📊 statistics

Generalized SEIR model and a new reproduction number reflecting the real epidemic dynamics

This paper proposes a generalized six-equation SEIR model incorporating re-infections, newborns, vaccinations, and an exposed compartment to address previous limitations, introduces a new reproduction number based on exposed population dynamics for epidemic control, and utilizes this framework to predict a resurgence of pertussis cases in England during mid-to-late 2027.

Original authors: Igor Nesterruk

Published 2026-06-25

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Igor Nesterruk

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: A Better Map for Invisible Epidemics

Imagine trying to navigate a city where half the cars are invisible. You can only see the red cars (people who are sick and showing symptoms), but you know there are also blue cars (people who are infected but feel fine) and purple cars (people who have the virus but haven't started showing symptoms yet).

Most old maps (mathematical models) only counted the red cars. This paper introduces a new, more detailed map called a Generalized SEIR model. It tracks not just the sick people, but also the "hidden" drivers and the people currently in the "waiting room" of the infection.

The author, Igor Nesteruk, uses this new map to study a specific outbreak: Whooping Cough (Pertussis) in England.

The New Map: Six Moving Parts

The old models were like a simple game of tag with three roles: Susceptible (untouched), Infected (tagged), and Removed (out of the game).

This new model adds three more roles to make it realistic:

Exposed (The Waiting Room): People who have caught the virus but are in the "incubation period." They aren't contagious yet; they are just waiting for the virus to wake up.
Hidden Infected: People who are sick but don't show symptoms (or aren't getting tested). They are invisible to the official count but can still spread the virus.
The "Re-Entry" and "Newcomer" Factors: The model accounts for people getting sick again (re-infection) and new babies entering the population, which changes the total number of people available to catch the disease.

The "Visibility" Problem

The paper points out a major flaw in how we usually count epidemics. We usually only count the people who go to the doctor.

The Analogy: Imagine a stadium where only the people wearing bright red hats are counted. But inside, thousands of people are wearing blue hats and spreading the flu. If you only count the red hats, you think the game is over when it's actually just getting started.

The author introduces a "Visibility Coefficient." This is a mathematical tool that estimates how many "blue hat" (hidden) people exist for every "red hat" (visible) person we see.

The New "Speedometer": A Fresh Reproduction Number

Epidemiologists usually use a number called $R$ to see if an epidemic is growing or shrinking. The author argues that the old $R$ is like looking at the rearview mirror—it tells you what happened to the people who are already sick.

The paper proposes a new reproduction number that acts like a speedometer for the "Waiting Room" (Exposed people).

The Analogy: Instead of counting how many cars have crashed (sick people), this new number counts how many cars are currently speeding toward the crash site (exposed people).
Why it matters: If you see the number of cars speeding toward the crash site increasing, you know the crash wave is coming before the cars actually hit. This allows for earlier warnings.

The Case Study: Whooping Cough in England

The author tested this new map on the Whooping Cough outbreak in England from 2023 to 2025.

The Scenario: The model looked at the data and realized that vaccination rates weren't high enough to stop the "newcomers" (babies) from entering the pool of susceptible people.
The Prediction: The model ran a simulation into the future. It predicted that after a dip in cases, the number of new Whooping Cough infections in England would start to climb again.
The Result: The paper predicts this increase will begin between July and September 2027.

What About Vaccinations?

The paper runs a "what if" scenario.

The Analogy: Imagine vaccination is like a giant vacuum cleaner sucking people out of the "Susceptible" line.
The Finding: If the vacuum cleaner is strong enough (high vaccination rates), the epidemic can be cut short. However, the author notes that for the recent Whooping Cough outbreak in England, the "vacuum" isn't strong enough to stop the wave because the virus is hitting older people who aren't being vaccinated as effectively.

The Bottom Line

This paper doesn't offer a new medicine or a clinical treatment. Instead, it offers a better calculator.

It admits that our current counts of sick people are incomplete because we miss the "hidden" and "waiting" cases.
It provides a new way to calculate the speed of an outbreak based on the "waiting room" of exposed people.
It uses this new calculator to predict that England will see a rise in Whooping Cough cases in the summer of 2027.

The author concludes that to make these complex calculations work perfectly, we will need supercomputers and artificial intelligence to crunch the numbers, as the model has 14 different moving parts that need to be tuned just right.

