Searching for ways to treat coronavirus with NLP
Imagine yourself a plague doctor fighting a disease outbreak in some god-forsaken medieval town.
You’ve probably never had proper medical training and have nothing else but a beak stuffed with herbs and spices to protect you from getting the illness. All your knowledge about treatment consists of a few weird facts about bloodletting and aromatherapy, and there is nowhere you can get any reliable information.
Sounds depressing, isn’t it?
Today we have better means of getting information about diseases, drugs, genes and proteins. It’s easy to open PubMed and find the latest articles on coronaviruses, for example, to help you in your research or practice.
But unfortunately, it’s even easier to get completely lost in the amount of data available. There are 2.5 million new scientific papers published each year, and the number grows exponentially. It becomes harder to find a needle in a haystack when there is more hay added.
This is why we created EBIO — a question-answering system that allows to ask biomedical questions in a natural way and instantly get answers from reliable scientific sources.
Let’s ask EBIO about the effectiveness of ribavirin in severe acute respiratory syndrome (SARS). SARS is a viral respiratory illness caused by the SARS coronavirus, and it was widely treated with ribavirin during the coronavirus outbreak in 2003.
You can see how EBIO finds the answer in the text and highlights it.
Build better voice apps. Get more articles & interviews from voice technology experts at voicetechpodcast.com
Furthermore, Natural Language Processing models allow EBIO to understand the question and find answers that don’t even contain the word “effective”. Here are some of them:
An example of an alternative opinion:
With EBIO, searching through the literature becomes faster and easier. We believe that our solution can help the healthcare industry speed up the drug-discovery cycle and improve patient outcomes.
If you’re interested in knowing more about the project or would like to try it out, let me know!
