Founder and CEO Suvas Vajracharya, Ph.D., had the opportunity recently to join Dr. Dianne for an episode of The Doctor’s Life podcast to discuss Lightning Bolt’s AI-based shift scheduling as a solution for preventing physician burnout. Enjoy part one of their conversation below, find part two here, or discover more episodes of The Doctor’s Life on iTunes or Google Play.
Dr. Dianne: Welcome to The Doctor’s Life podcast. I’m your host, Dr. Dianne. Think of the podcast as a mobile doctor’s lounge where we take some time out and explore topics that doctors are interested in related to what it means to be a doctor in today’s world and creating a rich, satisfying, and meaningful life as well as making the healthcare system better for doctors and patients. If you are a doctor, you’ll feel right at home, and if you aren’t a doctor, you’ll get a better understanding of doctors and what we think about and talk about. I know you will enjoy the conversation. Let’s jump in.
Dr. Dianne: Hey everyone. Welcome to The Doctor’s Life podcast. We’re so happy to have you. What are we talking about today? We’re going to be talking about artificial intelligence. More specifically, scheduling of docs and to their workflow. We’re going to be talking about how that relates to physician burnout. It actually does. And surprise, scheduling actually helps with physician burnout. I’m so fortunate to have today with us the CEO of Lightning Bolt Solutions, Suvas Vajracharya. Just a disclaimer, I am on the advisory board of Lightning Bolt Solutions, but I do really believe in this company and what they’re doing. I’m very happy to be on the advisory board. It’s an unpaid position. Full disclosure.
So, prior to founding Lightning Bolt, Suvas worked as a staff scientist at the Los Alamos National Lab, scheduling massively parallel supercomputers. That was a total mouthful, Suvas. So nice to have you on the show today.
Suvas: Thank you so much for having me. I’m really excited about the discussion today.
Dr. Dianne: Yeah, me too. We had a fun conversation when we talked before about what we were going to be talking about. It’s really cool. I was looking at your website. I love what you write there where it says, “The number of possible physician shift schedules can be greater than the number of atoms in the universe.” Which is so true. I experienced that myself when I was in anesthesia practice.
Scheduling is hard, and it’s difficult to make it fair, easy, and trouble free. We’ll talk about how Lightning Bolt Solutions does that, and your journey as to how you came to found Lightning Bolt Solutions is so interesting. I’m going to turn the mic over to you. I would love to hear, and I’m sure everyone else wants to hear, about how you came to found this company.
Suvas: Yeah, thank you. As you said in the bio, I used to work as a staff scientist at Los Alamos National Laboratory, and at that time, I was scheduling not people but supercomputers. One of my friends from high school knew about my expertise in scheduling with computers, so he challenged me to a problem that he was having with scheduling a group of internal medicine doctors in Wisconsin. I took up his challenge; it wasn’t intended to be a business. I thought I would create a software program to help Leo to schedule his group. The thing that I found out immediately was that this was not an easy task — not for people and even computers. As Dianne pointed out, part of the challenge is that the number of possible combinations in a schedule can be very confounding, more than the number of atoms in the universe. We can’t have people or computers look at every possible atom — that’s where the AI comes in. The artificial intelligence needs to just look at the ones that are reasonable, not look at every possible schedule and weed through that, and quickly come to a solution.
That’s how I got involved in the project, and I left it alone for a while; and many years later I thought if that was useful for Leo, maybe it’s also useful for other groups. Sure enough, we had customers that wanted to use our system right away, and to this day, Lightning Bolt is still bootstrapped, and we haven’t been taking any outside investment. It’s really been a product that’s been led by doctors from the very beginning. We are engaged with doctors on a daily basis and getting useful feedback to continue to bring solutions in scheduling.
Dr. Dianne: Yeah. You know, as doctors’ scheduling is really a challenge, you often have to have a dedicated person to do it or a dedicated persons. Usually, that is a really thankless job of that physician to do. It takes them out of their clinical work. Scheduling can seem really unfair, especially when it comes to our work-life balance, work-life integration. Sometimes it can feel like — and it may actually be true — some people are getting more holidays and weekends and others are… Your scheduling can make things more transparent. How does it do that?
