Chris Moses, cofounder and CEO of Smart Scheduling, was a neuroscience student at MIT looking for a way to have a broad impact on healthcare when he attended a hackathon at the school in 2012. It was at that event, called Hacking Medicine, where he heard a physician pose the problem he would eventually design a company to address.
The physician said whenever patients miss appointments or cancel them, providers lose revenue, and worse, appointment access for patients who need it is reduced. According to him, it was a big unsolved problem in healthcare.
"Another physician at the hackathon chimed in and said the answer to this problem was data," Mr. Moses says. "If you were to work with the EHR companies you could pull historical data, train a predictive model and try to figure out which patients were likely to no show or cancel. I stepped in with the business hat on and formed the team to help build and design what would become Smart Scheduling."
Soon after, Smart Scheduling teamed up with athenahealth, who invested in the startup in the summer of 2014. To date, the company has raised about $1.1 million, has a team of eight employees and has 600 providers across 25 organizations in 12 states using its product through athenahealth's EHR. Its algorithm is now able to predict which patients will cancel their appointments or no show altogether with 79 percent accuracy.
Mr. Moses took time to talk to Becker's Hospital Review about the difficulty of collecting data and what providers have to gain by predicting patient behavior.
Q: What were the healthcare-specific challenges you encountered as you designed the Smart Scheduling platform?
Chris Moses: One of the first big ones was being HIPAA-compliant. When you're first starting out you may not even know what HIPAA means and the regulations that go into it. Our company is HITRUST-compliant as well, which requires even more technical, administrative and operative processes. Just getting that first set of secure technologies and processes in place are a huge barrier to entering healthcare, and to do so for us was a big deal.
Another challenge was we need a lot of data to build predictive models, but the EHR companies don't necessarily expose that data. You've got everyone moving to digital medical records but not necessarily complying with any interoperability standards. When you look at some of the legacy vendors, like Cerner or Epic, each one of those integrations is going to be a one-off, custom, very difficult integration. They're not robust and scalable even between different sites in the same hospital system. So getting access to the data can be a tough problem. Luckily, athenahealth has a great application program interface and they expose a lot of their data. That allows us to do an integration that scales across all 70,000 of their providers.
Usability was another big challenge. Providers are extremely busy people. We have a product that front desk schedulers use, and it turns out they're just as busy as the doctors. When you're already working in kind of a poorly designed user interface in your EHR, and you're busy, it's hard for an EHR [company] to change any of its interface to support a third party partner. Then you've got to worry about getting the user out of the EHR application into your application in a separate window. It's both a design problem and an adoption problem specific to healthcare.
Q: How does the Smart Scheduling algorithm work?
CM: We pull the historical data and look at pretty much everything we can put our hands on — the clinical, the administrative, the scheduling data. The algorithm looks at trends in how and why people no show or cancel their appointments. We then come up with a prediction that says "Yes, I think this patient for this particular appointment on this day is more likely to cancel." Within athenahealth's EHR, we actually flag that patient and it allows the front desk scheduler to say, "Well, if I really need to squeeze someone in for the day and I need to double book, I can put them over this patient who is more likely to cancel rather than over another patient." These are called "smart open slots." It decreases wait times by improving appointment access. Instead of scheduling three weeks out, a patient can be scheduled today or tomorrow without disrupting the rest of the schedule. That's one of the first workflows.
A second feature is the ability to target those high-risk no shows and say give them a phone call, giving them the opportunity to cancel their appointment early enough that a scheduler has more capacity to refill the cancelled appointment, or more opportunity to allow the patient to confirm. Some practices devote resources to calling all patients that are booked to confirm appointments, which is a big time waster. Other practices don't call anyone, which doesn't give them a good idea of how the scheduled appointments will play out. What we do is pull in those patients who don't confirm a robo-call that goes out and then label them high-risk for cancellation or no show. So you're calling fewer patients and still getting bang for your buck by reducing the no-show rate and increasing your arrival rate.
The third main feature is called auto-rescheduling. Most of the time, appointments that are cancelled and not refilled with waiting patients happen within 48 hours of the appointment time. A scheduler doesn't have enough time to reach out to other patients and slot somebody. It takes a lot of outbound phone calls and coordination. So what we've done is keep an eye out for those appointments that cancel within 48 hours and send automated text messages to the patients who have previously no showed and don't have a future appointment scheduled. We're trying to improve the coordination of care there. We also contact any patients who are on the waiting list; they receive a text message notification about openings that are available for them. If they take one, Smart Scheduling automatically books it, saving any outbound or inbound calls to the practice.
Q: What is 2016 going to look like for Smart Scheduling?
CM: The biggest next step for us is that athenahealth is beginning to bundle Smart Scheduling into new customer implementations, so certain providers will have Smart Scheduling pre-integrated in their EHR. It's helping both us and athenahealth. While all of the features I mentioned are available, not all of the providers using our software have integrated them into their practices yet, so we expect to see more and more customers turning on all of the Smart Scheduling features and integrating them into patient scheduling workflows.