Role of relapse and multiple time delays in shaping Nipah virus epidemic dynamics: a mathematical modeling study

This study develops a delay differential equation model incorporating relapse and multiple time delays to demonstrate that while incubation delays affect epidemic peak timing, relapse is the critical mechanism driving Nipah virus persistence and altering endemic equilibrium conditions beyond the classical reproduction threshold.

The Gist

Imagine the Nipah virus (NiV) not just as a single attack, but as a ghost that refuses to leave the house.

This paper is a mathematical investigation into why the Nipah virus keeps popping up in Bangladesh, even when it seems like the outbreak should be over. The researchers built a digital "crystal ball" (a computer model) to simulate how the virus moves through people, but they added two special ingredients that most other models ignore: Relapse (the ghost coming back) and Time Delays (the lag between infection and symptoms).

Here is the breakdown of their findings using simple analogies:

1. The "Ghost in the Machine" (Relapse)

Usually, when you get sick, get better, and recover, you are done. You are immune, or at least you aren't contagious anymore.

But with Nipah, the story is different. Some people who seem fully recovered suddenly get sick again months or even years later.

• The Analogy: Imagine a fire that you think you've put out. You see no smoke, so you leave. But then, a hidden ember glows back to life, starting a new fire.
• The Finding: The researchers found that this "relapse" is the main reason the virus sticks around. Even if you stop all new infections (no new people catching it from others), the virus can keep spreading because recovered people are secretly becoming contagious again. It's like the virus has a "recharge button" hidden in the recovered population.

2. The "Slow Motion" Effect (Time Delays)

Viruses don't work instantly. There is a gap between when you catch it and when you get sick (incubation), and a gap between when you recover and when you might get sick again (delayed encephalitis).

• The Analogy: Think of a row of dominoes. If you push the first one, the last one doesn't fall immediately; there's a chain reaction time. In this virus, the "push" happens, but the "fall" (symptoms) happens days or months later.
• The Finding: These delays change when the peak of the outbreak happens.
  • Incubation Delay: This acts like a speed bump. If the virus takes longer to show symptoms, the big wave of sickness hits later and might be slightly less intense at the very start, but it shifts the whole timeline.
  • Recovery Delay: This is the "ghost" delay. It determines when the recovered people start causing trouble again.

3. The Two Phases of the Outbreak

The model revealed that the outbreak has two distinct acts, like a play:

• Act 1: The Initial Crash (The First Peak)
  • Who drives this? The incubation delay and how fast the virus spreads from person to person.
  • What helps? If we can detect cases faster (shortening the incubation delay in the model), we can stop the first big wave from getting too high. It's like catching the dominoes before they all fall.
• Act 2: The Long Tail (The Aftermath)
  • Who drives this? Relapse.
  • What happens? After the first big wave dies down, the virus doesn't disappear. It lingers because the "ghosts" (relapsing patients) keep re-entering the mix.
  • The Surprise: The researchers found that after the first peak, over 80-90% of new infections might actually come from people who had already recovered and got sick again, not from new transmissions. This is why the virus is so hard to kill off completely.

4. The "Magic Number" (R0) vs. Reality

In epidemiology, there is a famous number called R0. If R0 is below 1, the disease should die out because each sick person infects less than one other person.

• The Twist: In this study, the calculated R0 for Nipah in Bangladesh is very low (0.04). Mathematically, the virus should have died out years ago.
• The Reality: It didn't. Why? Because the standard R0 formula doesn't count the "ghosts." It assumes once you recover, you're out of the game. But because of relapse, the virus finds a way to keep the game going even when the math says it should stop.

5. Will the Virus Start Oscillating? (Hopf Bifurcation)

The researchers asked: "Could these delays cause the virus to start a rhythmic cycle of outbreaks, like a heartbeat?"

• The Theory: Yes, mathematically, if the delays get long enough, the virus could start flaring up and dying down in a perfect, repeating loop.
• The Reality: Based on the actual data from Bangladesh, the delays aren't long enough yet to cause these wild, rhythmic cycles. The virus is currently behaving more like a sporadic spark than a rhythmic drumbeat. However, the model shows that if conditions change, those cycles could happen.

