Institute of High Performance Computing


Powerding Discoveries!



Computing Science (CS)


Modeling Pandemic Influenza in Singapore

The 1918 Spanish flu pandemic is estimated to have killed 50-100 million people worldwide or 2.5-5% of the human population at the time. Each century has seen an average of three pandemics and the World Health Organization (WHO) anticipates an influenza pandemic within the decade. To prepare for a pandemic, we are working with the National University of Singapore (NUS), Nanyang Technological University (NTU) and the Ministry of Health (MoH) to model and study the spread dynamics of pandemic influenza in Singapore.

Traditional epidemic modeling methods have been centered on deterministic compartmental models in which the human population is partitioned into three “compartments”: Susceptibles who able to acquire the disease, the Infected who able to transmit the disease and individuals who have Recovered and are immune. This Susceptible-Infectious-Recovered (SIR) model has been extensively used to model diseases such as measles, mumps and influenza. However, the basic SIR methodology and its variants (S-Exposed-IR, SIR-Susceptible etc.) assume a homogenous, fully-mixing population. In other words, the SIR model assumes that (1) people are identical and (2) any one person can infect any other person in the population. Clearly, these assumptions are unrealistic.

To account for the differences between individuals and contacts inherent in the Singaporean population, we use recently developed contact network models. In these models, the population is represented as a graph or network where each node represents a person and contacts are represented as lines or edges between nodes (Figure 1). The number of contacts that a person has is called his or her degree.

Over the past year, we have been carefully constructing a statistically consistent population network for Singapore using data collated from the Census, surveys and agency databases. Using this network, we can quantitatively estimate the expected outbreak or epidemic size of a disease spreading in the community. For example, Figure 2 illustrates preliminary calculations of the size of an epidemic in Singapore compared to that of Portland, Oregon in the United States. As expected, the higher the transmissibility of a disease, the larger the expected size of the epidemic. We can also see that Portland is more susceptible to a larger epidemics compared to Singapore

From a scientific perspective, epidemic models are able to give insights into the nature of disease spread. In particular, our models will illustrate the dynamics of epidemics in a high-density urbanized environment such as Singapore. In addition to inherent scientific value, accurate epidemic models are of tremendous use to the public health authorities for rational decision-making. Epidemic models enable us to test and evaluate control strategies that will prove invaluable to minimize loss and suffering in the event of a global pandemic.


Back to Research Highlight

This page is last updated at: 26-MAY-2009