Matt McBride On Improving Telemedicine Through A Patient Attendance Predictor

Everyone in healthcare services knows how frustrating it is to manage appointment schedules, with many patients canceling or simply not showing up. Matt McBride of Mend addresses this problem by creating their very own Patient Attendance Predictor. He sits down with Chad Burmeister to discuss how AI can determine the likelihood of patients appearing in their telemedicine appointments. Matt explains the various factors considered in this process, particularly how the program changes every single day. He also delves into the other possibilities this technology could evolve into, especially in elevating customer experience and satisfaction.

---

Listen to the podcast here:

Matt McBride On Improving Telemedicine Through A Patient Attendance Predictor

I am thrilled to be hosting Matt McBride, the CEO and Founder of Mend, a terrific Telehealth application business and platform that we'll learn a bit about during our conversation. I'm heading to one of my health providers as soon as we leave this conversation. It's a timely conversation for me because some things that I've learned by doing my background research, Matt and how you use AI, I'm curious to dig into. Welcome. I’m glad you're here with us. If you wouldn't mind, give our readers a little bit of background and introduce yourself in any way you like.

Rich, thank you. It's great to be here. I’m Matt McBride, CEO and Cofounder of Mend. Mend is the leading telemedicine and patient engagement platform. Telemedicine is one of lots of different things that we do from reminders to scheduling, digital forms, payments and a lot of those complicated touchpoints with patients when they're not at a practice's location. That's where we come in and we help transform those workflows into a digital format.

Literally, with a text, you click a link, put in your date of birth, you can connect on Telehealth, you can do your forms and all sorts of things in preparation for your appointment. One of the things that make our platform unique is we span virtual and in-person appointments. It's one complete patient engagement and patient communication platform that practices, build on, and modernizes a lot of the workflows. It helps keep people safe in this COVID world that we live in.

We'll dig into that and how you got to where you are and where you're going with it from here. I want to start a little further back than that. I want to talk about 6- or 7-year-old Matt McBride. We’re coming upon the end of summer. I was thinking about how we spend our summers in our youth. What do we do with our time? How do we spend our day? What were the things that you enjoyed? Where did you get your energy? How did you spend your time doing the things that you like the best when you were in that period of your life?

I have kids as well. We had more freedoms when we were younger. I was lucky to have a lot of kids my age in my neighborhood. We were always out playing different sports, It might be baseball, tennis or whatever. In Michigan, it would snow and in the wintertime so we go out. We had some good sledding hills and things like that. I did a lot of the typical things playing with kids in the neighborhood in the summers.

The first ingredient in starting something is focusing on what you're really passionate about.

Active, outdoor and freedom are three things that I take away from what you described. How does that match what it takes to have an idea and collect a bunch of people together? I noticed you've got three other cofounders. How does that take that same spirit to found a business?

Another way that I spend my time as a kid was with computers. Gaming systems and computers were starting to become popular. I did work some of the summers caddying, too. I had saved up enough money when I was eleven so I could buy my own computer. I saved up and bought my first computer when I was eleven. Technology has always been there. It's something that I've been interested in.

The time I was growing up, the internet was beginning. It was in its infancy and it was this new, exciting thing you could all of a sudden connect with people all over the globe. You mentioned the three other cofounders in the business, they've always been more on the sales side and I've been the tech guy. That technology piece has always been something that I've been passionate about. The first ingredient to starting something is you want it to be something that you're passionate about and Mend was that business.

My friends and I have provided the bulk of the capital in the business. We've all built and exited other companies. Technology is something that we're passionate about but also, we wanted to do something that could have an impact and we're also achieving that as well at Mend. A lot of our customers are in the behavioral health space and primary care space. We played a role during this pandemic as well. Being entrepreneurial and starting a company, I feel like it was something that I was born with. My parents always thought I would start companies one day.

I'm going to give you maybe indirect credit and I don't know if this is you or if it's a combination of you and your partners but there's something that I noticed that I've never seen before even in small and mid-sized startups. Everyone in your company is represented on the people page of the About Us section of your website. The bigger the company, that goes in a different direction.

Even with startups and small to mid-sized companies, it's generally 3, 4, 5, 6 people that are leading functions in a company that is there but not Mend. With Mend, every single person in your business and the role that they're in is on the website. It seems like that must represent a piece of who you all are in terms of how you feel about the connectivity of the people in your business and the importance of that. Tell me about that.

Patient Attendance Predictor: In the United States, the healthcare system averages about a 23% no-show rate. Mixing that in with the late/cancel rate adds up to 30% of wasted schedule every day.