Suvas: Firstly, I completely agree with you about how difficult it is to build schedules. Typically, it’s actually a doctor that’s involved in creating schedules because they don’t trust that to be done by someone else because it impacts their lives, and it makes sense. But it does mean a commitment. To build schedules for a group can take ten, 20, or more hours per month when they should be using that time to see patients. It’s currently a big time waste. It’s something that’s not pleasant to do. It’s like having to do taxes. Every month you’ve got to do taxes. And despite the effort on the scheduler’s part, there’s always concerns on: Is the schedule fair? Am I taking more calls than my colleagues? I worked last Christmas and it seems like I’m always on call during holidays. How do you address that?
The thing that’s really surprising to us is that there’s always this concern that maybe the schedule was produced in this biased way, but in fact, when you look at the actual data, the statistics, the scheduler always ends up taking unwanted shifts or taking more calls because it’s difficult to meet everybody’s requirements. Our experience is that the scheduler always has good intentions, but when you put together all the requirements for doctors to be on vacation and we all have work-life balance to consider, or we may have family to take care of. An example may be a couple that works at a practice may be wanting… they can’t both be on call on the same day if they have children. Those are the nuances of scheduling that you need to consider, whether you’re a part-time person or they can only work certain days of the week.
What our system, Lightning Bolt, does is to individualize the schedule. So you create a profile for each doctor and their preferences on what dates they are available, what types of shifts they would like, and based on those rules, the system builds a schedule around that. The result is that it’s time saved for the person who does the schedule, but it also addresses the concern about bias and equity, and it produces a schedule that meets the individual work-life balance needs.
Dr. Dianne: When you’re saying that they take into account the rules… And we did find that there’s kind of these little rules that they don’t become obvious until you start unearthing them. They are sort of in the scheduler’s head or everybody’s head that the scheduler knows. For example, the guy that used to do my schedule when I was in anesthesia practice, he kind of knew that it was OK for me if I didn’t work a full day, or that kind of thing, or that I had kids, without asking, or that it was Mother’s Day or something like that. There’s all these little unwritten rules. Doctor so–and–so likes to work with that doctor. Or you’re given that example. Doctor so-and-so will work with doctor so-and-so only on Tuesdays if the cath lab is open. There’s so many little algorithms that we just keep in our head. How does the scheduler know all these little quirks? Well actually, they’re not quirks; they’re really important to…since we don’t want to be machines or cogs in the wheel. We’re individuals.
Suvas: Yeah. Definitely true, and when we set up a new group with scheduling rules, they often consist of hundreds of rules. Our clients are usually surprised by how many rules there are. You had mentioned in the back of their heads, the person who had been doing the schedule and they had been doing it for multiple years. They’ve not only learned all the rules but how do you actually build a schedule? What might be the tricks of the trade around the holidays or weekends? We capture all that when we first have a consultation with the group. We ask them what are your rules? Essentially doing a mind dump or brain dump. We call it a brain dump, which is to get all the information about what this current scheduler knows about exactly what you mentioned. What are the preferences of certain doctors? Some may actually prefer to work on certain weekends. And others, when they work on a weekend, they want both Saturday and Sunday to be on-call. There’s a few that say, “I’m okay with splitting my weekend.” There’s a lot of individual variance, and all of those need to be considered for it to be a good schedule.
Dr. Dianne: So how does the artificial intelligence part of it come in? How does that help out with scheduling? AI always sounds like some black box where like… I don’t get it, honestly.
Suvas: There’s many types of artificial intelligence. There’s a lot of discussion not only in healthcare but in all the industries about the use of AI for looking at data, machine learning, and there’s many different types of artificial intelligence that actually learn from data and recognize patterns. The one that we focus on is not so much on learning new things. Scheduling rules are what they are; there’s no need to derive new rules or come up with something new. The system is really looking to see how you can meet all the rules at the same time.
A lot of our clients think of scheduling as solving this really complicated puzzle. They liken it to solving a 20-dimensional sudoku puzzle or a Rubik’s Cube. There’s a lot of thinking involved. You mentioned the number of combinations or possible schedules being equal to or greater than the number of atoms in the universe. That’s often the case for these puzzles. Like, even playing chess, that would be another example. The number of possible chess games that could be had is similarly very complex and could be as many as the number of atoms in the universe. How do you navigate through all those possible schedules? You can’t look at every single one of them, whether you’re a person or a computer. Even the fastest computer in the world, if they would look at every possible chess outcome or every possible schedule, they wouldn’t finish within one’s lifetime.
