Turning to machine learning for better ROI
Connecting state and local government leaders
An automation tool on the Department of Health and Human Services website now brings together data from multiple sources so that agencies can decide where to invest their grant money.
To ensure that the $770 billion in grant money that federal agencies distribute each year is put to good use, GrantSolutions.gov is replacing manual processes with machine learning and artificial intelligence.
Intelligent Grants Automation
Department of Health and Human Services
Click here for all the 2018 Public Sector Innovation winners
An automation tool on the Department of Health and Human Services website now brings together data from multiple sources so that agencies can decide where to invest their grant money. The site uses a series of machine learning algorithms to analyze and identify patterns that can help predict potential behaviors and outcomes. For example, it might show, with 96 percent accuracy, that an applicant will have negative financial audit findings in the next year.
The tool synthesizes what could be hundreds of pages of information into a paragraph-long summary, said Michael Curtis, GrantSolutions’ executive director. And it can flag applications that require more in-depth review by a human analyst. As a result, a federal grant-maker can decide whether to proceed with an award at a glance.
“We want to know [whether we are] getting the output that we needed from those grants and [whether] those outputs are delivering the outcomes we want,” Curtis said. “Because 1,500 programs partner with us, the AI gives us insight into which programs are delivering the output and then [whether they are] delivering the tactical, economic savings and the strategic impact, such as saving lives.”
Without the use of AI, it would take grant-makers weeks to compile the information they need to decide on awards, said David Martens, GrantSolutions’ director of strategic initiatives.
“When they’re doing tens of thousands of these grant investments a year, they don’t really have that kind of time,” he said. “So they have two choices: They can spend a lot of time trying to gather the information, which means other grants they’re working on don’t get the attention, or they don’t spend the time and they end up making an investment in a place that may not be the right place.”
He added that “this is where machine learning is really an assistive tool for humans.”