New pandemic simulator helps to predict the evolution of pandemics in different epidemiological scenarios
Researchers from the Universitat Politècnica de València (UPV), belonging to the ALFA group of the Valencian Institute of Artificial Intelligence (VRAIN), have participated in the development and validation of a new pandemic simulator that helps to predict its evolution taking into account different epidemiological scenarios. The system, called LOIMOS, has been developed in the context of COVID-19 and its results are limited to the SARS-CoV-2 virus. However, it could be applied to the study of any other pandemic, other than the one caused by this virus.
“The versatility of LOIMOS makes it a very useful tool in making decisions about non-pharmaceutical measures to limit virus transmission, both in this pandemic and in others that may come. We can draw multiple scenarios, pose as many questions and hypotheses as we want and predict their effects. This helps a lot to decide what measures to take, to establish those that are most effective to prevent or at least limit the spread of the virus,” says José M. Sempere, a researcher in the ALFA-VRAIN group at the Universitat Politècnica de València.
LOIMOS has been developed by researchers from the Universitat Politècnica de València, the Biology and Evolution of Microorganisms Group of the Instituto Ramón y Cajal de Investigación Sanitaria (IRYCIS) in Madrid, the CIBER in Epidemiology and Public Health, the FISABIO Foundation, the Universitat Politècnica de València, the CIBER in Epidemiology and Public Health, and the FISABIO Foundation, the Universitat de València (UV), the Consejo Superior de Investigaciones Científicas (CSIC), the Hospital General Universitario de Valencia, the Hospital Universitario la Paz de Madrid and the spin-off of the Universitat de València Biotechvana, located in the Parc Científic of the academic institution. Their results have been published in the international journal microLife.
Virtual and hierarchical design of the behavior and evolution of viruses
The system is based on membrane computing models, which allow virtual design of the behavior of viruses in various environments, conditions and levels of severity.
“These models reproduce viruses and their interactions with an unprecedented level of detail. In this way, we can evaluate and predict the incidence of a virus in a neighborhood, a city or a country, under different situations and observe its evolution in the short, medium and long term,” explains José M. Sempere.
“The important thing is that we simulate under certain assumptions, for example, different types of preventive measures, and evaluate the infection rate and its variation over time in the population. Obviously, if the assumptions change, the effects of the virus could change. That is why it is so relevant to have a tool like the one developed here: because it can help to propose effective measures against the spread of this epidemic by the virus and, by extension, by any other pathogen”, explains Andrés Moya, researcher at Fisabio, UV and CSIC.
LOIMOS is also a model with different hierarchical levels that interact with each other, unlike others used to date. Thus, by modifying one parameter of these levels, it allows us to see the effects not only on that level, but also on all the others. “For example, we could increase in the model the period in which the virus can produce contagions and see how this would affect the number of people who go to work,” explains Marcelino Campos, IRYCIS researcher and also from the ALFA-VRAIN Institute group at the Universitat Politècnica de València.
LOIMOS variables
Among other variables, LOIMOS incorporates the type of infections – asymptomatic or symptomatic; the degree of immunity acquired by passing the infection or the period and rates of infection – LOIMOS allows different values to be defined depending on the area and age range of the infected person or the mechanics of the infection.
“In the latter case, when simulating the infection, we can define its growth, how the immune system acts at first, when immunity can be generated and the probabilities of this happening – they can even be different according to age range – and the effects on the infected person – no symptoms, mild symptoms, severe symptoms, critical symptoms or death,” Marcelino Campos points out.
The setting, a European city of 10,000 inhabitants
To validate LOIMOS, the research team applied it to a typical European city of just over 10,000 inhabitants, reproducing the dynamics of the epidemic and the effects of immunity on SARS-CoV-2 transmission in different age groups.
The model predicted the consequences of delaying the adoption of non-pharmaceutical interventions 15 to 45 days after the first reported cases and the effect of these interventions on infection and mortality rates. The researchers also simulated non-pharmaceutical interventions to reduce infections at three different levels: 20%, 50% and 80%.
And another of the most relevant conclusions was the verification of the importance of focusing the first efforts on the most sensitive and oldest people. “If the most sensitive and older people are isolated as soon as the infections start, they are able to slow down a little, but where it is most noticeable is in the healthcare resources used and in mortality, as these people are the ones most likely to suffer the worst symptoms when infected,” adds Campos.
Currently, the LOIMOS team is working on incorporating – and simulating in the model – the incidence of new strains during the vaccination period.
What data do we work with?
One of the most outstanding features of models such as LOIMOS is its contribution to provide information that covers the impossibility of testing in real life. The LOIMOS team works for this with three types of data: firstly, with those that can be extracted from existing knowledge; the second group is data that can be directly measured; and the third is data that has to be inferred. “In models like LOIMOS there is a great effort to fit the latter data as much as possible. We test many values in different experiments, looking for results that are as close as possible to reality. Finding possible values for these parameters can be very valuable to biologists for their own studies,” concludes Marcelino Campos.
Click here for more information (spanish)