What we need to do then is look at the ones that are possible solutions and so that makes it a little bit quicker. That’s where AI is useful because we want to intelligently make choices on who to schedule by looking forward. Usually, when you schedule — and you may have had this experience, Dianne, since you did the schedule — scheduling choices have a domino effect. If you make a choice, then they have a ripple effect.
An example would be if I scheduled somebody on a weekend, there might be a rule that says if you are on call this weekend, then I have to schedule that person again on call the next weekend. But if the person that is scheduled the next weekend is on vacation, then you need to really think ahead of time about who do you put on this weekend so you don’t put the person that’s on vacation next weekend. That’s an example of how you need to think several steps ahead, much like playing a chess game. You need to think about all the consequences about what you do today, not only next weekend but it can have an impact on the schedule months from now.
For example, how you build a schedule very early in the year, like if you’re building a call schedule for the whole year, your choices will be very limited in December because of holidays and doctors have vacations set up. If your goal is to balance the number of calls for the whole year, you really need to start thinking about that in January or February. Since there is a doctor that is going to be gone for all of December, you may need to start thinking about putting that doctor more frequently in January, February, and March when the doctor is available. Otherwise, when December comes, you’re kind of in a bind because you can’t catch up that doctor. It’s too late. They are already on vacation.
Those are some examples of analogies with a chess game because we need to think several steps ahead.
Dr. Dianne: There’s other industries that are using this AI scheduling too, right? Reading from your website, the NFL uses it and professional sports teams use it so that they can protect… I like the idea because I think of the same as doctors, right? They are very valuable assets as individuals. You want to protect them, and you don’t want them to get burned out and pissed off and leave.
Suvas: You’re right. There’s a lot of analogies, and you wouldn’t think about having a quarterback do the schedule of the NFL. In the case of healthcare, you do have doctors who do the schedule. And you’re right that the NFL, Major League Baseball, and professional basketball, they all use the same sort of technology that we use to build the schedule. It’s kind of interesting because sports scheduling went into operations research or AI-based technology around the same time that we did in the mid-2000s. That’s when this technology became possible, and there’s a great story how in baseball, there used to be a couple who built the schedule for Major League Baseball and they were experts at it and they got really good at solving this puzzle. As things got more complex and as the technology became more available, we transitioned to this technology, we call it operations research. That’s the technology we’re using.
They transitioned to this technology around the same time. We both transitioned to it. The reason for that is that just like when you build a schedule for doctor’s work, scheduling baseball teams is very complex because baseball teams want the same sort of thing. They don’t want to be working every July 4th when they could be home with their family. The bulk of teams don’t want to be playing football every Thanksgiving, and they don’t want to be traveling on consecutive days, just like doctors don’t want to be scheduled consecutive days on call.
There’s also studies about how there are more injuries for players when they play on Thursday and then again play on Thursday, when they play a Sunday and a Thursday. There’s actually a correlation between how frequently they play and the injuries, and that’s similar to the requirements for doctors — that they don’t take too many calls per week or per month and that you want to separate the calls as much as you can. All these variables about the frequency of games played, what days they’re played, are they played on holidays, have a lot of parallels to scheduling doctors. Professional sports, a long time ago, in the mid-2000s, switched to this technology that helps them not only from the perspective from the individual players but even helps improve their business because they are also looking to see what kind of schedule would also have the greatest audience. You don’t want to schedule at a time when there is some other event.
An example was they wanted to not schedule a game when the pope was visiting. So many variables and there’s the same sort of consideration in building the schedule for doctors because there is consideration for each doctor who has an amount of requirements on how they want to work, and that’s actually part of our solution to this burnout is giving more control to the physicians on how they like to be scheduled. Having the vacations they want and having the type of schedule. There’s part-time or working certain days of the week. Those are all variables you need to consider. Just like in the sports scheduling, there are many, many rules to consider.
Dianne Ansari-Winn, MD, MPH, is a board-certified anesthesiologist and nationally recognized author, speaker, and coach. Dr. Dianne is known for her powerful and practical coaching, which she shares with her exclusive one-on-one physician clients. She is the host of The Doctor’s Life podcast and is the founder of the Physician Vitality Institute, which has the mission of helping doctors to heal the world through being healthy in mind, body, and spirit. Dr. Dianne is a member of Lightning Bolt’s advisory board.
Continue reading part two of the interview here.
The interview originally aired July 27, 2018 on The Doctor’s Life. The transcript has been lightly edited for clarity.