The Big Takeaway for Public Health

If you want to stop Nipah, you can't just focus on stopping the initial spread (like washing hands or avoiding bat sap). You have to change your strategy for the long haul:

1. Early Detection is King: Shortening the time between infection and treatment stops the first big wave.
2. Watch the Recovered: You cannot just discharge a patient and say "they are safe." Because of relapse, recovered patients need long-term monitoring. They are the hidden reservoir keeping the fire alive.

In summary: The Nipah virus is a tricky opponent. It uses time delays to hide its timing and uses relapse to hide its source. To beat it, we need to look past the initial outbreak and keep a close eye on the people who have already survived it.

Technical Summary

1. Problem Statement

Nipah virus (NiV) is a highly lethal zoonotic pathogen with case fatality rates between 40% and 75%. While previous mathematical models have addressed NiV transmission, they often overlook critical clinical features documented in human cases:

• Relapse: Recovered individuals can develop symptoms again months or even years later.
• Delayed-onset Encephalitis: Severe neurological symptoms can appear long after apparent recovery.
• Incubation Delays: There is a significant lag between exposure and the onset of infectiousness.

Existing models fail to explicitly integrate these relapse mechanisms and multiple biological time delays into a cohesive framework. Consequently, the contribution of relapse to epidemic persistence and the potential for delay-induced oscillatory dynamics (recurring outbreaks) remain poorly understood. This study aims to fill this gap by quantifying how these factors shape NiV transmission dynamics, particularly in the post-peak phase.

2. Methodology

Model Formulation

The authors developed a Delay Differential Equation (DDE) model extending a previous compartmental framework (Tyagi et al., 2024). The human population ($N$) is divided into seven compartments:

• $S$: Susceptible
• $E$: Exposed (incubating)
• $I_1$: Asymptomatic infectious
• $I_2$: Symptomatic infectious
• $H$: Hospitalized
• $T$: Under treatment
• $R$: Recovered

Key Features of the Model:

• Incubation Delay ($\omega_1$): Individuals move from $E$ to $I_1$ after a delay, with a survival probability term $e^{-\mu\omega_1}$.
• Relapse Pathways:
  • Direct relapse: Recovered individuals ($R$) return to the asymptomatic class ($I_1$) at rate $\psi_1$.
  • Delayed relapse (Encephalitis): Recovered individuals return to the symptomatic class ($I_2$) after a delay $\omega_2$ at rate $\psi_2$.
• Mathematical Analysis:
  • Computed the Basic Reproduction Number ($R_0$) using the next-generation matrix method.
  • Analyzed the Disease-Free Equilibrium (DFE) and Endemic Equilibrium (EE) stability.
  • Investigated Hopf bifurcations to determine if time delays could induce sustained periodic oscillations (epidemic cycles).
  • Derived conditions for the existence of a positive endemic equilibrium, showing it is not solely determined by $R_0 > 1$ when relapse is present.

Parameter Estimation and Simulation

• Data Source: Annual NiV incidence data from Bangladesh (2001–2024).
• Calibration: Parameters (transmission rate $\alpha$, relapse rates $\psi_1, \psi_2$, delays $\omega_1, \omega_2$, etc.) were estimated using MATLAB's lsqcurvefit (least-squares minimization).
• Sensitivity Analysis:
  • Local Sensitivity: Varied key parameters ($\pm 10\%, \pm 20\%, \pm 30\%$) to assess impacts on peak timing and magnitude.
  • Global Sensitivity: Used Partial Rank Correlation Coefficients (PRCC) to identify parameters most influential on $R_0$.
  • Scenario Analysis: Examined joint effects of transmission, treatment, and relapse rates.
• Bifurcation Analysis: Numerical simulations were conducted under hypothetical parameter regimes to observe Hopf bifurcations and oscillatory dynamics.

