Scientists develop algorithm to identify animal viruses capable of transmitting to humans

Researchers from the University of Glasgow in the United Kingdom created an AI-based system to identify diseases that could pass from animals to humans.

They assembled a database of more than 800 zoonotic viruses. They then trained the algorithm to interpret their unique genomic structures.

According to the scientists, they were able to identify zoonosis candidates using only host-range signatures encoded in viral genomes.

“Viral genomes may possess generalizable characteristics that do not depend on the taxonomic relationships of viruses, and may contribute to their adaptation to human infection,” the paper states.

The researchers aimed to identify threats before they escalate into full-blown epidemics. Any potential zoonosis can be referred to specialists who will conduct more thorough investigations.

The team stressed that manually identifying animal viruses capable of transmitting to humans can be incredibly challenging.

“This undermines prioritising high-risk candidates for early investigation and ensuring readiness for future outbreaks,” they added.

In July, AI laboratory DeepMind specialists assembled the most comprehensive database of human protein structures, created by the AlphaFold neural network.

In late July, IBM trained an AI to predict the progression of Parkinson’s disease symptoms.

In June, the World Health Organization published the first global report on the use of artificial intelligence in medicine and outlined the key principles for its development.

Subscribe to ForkLog’s Telegram news: ForkLog AI — all the news from the AI world!