Author: Faculty of Science (Wiida Fourie-Basson) Research at Stellenbosch University (SU) into long Covid has received an “Artificial Intelligence” boost worth over R11.75 million (€600 000) over the next three years via the Erasmus.AI search engine. (Link)
On the photo above, from left to right, Prof. Louise Warnich, Daniel Erasmus, and Prof. Resia Pretorius. Photo: Ignus Dreyer
The Erasmus.AI engine, lead by SU alumnus Daniel Erasmus, is an immensely powerful search engine that can analyse web-scale large bodies of unstructured text – in practice over 250 000 000 URL’s are processed per day and translating from fifteen different languages. This includes all open access research articles and abstracts published on PubMed.
“If you consider that nearly 5 000 medical articles are published each day, one realises there is no way that a medical specialist or researcher can remain current. We are here to give overview, enhance collaboration, in short − to change the architecture of discovery,” Erasmus explains during a recent visit to SU.
The purpose of the visit was to sign a research agreement with physiologist Prof Resia Pretorius, head of the Department of Physiological Sciences and currently at the forefront of research efforts worldwide to understand the role of persistent microclots in the blood samples of individuals suffering from long COVID. Current thinking is that these insoluble microclots inhibit or even temporarily block blood flow in capillaries and subsequent oxygen transfer to tissue. The lack of oxygen in various parts of the body can account for many of the symptoms of long Covid.
Even more intriguing, however, is the presence of a significant number of large molecules known to significantly reduce the body’s ability to regulate clot formation, trapped in these microclots, as well as a range of antibodies. In an article published in October 2022 in the journal Cardiovascular Diabetology, Pretorius and co-authors write that these antibodies may have significance in the immune response following acute Covid-19 illness, or driving auto-immunity in some individuals with long Covid. There is however little to no data available on these antibodies.
Pretorius and her research team have since been combing research data bases for clues to better understand the role and function of some of these antibodies.
Now, with access to the Erasmus.AI search engine and platform, they can harness the super-computing power of AI to do the searching for them, as well as visualise in one go what is out there by means of a user-friendly interface.
Pretorius is also curious to better understand the ability of enzymes such as serrapeptase and nattokinase to break down blood clots: “Currently we only have anecdotal evidence from individuals who reported improved symptoms after taking these enzymes,” she explains. But maybe there is something in the literature out there that we have overlooked.
According to Erasmus, AI can provide researchers with clinical, laboratory and anecdotal reports from billions of sources out there: “AI can provide an overview where none existed. It is like a ‘super Google’ for professionals. Generative AI brings together the insight associated with a large-scale view, and when combined with serendipitous discovery, it most often leads to new insight and understanding,” he explains.
This combination of human and artificial intelligence is what is needed to tackle the global health crisis brought on by long Covid, he adds: “The impact of long Covid is going to dwarf the pandemic phase. We need to deal with long Covid as a matter of urgency,” he warns.
Erasmus, who obtained his degree in engineering at SU, is currently based in Amsterdam in The Netherlands.