Technical Summary: Generalized SEIR Model and a New Reproduction Number Reflecting Real Epidemic Dynamics

Problem Statement
Existing mathematical models for epidemic dynamics, such as the classic SIR model, often fail to account for critical factors including the incubation period (the exposed compartment), hidden (asymptomatic) infections, re-infections, and the effects of vaccinations and newborns. Previous models proposed by the author addressed hidden dynamics and re-infections but neglected the exposed compartment ( $E$ ). Furthermore, standard definitions of the effective reproduction number ( $R_t$ ) rely solely on visible (symptomatic) cases, which may not accurately reflect the true rate of infection spread when a significant portion of the population is asymptomatic or unregistered. The paper addresses the need for a more comprehensive model that integrates the exposed compartment with hidden dynamics and proposes a new metric to better reflect real epidemic trends.

Methodology
The author proposes a Generalized SEIR (Susceptible-Exposed-Infectious-Removed) model consisting of six coupled differential equations. This model expands upon previous work by introducing a compartment for exposed persons ( $E$ ) and dividing both infectious ( $I$ ) and removed ( $R$ ) populations into visible (registered) and hidden (unregistered/asymptomatic) sub-compartments.

Key features of the mathematical framework include:

Six Differential Equations: Governing the dynamics of Susceptible ( $S$ ), Exposed ( $E$ ), Visible Infectious ( $I^v$ ), Hidden Infectious ( $I^h$ ), Visible Removed ( $R^v$ ), and Hidden Removed ( $R^h$ ).
Parameters: The model incorporates 14 unknown parameters per epidemic wave, including infection rates ( $\alpha$ ), removal rates ( $\rho$ ), waning immunity rates ( $\delta$ ), the visibility coefficient ( $\beta$ ), and a coefficient ( $\mu$ ) representing the net rate of change in the susceptible population due to births, vaccinations, and re-infections.
Parameter Identification: The author proposes a method to identify these 14 parameters using the method of least squares, fitting the model to observed daily, weekly, or monthly numbers of registered cases ( $D_j$ ). The model relates the derivative of visible removed cases to the observed data.
New Reproduction Number ( $R_\tau^E$ ): A new reproduction number is defined based on the ratio of exposed persons at time $t$ versus $t-\tau$ . Crucially, the paper demonstrates that this rate can be estimated using only the ratio of daily (or weekly) registered cases, effectively linking visible data to the dynamics of the hidden exposed population.

Key Contributions

Generalized Model Structure: The introduction of a six-equation SEIR model that simultaneously accounts for the incubation period, hidden infections, re-infections, and demographic changes (births/vaccinations).
Parameter Estimation Method: A specific procedure for estimating the model's 14 parameters using standard epidemiological data (accumulated or daily case counts) without requiring direct observation of hidden cases.
Revised Reproduction Number: The proposal of a reproduction number ( $R_\tau^E$ ) that reflects the increasing or decreasing rate of exposed persons. The paper establishes that the ratio of consecutive daily (or weekly) registered case counts serves as a valid estimator for this rate, providing a practical tool for epidemic control decisions.
Equilibrium Analysis: The paper provides an analysis of equilibrium points, noting that a non-zero equilibrium (endemic state) is only possible when the net susceptible growth rate ( $\mu$ ) is non-zero.

Results
The model was applied to simulate the pertussis (whooping cough) epidemic in England from 2023 to 2025.

Simulation Scenarios:
- $\mu = 0$ : Simulations assuming no net growth in the susceptible population (e.g., perfect vaccination balancing births) resulted in a monotonic decrease in cases after a peak, suggesting a limited epidemic duration.
- $\mu > 0$ : Simulations incorporating a positive net growth rate (reflecting birth rates exceeding vaccination efficacy) produced oscillating epidemic waves with a period of approximately 2500 days (6.85 years).
Validation: The model with $\mu = 7.48 \text{ day}^{-1}$ showed strong agreement with historical data and recent observations (2025) of pertussis cases in England.
Prediction: Based on the calibrated model, the paper predicts that the number of new pertussis cases in England will begin to increase again in July–September 2027. The model indicates that the number of exposed persons will begin to rise immediately following the minimum in daily registered cases.

Significance and Claims
The paper claims that the proposed Generalized SEIR model offers a more realistic representation of epidemic dynamics by integrating the incubation period and hidden transmission. The primary significance lies in the ability to:

Predict future epidemic waves (specifically the 2027 pertussis resurgence in England) with high accuracy using only registered case data.
Provide a practical, data-driven definition of the reproduction number that accounts for hidden dynamics, allowing epidemiologists to monitor the true rate of infection spread using standard surveillance data.
Demonstrate that while intensive vaccination can theoretically reduce epidemic duration, it may be insufficient to prevent long-term oscillations if birth rates and waning immunity are not adequately compensated.

The author notes that the large number of parameters (14) complicates optimization, suggesting that future research should utilize High Performance Computing (HPC) and AI methods for large-scale parameter identification. The study concludes that the model successfully bridges the gap between theoretical compartmental dynamics and observable real-world data.