Logistics is something we traditionally associate with the trucking or the package delivery industry. In fact, a recent article in the Economist estimates that the delivery of 25 packages equals roughly 15 septillion (trillion trillion) possible routes. If these kinds of logistics blow your mind, you’re not alone. That’s why many companies dealing in complicated webs of variables like this are turning to new technologies like artificial intelligence to help streamline and optimize their operations.
What if we took the concepts behind shipping logistics and applied them to the healthcare space?
Imagine a healthcare organization with multiple locations, each staffed with providers across multiple specialties—individuals who are not interchangeable—operating under a wide range of room availability and scheduling constraints. Maybe not 15 septillion variables but, in some cases, we’re getting close. This is a very typical scenario, and one that is all-too-often addressed with clunky spreadsheets or whiteboard-based planning. Resources are allocated, but not optimally. “Staffing to peak demand” or “reserving capacity just in case” leads to poor utilization, which is not only wasteful (and costly), but also represents a missed opportunity to better deploy scarce resources to serve the patients who need them.
One area of healthcare where approaching challenges with a logistical mindset makes the most sense is telemedicine. This growing space is predicted to be worth more than $66 billion by the end of the year 2021. Telemedicine, or telehealth, originated from the need for practitioners in rural areas—or other settings having a shortage of specialists—to be able to consult with experts to provide care. Historically, regulatory barriers, unfavorable reimbursement models, and inconsistent coverage for these kinds of services by private insurers have slowed adoption for telemedicine. However, as these barriers start to fall, potential for heightened use of this model increases.
Perhaps the biggest opportunity for the effective use of telemedicine is in the area of “teleconsults,” where a local care provider is able to quickly confer with a remote specialist colleague. By routing slices of top specialists’ time to distant locations in a real time, the door is opened for previously-isolated populations of patients to benefit from better diagnosis and treatment decisions. Ultimately, this ability to dynamically pull in experts where needed, even across large and isolated geographic areas, promises to improve patient access to care, improve outcomes and reduce costs.
As telemedicine adoption increases, the healthcare space is faced with scaling issues. When you have hundreds of doctors, working at all different times of day at different facilities, all with their own specialties you have a “combinatorial explosion” (or what we internally sometimes call a “big mess”). This sheer number of combinations is one of the biggest challenges in constructing facility coverage rosters and matching physicians to consults. The end goal is to operate a lean network of specialists, providing many different areas of expertise, available at all times of the day for on-demand consults. If you look at this from a purely mathematical standpoint, you can start to see the massive logistics involved in making this type of model efficient and effective.
In order to put the scale of these challenges in perspective, let’s imagine that there is a healthcare organization that has three facilities and three doctors. For each facility, one of seven possible physician rosters can be selected.
|Facility A||Facility B||Facility C|
|Physician one only|
|Physician two only|
|Physician three only|
|Physicians one & two||X||X|
|Physicians one & three|
|Physicians two & three|
|All three physicians||X|
In the above example, Facility A has physicians one and two covering. Facility B also has physicians one and two covering, and Facility C has all three physicians covering. Because remote physicians can be on more than one roster, you can start to see where the combinations, depending on specific constraints, can start to rise exponentially.
In this example, there are 343 possible remote roster possibilities, or (7 x 7 x 7) across the three facilities combined. For the more mathy folks out there, the formula for this, excluding the case of an empty roster, is: (2physicians – 1)facilities. In this case of 3 physicians for 3 facilities, we have (23 – 1)3 = 73 = 343 possible facility-physician assignments. As you add physicians and facilities to your telemedicine network, the number of possible rosters starts to explode exponentially:
A moderate telehealth network size might be 100 doctors and 1000 facilities, so you can see how this quickly becomes a mess, requiring human schedulers to resort to “best effort” or “over coverage” approaches, which may be far from optimal. Wrapping the human brain around this kind of scheduling complexity is like playing a game of chess against a grandmaster, day after day. But unlike chess, attempting to optimize a telemedicine network by hand isn’t fun. Fortunately, advances in artificial intelligence (AI) make it possible to automate away this complex and thankless workflow.
Using technology to solve logistical challenges
When you start to look at telemedicine through the lens of logistics, some of the solutions that are currently solving logistical problems in other industries – say in the shipping sector – are appealing for this space as well. Specifically, technology such as artificial intelligence can make a huge difference – taking in massive amounts of data, understanding complex constraints, and then intelligently routing little packets of physician time around the network in a way that makes the best possible logistical sense.
Looking beyond the hype surrounding AI and understanding its practical application in the healthcare space, it’s easier to see the benefits. The real power of AI is that it enables computers to more intelligently search through all of the possible doctor, facility, consult, time, space combinations, such as those outlined above. So rather than wandering aimlessly through this unfathomable universe of possibilities, artificial intelligence allows the computer to chart a path to the right answer. In the case of telemedicine, this “right answer” is a set of rosters that ensures consult availability and low wait times, as well as ensuring doctors are well utilized by specialty.
This kind of proactive, pragmatic approach to telemedicine can help to streamline the scaling challenges that come up as networks grow and adoption of the model increases. According to the Council of Accountable Physician Practices (CAPP), telemedicine has the “potential to transform health care delivery….access, quality, and efficiency.” Overcoming barriers steeped in regulatory and payment issues is definitely a challenge to widespread adoption. Despite this, adoption continues to rise because the model makes good sense in a connected world and is transforming from a “vision to a business model.”
AI can solve, at the very least, the logistical challenges that come as telemedicine continues to grow exponentially. By letting the machine do its job of solving the complex game of telemedicine logistics, we empower physicians to do their job—helping more people across more settings.
Photo: Filograph, Getty Images