This article is more than 5 years old. Information may no longer be current.
Novel tool ‘leverages the power of AI’ to match patients with cancer to clinical trials
A digital tool developed by researchers from the Department of Energy’s Oak Ridge National Laboratory Health Data Sciences Institute uses artificial intelligence to better match patients with cancer to clinical trials, according to a press release.
“One of the major obstacles facing cancer trial eligibility is the unstructured nature of the data,” Ioana Danciu, researcher in the Biomedical Sciences, Engineering and Computing Group at Oak Ridge National Laboratory, said in the release. “Artificial intelligence and natural language processing tools refine and advance the process of matching [patients with] cancer to promising clinical trials.”
Danciu and colleagues comprised one of 10 research teams that developed digital tools to address complex challenges relevant to medical conditions, including cancer, as part of The Opportunity Project Health Sprint — a 14-week effort sponsored by the U.S. Census Bureau, coordinated by HHS, and led by two Presidential Innovation Fellows, according to the release.
HemOnc Today spoke with Danciu about how the tool works, what makes it unique, and whether it will become available to the public.
Question: What is the need for a tool such as this ?
Answer : Clinical trials have great potential in advancing the standard of care. However, matching patients with cancer with clinical trials remains a challenge, mostly due to the unstructured nature of eligibility criteria as well as the clinical documentation. This tool was created out of the first health challenge as a part of The Opportunity Project Health Sprint.
Q: How does it work?
A: The tool leverages the power of artificial intelligence to build large-scale knowledge graphs using deep learning and exascale graph analytics. The tool brings together cancer registry data, medical ontologies and clinical trials among other categories to answer complex questions.
Q: How is this tool unique?
A: The tool is innovative in three ways. First, it processes unstructured eligibility criteria using artificial intelligence and natural language processing techniques. Second, the tool provides recommendations for trials using a machine approach, similar to how Netflix and Amazon provide movie and product recommendations. Third, the tool integrates the data generated in the previous steps with patient-level data and ontologies to paint a bigger picture.
Q: Do you have plans to expand your work ?
A: We plan to collaborate with industry to further realize the potential of artificial intelligence in revolutionizing cancer management. Our research group is developing novel natural language processing techniques, including supervised clustering methods and supervised deep learning for information extraction and classification of medical text. As a part of this larger effort, our participation in The Opportunity Project Health Sprint produced a clinical trials data model that will be the underpinning of an exascale knowledge graph. The data model incorporates feedback from physicians and patient advocates and convenes data from the clinical trials inclusion criteria. Using state-of-the-art text processing techniques, we can extract discrete values from the inclusion criteria. These artificial intelligence capabilities are accessible to other services, such as with our community project clinical model, to provide real-time feedback for patients and clinicians on novel experimental treatments available to them.
Q: When might the tool become available to others ?
A: This approach is available in our environment, but due to HIPAA requirements, not to the entire world. It is yet to be determined if this will become available outside of our environment. Everything that we develop is a part of federal intellectual property.
Q: Is there anything else that you would like to mention?
A: These tools will help the research community and help realize the overarching goal of the cancer moonshot initiative to rapidly advance the treatment of cancer. – by Jennifer Southall
For more information:
Ioana Danciu can be reached at Oak Ridge National Lab, 1 Bethel Valley Road, Oak Ridge, TN 37830; email: danciui@ornl.gov.
Disclosure: Danciu reports no relevant financial disclosures.