A 14-week health technology “sprint” sponsored by the U.S. Census Bureau and coordinated by the HHS has produced an AI tool that developers claim could revolutionize the way researchers match cancer patients with clinical trials.
The tool, a novel knowledge graph created by a team at the Department of Energy’s Oak Ridge National Laboratory Health Data Sciences Institute, was developed as part of The Opportunity Project Health Sprint, according to a release. Oak Ridge’s team was just 1 of 10 to develop a digital tool to address challenges relevant to medical conditions like cancer or Lyme disease.
Spearheaded by project lead Ioana Danciu, the scientists applied unsupervised machine learning and large-scale graph analytic methodologies to open datasets provided by governmental agencies like the National Cancer Institute and Department of Veterans Affairs. Danciu and her colleagues added clinical trial data to an exoscale knowledge graph that continuously collects and connects data across sources like EHRs and medical ontologies.
“One of the major obstacles facing cancer trial eligibility is the unstructured nature of the data,” Danciu said in the release. “Artificial intelligence and natural language processing tools refine and advance the process of matching cancer patients to promising clinical trials.”
Using AI approaches similar to what Netflix uses to tailor its movie suggestions, the team was able to present information in a much clearer, more meaningful way.
According to the release, the TOP project built upon existing AI research at the Oak Ridge lab, which so far has been used to extract information with signals from low signal-to-noise ratios, developed algorithms capable of accelerating modeling and simulation with little training data and designed novel biomimetic neuromorphic devices capable of detecting epileptic seizures.
“This was an exploratory project,” Danciu said. “Ultimately, we are interested in unleashing the power of the unique computing resources and data science expertise available at ORNL to find end-to-end solutions to big problems with broad societal impact, and clinical trial patient matching certainly fits that description.”
Date: March 13, 2019
Source: AI in Healthcare