Tuberculosis (TB) is a deadly infectious killer but Simon Fraser University computing science professor Maxwell Libbrecht is determined to fight it — using artificial intelligence.
Antibiotic resistance occurs when the TB bacterium evolves its ability to fight against the antibiotics meant to cure the patient. This poses an increasing challenge for healthcare practitioners. The WHO estimates that by 2050, TB will be responsible for more than 25 per cent of the 10 million annual deaths from drug-resistant infections. And while recent efforts to improve access and adherence to TB treatment have shown some promise, drug resistance could thwart these efforts by limiting the pool of effective treatment options.
The project will combine the researchers’ expertise in infectious disease, machine learning, and genomics to create a new, artificial intelligence platform for understanding and predicting drug resistance in TB. To do this, the team will develop a new computer algorithm that, when given information about the genomics of drug-resistant and drug-sensitive TB bacteria, will learn from the data and make predictions about new samples of the bacteria based on their genetic traits alone.
Libbrecht and his research collaborators are combining their expertise in infectious disease, machine learning and genomics to create an artificial intelligence platform for understanding and predicting drug resistance in TB. To do this, the team is developing a computer algorithm that, when given information about the genomics of drug-resistant and drug-sensitive TB bacteria, will learn from the data and make predictions about new samples of the bacteria based on their genetic traits alone.
Investigating the genomic mechanisms of TB
“We have a lot of information about the human genome, however we do not have the same depth of information about bacteria. We do not necessarily know what gene or gene combination is getting mutated that is granting the TB its resistance. This study may help us better understand these mechanisms,” says Libbrecht.
And, just as the bacteria will continue to evolve over time, so too will the predictions made by the computer algorithm as it learns from new data given to it. New data sets with the genomes of drug-resistant and drug-sensitive TB and other bacteria provided by researchers from the Harvard Medical School and University College London, among others, will advance the algorithm, improve its reliability, and inform the predictions it makes.
It is an important example of applying the latest in artificial intelligence to a pressing public health challenge.
The study could also have important implications for other types of drug-resistant infections such as malaria, gonorrhea, and diarrheal diseases such as Clostridium difficile (also known as C. diff).