Modelling the spread of infectious diseases using stochastic processes

The spread of infectious diseases has long been a major concern in public health, and mathematical modelling plays a crucial role in understanding and controlling outbreaks. While deterministic models such as the classical SIR (Susceptible–Infected–Recovered) framework provide useful insights, they often assume uniformity and certainty. In reality, disease spread is influenced by random events such as chance encounters, variations in immunity, or changes in behavior. This is where stochastic processes become powerful tools, allowing researchers to capture the inherent randomness of epidemics.

---

Why Stochastic Modelling?

Deterministic models predict an average trajectory of an epidemic, but real-world epidemics do not follow a smooth path. Stochastic modelling accounts for:

• Random contact patterns between individuals.
• Uncertainty in transmission probabilities.
• Small population effects where chance events can determine whether a disease dies out or escalates.
• Variability in recovery and incubation periods.

By incorporating probability distributions rather than fixed values, stochastic processes provide more realistic simulations of disease spread.

---

Basics of Stochastic Processes in Epidemiology

A stochastic process is a collection of random variables representing the evolution of a system over time. In epidemiology, this translates to the number of susceptible, infected, and recovered individuals changing unpredictably.

Common stochastic tools in disease modelling include:

1. Markov Chains

   • Disease states (Susceptible, Infected, Recovered) are modeled as states in a Markov chain.
   • Transition probabilities define how individuals move between states over time.
   • Example: The probability of an infected person recovering in the next time step.

2. Poisson Processes

   • Used to model random events such as infection occurrences or contacts.
   • Assumes infections occur independently at a certain average rate.

3. Stochastic Differential Equations (SDEs)

   • Extend classical epidemic models by adding random noise terms.
   • Capture fluctuations around deterministic epidemic curves.

4. Branching Processes

   • Often used to model early stages of an epidemic when each infected person may or may not cause further infections.
   • Helps estimate the probability of disease extinction vs. epidemic growth.

---

The Stochastic SIR Model

The classical SIR model divides the population into three groups:

• S (Susceptible): Individuals who can contract the disease.
• I (Infected): Individuals who have the disease and can transmit it.
• R (Recovered/Removed): Individuals who are immune or no longer spreading the disease.

In a stochastic version:

• The number of new infections in a time step follows a Binomial distribution, depending on the number of susceptible and infected individuals.
• Recovery events occur randomly, often modeled with exponential waiting times.
• Outcomes differ across simulations—sometimes the disease dies out early, other times it spreads widely.

---

Applications of Stochastic Modelling

1. Predicting Outbreak Probabilities

   • Stochastic models estimate the likelihood that a single infection sparks a large outbreak.
   • This is especially important in emerging diseases where early extinction is possible.

2. Evaluating Control Strategies

   • Vaccination, quarantine, and social distancing can be tested under random scenarios.
   • For example, models can determine how many random contacts must be reduced to halt an outbreak.

3. Modeling Superspreading Events

   • Real outbreaks often involve individuals who infect disproportionately many others.
   • Stochastic approaches capture such events better than deterministic averages.

4. Small Population Settings

   • In schools, villages, or isolated communities, chance events matter more.
   • Stochastic processes predict variability in outbreak sizes across different runs.

5. Uncertainty Quantification

   • By running many simulations, stochastic models provide confidence intervals for infection counts, unlike single deterministic curves.

---

Case Example 

Imagine a population of 1,000 individuals with 1 infected person initially.

• Deterministic model: Predicts a smooth curve where infection peaks at 300 individuals before declining.
• Stochastic model: In repeated simulations, some outbreaks die out after 10 cases, while others infect more than 500 people.

This variability highlights the importance of stochastic approaches in public health planning.

---

Challenges in Stochastic Modelling

• Computational Intensity: Requires repeated simulations and complex algorithms.
• Parameter Estimation: Accurate probability distributions depend on high-quality data.
• Interpretation Complexity: Results often come as ranges and probabilities, not precise forecasts.

Despite these challenges, the realism offered by stochastic modelling makes it indispensable.

---

Conclusion

Stochastic processes enrich our understanding of infectious disease dynamics by capturing the randomness, uncertainty, and variability inherent in epidemics. They complement deterministic models, providing a fuller picture of outbreak risks, intervention effectiveness, and potential outcomes.

As global health systems face challenges from pandemics and emerging diseases, stochastic modelling will continue to be a vital tool in shaping evidence-based strategies and ensuring preparedness.

Key takeaway: In the fight against infectious diseases, embracing randomness through stochastic modelling leads to better predictions, smarter policies, and ultimately, stronger public health outcomes.