You can now read 10 articles in a month for free on BostonGlobe.com. Read as much as you want anywhere and anytime for just 99¢.

The Boston Globe

Business

Data-driven scheduling predicts patient no-shows

With all the advancements in health care, the medical profession still cannot get its appointment book in order.

Doctors are constantly overbooked. Patients constantly rescheduling. One day a waiting room is packed, the next it’s empty.

Continue reading below

So when Gabriel Belfort attended a health care hackathon at the Massachusetts Institute of Technology in 2012, he challenged the coders, engineers, and clinicians there to fix that nagging issue.

“There’s a scheduling problem in medicine,” said Belfort, who at the time was a postdoctoral student studying brain science at MIT. “If you’ve had an appointment and you’ve showed up on time, you’ve probably had to wait.”

That dilemma posed by Belfort generated a very MIT proposal: What if you could use data science to determine which patients are likely to show up and which ones will be no-shows and manage office appointments around those tendencies?

“It was immediately clear to me that this is a problem that computers could solve,” Belfort said.

In short order, Belfort and an ad hoc team of nine people — students and health care professionals — at the hackathon built a prototype to prove out the concept. Then, so excited by the prospect that they could solve one of health care’s chronic pains, Belfort and three others who were strangers before that weekend launched a startup, aptly named Smart Scheduling Inc.

Continue reading below

Here’s the gist: Smart Scheduling mines patient scheduling histories to determine who is more likely to cancel or miss an appointment. It then sends alerts to the scheduling programs that doctor offices use to book appointments.

If a patient is in a high-risk category, for instance, it prompts office schedulers to call with a reminder. If the patient cannot be reached, there is a good chance he will not show up at all. So, the doctors could then book another patient for that time slot, keeping the patient flow consistent throughout the day.

Within months of forming, Smart Scheduling attracted the interest of Healthbox, an accelerator program that invests $50,000 in promising startups and gives them free office space and mentoring. It also landed a meeting with executives at athenahealth Inc., which eventually resulted in Smart Scheduling’s becoming the first startup in the Watertown health information company’s new accelerator program. Athenahealth also made an undisclosed investment to help the company build out its marketing and sales efforts.

So far, Smart Scheduling has attracted some $500,000 in early-stage investment.

Illustration by Peter Hoey

And already it has two large health systems signed up as customers: Martin’s Point Health Care, which runs health centers in Maine, and Steward Health Care System, one of the biggest hospital groups in Massachusetts, where the software is being used by about 40 of its doctors offices.

Dr. Michael Callum, president of Steward Medical Group, said Smart Scheduling helps take some of the ambiguity and guesswork out of making appointments; by eliminating unexpected down time, Steward doctors systemwide are able to see 100 more patients every week.

“When you leave it to the front-desk people in the office, they are not all that good of predicting flow in terms of when patients will show up,” Callum said. “It turns out that Smart Scheduling is much better at predicting that.”

Here is what Smart Scheduling has learned about us as patients: If we are single, or under 40, we are more likely to cancel an appointment than an older or married patient. New patients miss more appointments than regulars.

In general, expecting patients to show up for the 1 p.m. slot is a bad idea. On the other hand, Wednesdays are great, as patients are not likely to cancel on those days.

So far, Smart Scheduling has developed 722 variables that it uses to make predictions, based on an analysis of millions of data points about patients from athenahealth. And the more data Smart Scheduling can crunch, the better it gets at predicting behavior

The company says that, so far, its analysis has proven accurate 70 percent of the time when predicting cancellations.

“If everybody got a better schedule, we’d all be happier,” said Ateet Adhikari, director of the Healthbox accelerator program. “The patients benefit, the doctors benefit, and the insurer benefits. A more efficient system trickles down.”

Smart Scheduling was among the first companies that Healthbox invested in when it launched in Boston in 2012. Since then, it has backed 19 health-related startups.

Smart Scheduling exemplifies a new type of health care startup; instead of going after the big issues in health care — curing cancer, for instance — they are targeting more modest changes to improve the medical experience with technology.

“Companies like Smart Scheduling are dramatically improving health care not by producing a new drug,” said Bill Aulet, director of the Martin Trust Center For MIT Entrepreneurship. “It’s by streamlining the process and getting increased efficiencies.”

Belfort has since gone on to work at a local biotech company, although he remains an adviser to Smart Scheduling. Out of the group that came together to build the original product at the MIT hackathon in 2012, only Chris Moses has stuck around full time, and is now the company’s chief executive.

Improving patient flow in the doctor’s office is just the first step, Moses said. “The next step,” he added, “is to try to figure out who are the sickest patients and who the ones are that need to be seen first.”

You have reached the limit of 10 free articles in a month

Stay informed with unlimited access to Boston’s trusted news source.

  • High-quality journalism from the region’s largest newsroom
  • Convenient access across all of your devices
  • Today’s Headlines daily newsletter
  • Subscriber-only access to exclusive offers, events, contests, eBooks, and more
  • Less than 25¢ a week