Using machine learning, researchers can predict outbreaks of pathogens such as coronavirus and monkeypox
The rate at which emerging wildlife diseases infect humans has steadily increased over the past three decades. Viruses, such as the global coronavirus pandemic and the recent outbreak of monkeypox, have heightened the urgent need for disease ecology tools to predict when and where outbreaks are likely.
An assistant professor at the University of South Florida has helped develop a methodology that will do just that – predict disease transmission from wildlife to humans, from one wild species to another, and determine who is at risk of infection.
The methodology is a machine learning approach that identifies the influence of variables, such as location and climate, on known pathogens. Using only small amounts of information, the system is able to identify community hotspots at risk of infection globally and locally.
“Our main goal is to develop this tool for preventive measures,” said co-principal investigator Diego Santiago-Alarcon, assistant professor of integrative biology at USF. “It is difficult to have a general-purpose methodology that can be used to predict infections in all the various parasitic systems, but with this research we are helping to achieve this goal. »
With the help of researchers from the Universiad Veracruzana and the Instituto de Ecologia, located in Mexico, Santiago-Alarcon examined three host-pathogen systems – avian malaria, birds infected with West Nile virus and bats. mice with the coronavirus – to test the reliability and accuracy of the models generated by the methodology.
The team found that for all three systems, the most frequently infected species was not necessarily the most susceptible to the disease. To better identify hosts at higher risk of infection, it was important to identify relevant factors, such as climate and evolutionary relationships.
By integrating geographic, environmental, and evolutionary developmental variables, researchers identified host species that had not previously been recorded as infected with the parasite under study, providing a means to identify susceptible species and possibly to mitigate pathogen risk.
“We are confident that the methodology is a success and can be widely applied to many host-pathogen systems,” Santiago-Alarcon said. “We are now entering a phase of improvement and refinement. »
The results, published in the Proceedings of the National Academy of Sciencesprove that the methodology is capable of providing reliable global predictions for the host-pathogen systems studied, even using a small amount of information. This new approach will help guide infectious disease surveillance and field efforts, providing a cost-effective strategy to better determine where to invest limited disease resources.
Predicting what type of pathogen will produce the next medical or veterinary infection is difficult, but necessary. As the rate of human impact on natural environments increases, the possibilities for new diseases will continue to increase.
“Humanity, and indeed biodiversity in general, are facing increasing challenges from infectious diseases due to our incursion and destruction of the natural order in the world through things like deforestation. , global trade and climate change,” said researcher Andrés Lira-Noriega. at the Institute of Ecology. “This mandates the need for tools like the one we are publishing to help us predict where new threats in terms of new pathogens and their reservoirs may occur or arise. »
The team plans to continue their research to further test the methodology on other host-pathogen systems and expand the study of disease transmission to predict future outbreaks. The goal is to make the tool easily accessible via an app for the scientific community by the end of 2022.
Source of the story:
Materials provided by University of South Florida. Note: Content may be edited for style and length.
#machine #learning #researchers #predict #outbreaks #pathogens #coronavirus #monkeypox