A scientific twist on an iconic video game just might prevent the apocalypse
With a little less dancing and a little more traffic modeling.
Whether it's perfecting your Nintendo Mii or perfectly replicating your childhood cul-de-sac in The Sims, we're suckers for recreating our lives in a virtual world. And this is an obsession that might just help avert the world's biggest existential threat.
The spread of Covid-19 in 2020 has already been a stark wake-up call to the kinds of damage a global virus can wreck on our cities if we're unprepared — and scientists predict that these kinds of pandemics are only going to become more common.
To meet this oncoming apocalypse head-on, a team of scientists have designed a much more hardcore version of the iconic video game The Sims that could stop future pandemics in their tracks.
Using supercomputers and analytic techniques borrowed from high-level chemistry, a team of engineers and epidemiologists have designed a new approach to modeling the spread of infectious viruses, like Covid-19, that could accurately predict how new risks — like super spreader events — or precautions — like mask mandates — affect the ebb and flow of disease spread.
In their work, the team was able to accurately model how lockdowns flattened the infection curves in Birmingham, England, and Bogotá, Colombia using only "hard" data like population statistics and estimated movement maps.
Why it matters — Immediately, this powerful estimation method like this could help epidemiologists better understand and target new Covid-19 restrictions to effectively slow the virus's spread, but the use of this model may be much more widespread. With enough time and computing power, the researchers write it could "simulate the mobility of the entire human population."
The findings were published Tuesday in the journal Proceedings of the Royal Society A: Mathematical and Physical Sciences.
Here's the background — It's romantic to imagine our lives as individualistic narratives driven by our own free will and dreams, but in reality we're actually extremely predictable, write the researchers in their paper.
Just as it is commonplace in modern chemistry and particle physics to model the complex movement of atoms using supercomputers and highly advanced stochastic (or, probabilistically) algorithms, the researchers say that the movement of people through a city can be similarly modeled.
"The epidemiological models used to predict the spread of infectious diseases are similar to the mathematical models used in chemistry. It is not a coincidence," write the authors. "The complex dynamics of large numbers of molecules or individuals is condensed into a few ordinary differential equations."
Borrowing from techniques used in the physical sciences, the researchers present in their paper a new approach to model epidemiology (the study of disease spread and distribution) using digitally modeled cities. They dubbed the approach discrete epidemiology as a nod to its mathematical roots.
When modeling Bogotá the researchers broke up the city's map into three main sections for living and working.
Digging into the details — To put their model to the test, the team set out to create exact replicas of two cities: Birmingham (which has a population of about 2.6 million) and Bogotá (which has a population of nearly 11 million.)
To make their models, they looked at cities' average household size, layout, infrastructure (such as public transportation,) and age distribution. They then randomly assigned "particles" in the simulation, which represented residents, an employment status to determine their day-to-day activity (the options included student, employed, and unemployed), and used average distances between these locations to estimate their daily commute activity.
These daily commutes were categorized as "predictable paths" and so-called random walks (a computer science term that is essentially exactly what it describes) to simulate unpredictable trips like shopping or picking up a latte.
With all this information baked in, they then let the city simulations run for several months in simulation-time or a few hours in real-time to simulate both pre-lockdown and lockdown dynamics and virus spread.
What they discovered — While their model was not completely accurate in simulating the behavior of these virtual residents, they were able to very closely match the cities' true pre- and post-lockdown infection numbers by applying only these few parameters to their model.
Using estimations based on the city's size and predicted movements, the team's model was able to accurately match infection spikes and declines in Birmingham.
The authors write that when modeling Brazil's early opening they did notice that the model tended to over-estimate the amount of new infections that would crop up versus true data from the openings.
"A possible explanation is that, after lockdown is lifted, people tend to be more careful about social distancing," explain the authors.
What's next — With this proof-of-concept under their belt, the researchers are excited for the possible improvements their model will undergo in the future. For example, future models could use the city's real topology to exactly model where homes and offices are (instead of randomly generating) and can include more detailed information about the residents' habits and jobs.
They also write that — with the right amount of resources — it might be possible to model the entire world's movement just as accurately. This could be huge for scientists working to model future infection outbreaks or pandemics by helping them test and predict new treatments or parameters even faster.
These models might not be as addicting to play with as The Sims, but this virtual worlds will play a crucial role in saving the real world.
Abstract: This study develops a modelling framework for simulating the spread of infectious diseases within real cities. Digital copies of Birmingham (UK) and Bogotá (Colombia) are generated, reproducing their urban environment, infrastructure and population. The digital inhabitants have the same statistical features of the real population. Their motion is a combination of predictable trips (commute to work, school, etc.) and random walks (shopping, leisure, etc.). Millions of individuals, their encounters and the spread of the disease are simulated by means of High-Performance Computing and Massively Parallel Algorithms for several months and a time resolution of 1 minute. Simulations accurately reproduce the Covid-19 data for Birmingham and Bogotá both before and during the lockdown. The model has only one adjustable parameter calculable in early stages of the pandemic. Policymakers can use our digital cities as virtual labs for testing, predicting and comparing the effects of policies aimed at containing epidemics.