3. Key Contributions

1. Novel Modeling Framework: First study to explicitly integrate relapse rates and multiple time delays (incubation and post-recovery encephalitis) into a single DDE model for NiV.
2. Redefinition of Persistence Thresholds: Demonstrated that the existence of an endemic equilibrium is not solely determined by the classical threshold $R_0 = 1$. Relapse feedback creates a mechanism where the disease can persist even if primary transmission is sub-critical ($R_0 < 1$).
3. Quantification of Relapse Contribution: Introduced a "relapse-driven replenishment fraction" to quantify the proportion of new infections arising from relapse versus primary transmission.
4. Bifurcation Analysis: Provided theoretical and numerical evidence that biological time delays can destabilize the endemic equilibrium, leading to Hopf bifurcations and sustained oscillations, though these require specific (hypothetical) parameter regimes.

4. Key Results

Epidemiological Dynamics

• Role of Relapse: Relapse is the dominant driver of post-peak dynamics. While primary transmission drives the initial outbreak peak, relapse sustains low-level transmission and creates "secondary humps" or extended tails in the epidemic curve.
• Relapse Contribution Fraction: In the post-peak phase (after 2007 in the Bangladesh data), the relapse contribution fraction exceeds 80–90%. This indicates that most new infections in the later stages of an outbreak stem from recovered individuals relapsing, not new human-to-human transmission chains.
• Impact of Delays:
  • Incubation Delay ($\omega_1$): Primarily affects the timing and intensity of the initial peak. Reducing $\omega_1$ (faster detection) significantly lowers cumulative cases.
  • Encephalitis Delay ($\omega_2$): Primarily affects the timing of the secondary phase. Increasing $\omega_2$ delays the return of infectious individuals, shifting the secondary peak later but reducing immediate post-peak burden.

Stability and Oscillations

• Stability: Under empirically estimated parameters (Bangladesh data), the system is stable with damped fluctuations; sustained oscillations are unlikely under current conditions.
• Hopf Bifurcation: Theoretical analysis shows that if delays ($\omega_1$ or $\omega_2$) exceed critical thresholds, the endemic equilibrium loses stability via Hopf bifurcation, leading to sustained periodic oscillations.
• Hypothetical Scenarios: Numerical simulations with hypothetical parameters confirmed that prolonged delays can generate recurring epidemic waves, highlighting the model's theoretical capacity for complex dynamics.

Sensitivity and Control

• $R_0$ Estimation: The estimated $R_0$ for Bangladesh is $\approx 0.047$ (well below 1), suggesting primary transmission alone cannot sustain an outbreak.
• Control Implications:
  • Reducing Relapse ($\psi_1$): A 30% reduction in relapse rate decreases total cumulative cases by ~14%. This is crucial for long-term elimination.
  • Reducing Incubation Delay ($\omega_1$): A 30% reduction in incubation delay (via faster detection) decreases total cases by ~23%, making it the most effective lever for reducing the initial peak.
  • Combined Effects: Joint reductions in transmission and improvements in treatment/hospitalization significantly suppress both initial and secondary peaks.

5. Significance and Conclusion

This study fundamentally shifts the understanding of NiV control by highlighting that relapse is a key mechanism for epidemic persistence, independent of primary transmission intensity.

• Public Health Strategy: Traditional control measures focusing solely on reducing contact rates (to lower $R_0$) may be insufficient for eradication if relapse is ignored. Strategies must include long-term monitoring of recovered patients to detect and isolate relapse cases.
• Modeling Implications: The study underscores the necessity of incorporating biological time delays and relapse pathways in modeling emerging zoonotic diseases (like Ebola or Lassa fever) where post-recovery transmission is documented.
• Future Outlook: While delay-induced oscillations are not currently observed in Bangladesh, the model warns that under changing biological or environmental conditions, these delays could trigger recurring epidemic cycles. Accurate estimation of relapse rates and timing is critical for reliable long-term forecasting.

In summary, the paper provides a rigorous mathematical basis for integrating relapse and time delays into epidemic models, offering actionable insights for managing the complex, long-tail dynamics of Nipah virus outbreaks.