Can NLP help patients understand clinical trials?
We’re giving ourselves 12 weeks to try something new
A clinical trial is a research study to see if a new medical treatment (like a drug, diet, or device) works and is safe for people. To join a clinical trial, you need to understand what you’re signing up for. But too many clinical trials are described using inaccessible medical jargon only a scientist could love.
When potential participants have trouble understanding what’s involved, it makes it harder to make informed choices, and slows down the pace of research. In an ideal world, every medical researcher would work with a trained expert who can write simple but accurate medical language. For example, the BreastCancerTrials.org website hires experts to write high-quality summaries of hundreds of clinical trials. But until we have better trial descriptions, participants still have to fight their way through complicated text.
Our multidisciplinary research technology team at UC San Francisco launched the UCSF Clinical Trials website in 2016, and we’ve been frustrated at how difficult it’s been to provide simple explanations of each trial.
That’s why UCSF is teaming up with information scientists at Drexel University, to see if we can use natural language processing (NLP) techniques to help people make sense of trials.
Can computers help patients make sense of clinical trial descriptions? We’re launching a 12-week clinical trial NLP challenge to find out.
Over the next twelve weeks, the Drexel team is going to apply a series of NLP techniques to clinical trial descriptions and metadata — and see if we can come up with ways to help patients better understand what’s going on with a given trial, inspired by prior work extracting meaning from trial descriptions.
Alexa and Siri can make NLP look easy, but making sense of complex human language is much easier said than done. We don’t really know what’s going to happen with this project. It might be an embarrassing failure, or wildly successful, or more likely something in between.
But no matter the outcome, we expect to learn a lot about applying NLP techniques to this very real-world problem.
We’ll be blogging and posting our code on GitHub, so you can learn along with us. Please subscribe to follow along!
