Hospitals struggle to retain employees in every role, from the C-suite to the front desk. Nearly 37 percent of hospital workers say they plan to leave their current organization within the next two years, forcing hospitals to replace almost half their staff every five years.
But what if hospital hiring processes could be expedited and simplified such that a computer algorithm could predict how successful a particular applicant would perform in a particular position?
Data analytics tools, or more specifically predictive analytics and artificial intelligence, may be the answer to today's healthcare staffing woes. At least that's what Mike Rosenbaum, CEO of Arena, believes and has built a company around.
For Mr. Rosenbaum, it all started with a question: How can data be applied to human behavior? He first became interested in this question while conducting research at the Cambridge-based Harvard Law School in the 1990s.Then, while working for the White House during the Clinton Administration, he made the argument that labor markets were inefficient but could benefit from data. By combining his passion for numbers with his knowledge of software, Mr. Rosenbaum found ways to calculate the likelihood certain people could achieve certain outcomes in the workforce.
Although Mr. Rosenbaum knew his solution could potentially change the way hiring was done, he wasn't sure which industry could benefit most from his theory. He started a business that applied these ideas to the software engineering industry but did not know much about healthcare or the challenges it faces.
That is until 2009, when he met a hospital executive in a Washington, D.C., suburb with a big problem. She told him how hospitals turnover 25 percent of their staff every year, with some positions in certain departments losing 40 percent of their new hires before they hit the 12-month mark.
Mr. Rosenbaum found the market to test his theory. Since then, Arena's technology has been deployed in nearly 486 healthcare-focused organizations, including New York City-based Mount Sinai Health System, McLean, Va.-based Sunrise Senior Living and Gaithersburg, Md.-based Adventist HealthCare. The software processes millions of job applications annually and boasts a 100 percent success rate in helping organizations improve employee retention. Specifically, Arena has reduced employee turnover at healthcare facilities by a median of 38 percent.
"People sort of get that Amazon predicts what you're going to buy or Netflix predicts what you're going to want to watch, but the idea that we can predict the likelihood that a job applicant is going to achieve at a job is a new experience," Mr. Rosenbaum says.
Here are four thoughts from Arena CEO Mike Rosenbaum on using predictive analytics for hiring in healthcare.
Editor's note: Responses have been lightly edited for length and clarity.
Question: What is the idea behind Arena?
Mike Rosenbaum: We fit into an area called predictive analytics. That's the idea that with all of this data, you not only can look at what is or what has been, but you can look at what is likely to be in the future. What we're doing is collecting large amounts of data on job applicants or employees considering moving into a different role inside of an organization and using that information to predict the likelihood they will to achieve a particular outcome in a particular role. Some outcomes include how likely is this person to stay in the job they applied to; how likely are they to stay in some other job that perhaps they didn't apply to; are they likely to drive up an HCAHPS score; or, are they likely to be an engaged employee?
Q: What types of data do you collect?
MR: We collect data from three buckets … then use predictive analytics and AI to compare that information to outcomes. The first, which is the smallest bucket, is third party data, or data that is already out there in the world — like a person's digital footprint. The second bucket is application data, like a resume. The third bucket is interaction data, which we get directly from a job application. It could be open text information from application criteria questions, or we'll ask a series of questions designed to look a little bit like a test. There might be some personality questions or there might be math cropped in, but when we ask these questions, we may or may not care what the answer is. What we're often looking at is folks' keystrokes, seeing how many seconds they look at something, seeing if they close the browser or open a new browser, or seeing if they skip a question. In other words, we're collecting behavioral data.
Q: How is the data analyzed, or rather, what does the data tell you?
MR: One example of what the data tells us is that there's no such thing as a good nurse. There is certainly a person who is going to achieve a particularly positive outcome in a particular department at a particular facility, maybe on a particular shift. But that same nurse may not achieve those outcomes inside the same facility in a different department, or in the same department but a different facility. What we find is those algorithms, or those predictions, are calculating differently for every facility, as well as every department and every role.
Q: What's next for Arena?
MR: We're expanding into predicting more and more things from a workforce perspective, things like the likelihood someone cares for a patient who is readmitted, or the likelihood that someone is involved in a specific kind of incident. I think about all the disruptions healthcare providers are facing — how rapidly the industry is changing — and those disruptions create huge opportunities for healthcare organizations to think about what the workforce needs and where it needs it.
More articles on artificial intelligence:
Apple's Siri team seeks software engineer with psych experience