Department of Energy (DOE) researchers developed a new tool to connect cancer patients with clinical trials. The tool uses a Netflix-style of analytics to recommend studies that would be a good fit for particular patients.

This technological solution could help improve trial enrollment in several important ways, says Ioana Danciu, MS, biomedical scientist, engineering and computing group at the Health Data Sciences Institute of Oak Ridge National Laboratory (ORNL) in Oak Ridge, TN.

The approach improves trial matching by using natural language processing techniques on the unstructured eligibility criteria to extract certain data elements such as labs. It also clusters clinical trials that are similar, using an approach called agglomerative clustering — similar to what Netflix uses for movie recommendations, Danciu explains.

“The approach also improves matching by adding this information into an exascale knowledge graph that connects data from disparate sources, such as electronic health records, medical ontologies, and public datasets,” she adds.

ORNL’s SmartClinicalTrials capability builds on an existing collaboration between the DOE and the National Cancer Institute, Danciu notes.

ORNL researchers are leading a pilot effort to expand cancer surveillance capabilities and build statistical models that are capable of predicting the clinical course and outcomes for different types of cancer, she explains.

“As part of The Opportunity Project, we collaborated with data owners from the National Cancer Institute and the Department of Veterans Affairs,” Danciu says.

The tool has artificial intelligence (AI) capabilities as it can store data, continuously add new information, and allow for computational analysis and knowledge discovery.

“We are using natural language processing in the context of AI to understand the unstructured inclusion criteria, group similar trials together, and perform exascale graph analytics,” Danciu says.

ORNL’s AI Initiative helped make development of the tool possible, accelerating scientific research breakthroughs.

It is a benefit to cancer trial enrollment because eligibility data often are unstructured in nature, whereas the use of an artificial intelligence tool helps to improve the process of matching cancer patients to clinical trials, Danciu notes.