Talk to the Veterans Crisis Line now
U.S. flag
An official website of the United States government

Office of Research & Development

print icon sign up for VA Research updates

View: Recent Awards || All Awards


November 22, 2019

VA project receives innovation award for novel AI ecosystem framework

thumbnail
Dr. Gil Alterovitz at the Government Innovation Awards. (Photo by: Neeyanth Kopparapu)

Dr. Gil Alterovitz at the Government Innovation Awards. (Photo by: Neeyanth Kopparapu)

Dr. Gil Alterovitz, VA director of artificial intelligence, was recognized by the Government Innovation Awards for his role in developing a novel data framework to foster collaboration between private industry and government around AI applications.

The awards were presented on Nov. 7, 2019, at a ceremony in Washington D.C. They are given to celebrate technology disruptors, innovators, and emerging leaders in both the public and private sectors. Close to a dozen participants from Alterovitz' project attended the awards banquet.

Alterovitz and his team developed and piloted an AI-able Data Ecosystem Framework. "AI-able" is a term that refers to data that is formatted to facilitate the use of AI technology. A data ecosystem is a collection of infrastructure, analytics, and applications that together can analyze big sets of data.

Alterovitz' project pioneered a new way of creating and leveraging an AI ecosystem where collaborators use AI-able data as a two-way link between government and industry. A traditional data ecosystem typically generates and releases information via a "broadcast" model, which primarily uses one-way communication.

The new AI-able data ecosystem functions like a group telephone call. It uses open, two-way communication for quick, iterative feedback. It permits data, and the final AI-based results, to be more usable to all parties.

Alterovitz' team used a voluntary incentivization framework to encourage collaboration between VA and nongovernment agencies. They developed criteria called AI's Choice and Data's Choice. The criteria addressed these questions:

  • Data's Choice: What makes federal data useful to private industry to help build tools? How can this be measured and incentivized?
  • AI's Choice: What makes private industry's AI-data results usable by federal agencies? How could this be measured and incentivized?

AI Tech Sprint at VA

Alterovitz is now leading a "tech sprint" at VA, which started in the fall of 2018. The AI tech sprint seeks to develop partnerships with outside organizations to apply AI tools to VA data. As part of this effort, more than 10 teams worked together to apply the AI-able Data Ecosystem Framework to leverage VA and other agency data to build AI tools.

One application developed by tech sprint team members used AI to match Veterans with cancer to relevant clinical trials. The program was designed for eventual use by patients and physicians from many different organizations, not just VA.

The application compiled select patient information drawn from VA and the Center for Medicare and Medicaid Services. Using a specialized interface, it applied AI to sort through patient characteristics in the medical record, like age, gender, lab results, and type of cancer. The team then ran that patient information through the National Cancer Institute's interface to obtain a list of potential matches to clinical trials for cancer.

The interface used natural language processing—technology that recognizes free text, rather than structured data fields. Once the program extracted patient information from text descriptions within electronic health records, it matched eligible Veterans to clinical trials for specific types of cancer.

"Given how health care is evolving," Alterovitz told VA Research Currents, "AI is really the only way to move forward in terms of reducing costs and providing better care. AI is key to really taking advantage of data to help Veterans and potentially others, as well."

VA is already using AI to foster improvements in health care for Veterans. Examples include reducing wait times for patient appointments; identifying Veterans at high risk of suicide; and sequencing tumor characteristics to match Veterans with the best cancer treatments.

Alterovitz is a professor at Harvard Medical School and is also on the faculty at Boston Children's Hospital in the Computational Health Informatics Program. He is a recipient of the 2019 Federal 100 Award for his work to help government get ready to adopt AI technology.


Questions about the R&D website? Email the Web Team.

Any health information on this website is strictly for informational purposes and is not intended as medical advice. It should not be used to diagnose or treat any condition.