilde{\epsilon }}_1,{\tilde{\epsilon }}_2$) space where this occurs—the “instability window.” Within this zone, trajectories approach a limit cycle, meaning prevalence rises and falls periodically without damping. These are not transient fluctuations; they are permanent, self-sustained waves.

For the transmission-driven model, this instability window appears as a closed loop in parameter space (see

). Outside it, the endemic equilibrium is stable. Inside, the system oscillates indefinitely.

The size and location of this window depend on other parameters like transmission rate ${\beta }_0$ and partial infectivity $\sigma$. Higher ${\beta }_0$ widens the range of feedback strengths that induce oscillations, suggesting that more transmissible pathogens are more prone to self-driven recurrence.

Even more strikingly, these oscillations occur without seasonal forcing. Traditional models rely on external drivers—school terms, climate shifts, holiday gatherings—to explain annual or biennial epidemic cycles. SIRCm shows that internal feedback alone can generate such patterns.

Why This Changes Things

We’ve long known that viruses mutate. We’ve seen immune escape in real time with SARS-CoV-2’s Omicron variant, which evaded prior immunity so effectively it sparked global waves despite high vaccination rates. But we’ve treated these events as exogenous shocks—surprising, unpredictable, driven by biology outside the model.

SIRCm flips the script. It treats immune escape not as an external perturbation, but as an endogenous process fueled by the epidemic itself. High case counts aren’t just a consequence of spread—they’re a cause of future spread.

This reframes our understanding of recurrent epidemics. Diseases like influenza, respiratory syncytial virus (RSV), and now COVID-19 exhibit repeated waves even in temperate zones with mild seasonal variation. Public health responses often assume these are driven by behavior or environment. But what if the engine is internal?

Consider a city emerging from a winter surge. Cases decline. Schools reopen. People relax precautions. Then, unexpectedly, another wave hits in spring. Was it behavior? Or did the high winter prevalence generate enough mutations to reset population immunity, priming the system for rebound?

The model suggests the latter is plausible—even likely—under realistic conditions.

Moreover, SIRCm bridges two classical paradigms. At one extreme—low mutation feedback—it reduces to SIRC, capturing gradual waning of immunity. At the other extreme—high feedback—it approaches SIS (Susceptible-Infected-Susceptible), where re-infection is immediate. Between them lies a dynamic regime where neither model suffices.

This has practical implications. If we assume immunity wanes slowly (SIRC-like), we might expect long intervals between waves and plan accordingly. If we treat re-infection as rapid (SIS-like), we might accept constant circulation. But if the truth is somewhere in between—and context-dependent—then static models will mislead.

Public health planning must account for state-dependent vulnerability. The risk of resurgence isn’t just a function of time since last exposure. It’s shaped by how many people were infected, and when. A massive wave today could shorten the window until the next one—not because immunity fades faster, but because the virus evolves faster.

This also affects surveillance priorities. Genomic monitoring isn’t just about identifying variants—it’s about measuring the rate of immune escape. If mutation scales with prevalence, then high case counts should trigger intensified sequencing, not just clinical response.

What's Next

The SIRCm model is a theoretical advance, but its assumptions need testing against real-world data. Does immune escape truly scale with $I$ or $SI$? Are there thresholds beyond which feedback becomes dominant?

Empirical validation will require linking epidemiological curves with genomic data. For instance, one could regress the rate of amino acid substitutions in spike proteins against contemporaneous case counts. Preliminary evidence supports this: during the Delta and Omicron surges, mutation rates spiked alongside prevalence (Zhang et al., 2022).

Another open question is whether this feedback operates symmetrically across age groups or immune backgrounds. Children may contribute disproportionately to transmission (high $SI$), while immunocompromised individuals harbor prolonged infections (high $I$). Each could drive distinct flavors of immune escape.

The model also assumes homogeneous mixing—a limitation shared by most compartmental frameworks. Future work could incorporate network structure or spatial heterogeneity to explore how localized outbreaks seed global immune escape.

Perhaps most importantly, SIRCm challenges the assumption that endemicity implies stability. Many hoped that SARS-CoV-2 would settle into a predictable, seasonal pattern. But if infection-driven feedback is strong, endemic may not mean boring. It could mean perpetual motion.

That doesn’t doom us to endless crisis. But it does suggest that control strategies must be adaptive. Vaccination campaigns timed to anticipated waves may fail if those waves are generated internally and unpredictably. Instead, we may need responsive surveillance systems that detect rising mutation pressure—not just rising cases.

One policy implication stands out: suppressing large outbreaks does more than save lives in the short term. It may delay or prevent future waves by reducing the mutation pool. In this view, mitigation isn’t just emergency response—it’s long-term prevention.

The SIRCm model reminds us that epidemics are not passive processes. They are adaptive systems, shaped by feedback loops that can amplify small changes into large consequences. We ignore them at our peril.

As the authors conclude: “Recurrent epidemic waves can emerge from within the system—no external driver required.”

That’s not just a mathematical curiosity. It’s a warning—and a call to rethink what we thought we knew about how diseases persist.