Preventing pandemics with a simulator

Researchers at the Technical University of Valencia, (UPV), part of the ALFA group of Valencia’s Artificial Intelligence Insititute (VRAIN), have developed and validated a pandemic simulator. This will help prevent the development of pandemics by taking different scenarios into account. The system has been dubbed LOIMOS. It was developed during the COVID-19 emergency and as such, its results are linked to SARS-CoV-2. However, it could be applied to any other pandemic, writes the Technical University of Valencia in a press release.

“LOIMOS’ versatility makes it a useful tool for making decisions about non-pharmaceutical measures aimed at containing the spread of a virus. We can outline several scenarios, we can pose all the questions and hypotheses we want, and we are able to predict the repercussions. This helps a lot in deciding which measures to put into place. And to establish which ones are most effective to avoid or to contain the spread of a virus,” stresses José M. Sempere, researcher of the ALFA-VRAIN group from the Technical University of Valencia in Spain.

LOIMOS was developed by researchers from the Technical University of Valencia, Biology and Microorganisms Evolution Group of the Ramón y Cajal Insitute of Sanitary Investigation (IRYCIS) of Madrid, the CIBER for Epidemiology and Public Health, FISABIO foundation, the University of Valencia, the Superior Council for Scientific Investigations (CSIC), the General University Hospital of Valencia, University Hospital La Paz in Madrid and Biotechvana, a spin-off from the University of Valencia, based in the Scientific Park of the academic institution. Its results have been published in the international journal microLife.

Virtual and hierarchical design of the virus’s behavior

The system is based on computing models which is capable of designing the virus’ behavior virtually in different environments, conditions and severity levels.

“These models reproduce viruses and their interactions in an unprecedented way. We can assess and predict the virus’ impact on a neighborhood, on a city or on a village under different conditions. Plus we can observe its evolution on a short, medium and long span of time”, Sempere explains.

“What matters is the ability to simulate different hypotheses. For instance, a range of containment measures. Depending on these, the infection rate of the population can be assessed as time passes. If the scenario changes, the effects of the virus will change too. That’s why it’s important to have a tool like LOIMOS. It can help with providing effective measures against the spread of an epidemic or of any other pathogen,” says Andrés Moya, researcher at Fisabio, UV and CSIC.

What’s different from earlier tools, is that LOIMOS comprises hierarchical levels that interact with each other. By modifying a setting of one of the levels, the effects on all the others are shown this way. “For instance, we could set the model so that there is a period with a higher number of infections to see how this would affect the number of people going to work”, explains Marcelino Campos, researcher at IRYCIS and also at the ALFA-Instituto VRAIN group of the Technical University of Valencia.

LOIMOS variables

LOIMOS embeds in the variables, amongst other things, the type of the infection – symptomatic or asymptomatic, and the immunity level acquired after contracting the virus during specific periods and the contagion rate. LOIMOS enables different values to be defined depending on the area and the age of the infected person or the way he/she was infected.

“In this case, when simulating an infection, we can define its spread, how the immune system reacts in the first stage, when immunity levels can be reached and the chances that this will happen. These can be different depending on the age group. As well as the effects on the infected person – no symptoms, severe symptoms, critical conditions or death,” Marcelino Campos points out.

A simulation on a European town of 10,000 people

To validate LOIMOS, the research team applied the model in a fictional standard European town of around 10,000 inhabitants. They reproduced the epidemic’s dynamics and immunity effects of the SARS-CoV-2’s virus within different age groups.

The model predicted the consequences of delaying non-pharmaceutical measures by 15 – 45 days after the first confirmed cases and the effects of such measures on the mortality and infection rates. The researchers also simulated non-pharmaceutical measures with the aim of reducing contagions on three different levels: by 20%, 50% and 80%.

Another finding is the confirmation that the first measures need to be made to protect the elderly and individuals whose health is compromised. “If a contagion starts to spread, and the more vulnerable individuals and the elderly are isolated, infections can be slightly contained. However, these kinds of measures are key when it comes to reducing hospitalizations and mortality rates, as these groups are most prone to becoming severely infected,” Campos notes.

At the moment, the group is working on incorporating – and simulating – the impact of new variables during vaccination phases.

What data is used

One of the main features models like LOIMOS offer, is the contribution it makes in bringing information that can compensate for the lack of proof in real life.The LOIMOS team works with three types of data: information that can be derived from data that is already known, information that is directly measurable, and information that needs to be distilled. “Models like LOIMOS work on refining this third group of data. We tried different values in different tests which obtained results comparable to real life practice. Finding values to these parameters can help biologists as well,” Campos adds.

Also interesting: Study suggests R value for tracking pandemic should be dropped in favour of ‘nowcasts’