2023-12-11 16:00:00
CHRISTMAS SPECIAL! Get an exclusive GIFT with your subscription to National Geographic magazine.
Enjoy National Geographic Special Editions at an exclusive price for being a subscriber.
The arrival of COVID-19 changed societies as they were known on a global scale. The planet was forced to stop, economies faltered and citizens suffered the consequences of a virus, SARS-CoV-2, which left an indelible mark on the collective imagination. Science, once it overcame the first challenge by developing vaccines that mitigated the effects of this disease, was forced to continue researching.
The ravages caused by COVID-19 set off alarms and, now, the Foresight has taken on a leading role in laboratories and research centers around the world. In an increasingly interconnected planet in which viruses emerge and reach the other side of the Earth in a matter of days, detecting cases of each new pathogen early is crucial. Identify areas where these zoonoses could occur even more.
For example, this study published in the magazine Proceedings of the National Academy of Sciences presents a totally new approach: although previous studies had already investigated how environmental, phylogenetic and geographical variables determined the infection of numerous pathogens, none had until now developed a method applicable to a large number of systems in which that relationship between pathogen and host is created.
This is how the coronavirus infects human cells
Artificial intelligence applied to research
In the study, the researchers provide a artificial intelligence based approachspecifically through machine learning, which can integrate a large number of variables and be applied to any system of relationship between pathogen and host. That is why these results not only agree with the data of each system analyzed, but also provide this new tool that can help discover potential host species and new “hot spots” in planetary geography in which this interaction between pathogen and species could occur.
Although previous studies had already investigated how variables influenced the infection of numerous pathogens, none had developed a method applicable to a large number of systems.
Ángel Luis Robles Fernández, of the Veracruzana University of Mexico and one of the authors of the study, explains the methodology in an email interview with National Geographic España: “this method only uses a proportion of the information because we start from the assumption that the observed interactions have the same nature and are explained with the same variables, so the species with which there is more information are taken as a starting point.”
What is ChatGPT and how does it work?
How is the machine learning? Robles explains it like this: “We start from a statistical hypothesis and a generalized linear model, but by combining more variables and proposing a model that is not necessarily linear, the machine learning allows a more precise prediction if the phenomenon being modeled is more complex”, a solution that allows numerous variables to be added for prediction with truly relevant information.
By entering data on known incidents between pathogens and hosts, the algorithm used learns by itself to organize them to classify them in detail. “These procedures They are repeated thousands of times and therein lies the power of searching for patterns using these tools. of machine learningsince otherwise it would be a perhaps less robust process statistically speaking,” says Andrés Lira-Noriega, another of the authors of the study.
Predictions point to Eurasia
Together with Andrés Lira-Noriega, Researcher at the Advanced Molecular Studies Networkand Diego Santiago-Alarcón, professor at the University of South Floridathe authors applied the model to three systems of relationship between pathogens and hosts: coronaviruses and bats, as well as West Nile virus and malaria with birds. Their results suggest that the transmission of avian malaria depends largely on the distance between hosts, while the transmission of coronaviruses between bats is mainly affected by the geographical distribution between species.
For its part, the transmission of West Nile virus is largely influenced by a combination of environmental, geographical and phylogenetic factors. “We are working on modeling different host-parasite systems, but we are also studying whether we can model other biological interactions, not just host parasitebut in general predict biological interactions,” adds Robles.
From this information, the researchers identified numerous hot spots around the world, detailing how the region of Eurasia is particularly susceptible to these interactions with avian malaria pathogens (caused by Plasmodium left) and West Nile virus.
ChatGPT Artificial Intelligence Interview
A method to predict future pandemics
Given the current global epidemiological situation, which poses infectious disease challenges as a result of the deterioration of the natural order around the world, the need to anticipate new risks plays a major role. The best example of this has been COVID-19.
“This is just an example of other threats that we can expect and therefore anticipating these risks is necessary to be able to carry out preventive actions that help us avoid the havoc that they can cause. The predictions that we can make thanks to this type of analysis models and tools can help us help optimize resources (economic, human, institutional) as well as strategies so that governments at different levels act preventively,” details Andrés Lira-Noriega.
But this tool goes one step further, allowing the possibility of extrapolate it to other virus systems and hosts around the world, so that we can predict where future pandemics could occur. “Our hope is that this tool can be used to have spatially and temporally explicit predictions to explore reservoirs and vectors of multiple pathogens. A clear example would be its application for know which bat species to sample to continue working on the development of universal vaccines for COVID-19,” says Lira-Noriega.
#Artificial #intelligence #predict #future #pandemics