Without all of the amazing people that we have, this is impossible. The ideas are the easy part and starting a business is the actual hard. Now we have to figure out how to make this thing happen. Doing that by yourself to build a company of our scale and size, there's no way. Everybody is a valuable member of the team. We see ourselves as a team. There are no egos at the top of our company. Everybody is extremely important and valuable and yet it's part of our onboarding process that we make sure we get a picture and we get you on the website. I think it’s something to people. It's like maybe when you got your first business card, too. People like having that little recognition, seeing this big team that they're a part of, helping to build this company and build towards our mission of helping people lead happier and healthier lives.

It's an unusual way to express an outward sign of your culture that you don't see that often. It shows great things where you can take your business based on that foundation. Help me a little bit. I want to dig into the AI side of your business since this is an AI discussion. One of the things I saw were these data points about cancellations, the cost of cancellations to practice and how AI can help predict cancellations. Here I am about to go to an appointment when we finish our conversation. I'm certainly not about to cancel my appointment. Help me understand, number one, the value of AI to your business, your clients and how it works. What's the engine inside that helps it drive that value proposition for your clients?

We see AI as being a major part of our business. Long-term, the value that it will create is huge. We're looking at our patient attendance predictor as helping to solve the patient access problem. We mirror this trend globally. In the United States, our healthcare system averages about a 23% no-show rate. If you mix in a 23% no-show rate with the late cancel rate, we can see groups with 30% waste on their schedule every day. There are different tactics with overbooking and whatnot that aren't great but if you think about that for a second, if somewhere in the neighborhood of 20% to 30% of the appointments on a given schedule are never going to happen, that's about 1 in 4 to almost 1 in 3 appointments that you're putting on the schedule are not going to happen.

This compounds every day for healthcare practice and then leads to long wait times even to get in. We're applying AI to other parts of the experience and patient experience between their healthcare providers. We're looking at the schedule first and how we can help predict when these types of events are going to occur. That’s the first part for us because going back to Mend being a patient engagement platform with Telehealth and forms and all these other things that we do, we are good at communicating with the patient. We have a combination of AI to look at the data and find who's not going to attend and then we have all the automation built-in behind that to go out and ask the person, “Are you coming in or not?” and make it easy for them to tell us. It's a combination of AI, in-patient engagement and our patient attendance predictor.

Like everybody else nowadays, I get text messages from my providers, “Confirm that you're coming. Reminders of your appointment.” All those things help solve that problem. How in the world do you do the predictors? How does AI go through and help the provider predict who's not going to come so that they can take proactive action?

Coming up with ideas is the easy part of starting a business. The hard part starts with figuring out how to make them happen.

The first is we work with standard demographics and appointments interfaces. We're not asking for a bunch of additional data that would be hard for us to capture. The data that's pretty standard within healthcare to plug into any other system. From there, we're deriving about 50 to 60 different factors. None of which in healthcare and AI ethics are important. We're not looking at any data like race, religion or anything like that. These are things like, what type of appointment is it? When is it happening during the day? What's your distance to that appointment? What’s your history of attending appointments? There are about 50 to 60 different factors and we try to predict attendance, no-show and late cancel.

We predict those three independently and then we put that all together again to make a final prediction. That's where the technology begins. You were talking about the reminders and the notifications. We have reminders. They go out via text, email and phone as part of the system. We can now start communicating differently with you. If we thought you wouldn't show up to your appointment, we might start communicating differently and give you different options.

For example, we allow you to cancel it through a text message or to confirm it. The AI is never going to quite be 100% sure. It might be confident that perhaps you're not going to attend. We send those notifications and you might get a different phone call, email or text message if we thought that you weren't going to come in. The AI is then changing every day as you get closer to your appointment, it's dynamic and the prediction is changing. If we see you opening or clicking, you tell us you're coming or not coming, it's a combination of the AI and then we're engaging intelligently and differently. If somebody is not going to come in, making it easy for them to say they're not going to come in so we can free up that slot for somebody that does need it.

At the same time, your profile of that individual is learning from that behavior so that you can continue to predict at an even more accurate rate.

That’s one of the key things with AI. We can continue to change it daily. Some other interesting things that we want to do in this COVID environment or this post-COVID world is if we detected that you weren't going to come into your appointment, maybe it's because you're having a transportation issue or something else. What if we offered you a telemedicine visit instead? That would be appealing and more convenient so you didn't have to take time off of work, get childcare or whatever the case may be. You could still stay engaged in your care. It's the combination of AI and patient engagement that makes it unique because we can reach out intelligently and start communicating with people to see if they're going to come in or not.

Where do you go next? What's on the roadmap for how to apply AI in the next area that you're heading to?

In some other areas for AI, we are looking at natural language processing for transcription during telemedicine visits and calculate wait times. Everybody is a patient on the planet, even if you're a doctor. We've all experienced the wait time. When you go to a physical location and you can see other people waiting around, you could go to the front desk and see what the wait time is. When you're in a telemedicine visit, you don't know who else is around you. Maybe that person at the front desk is accessible or they're not.

Patient Attendance Predictor: With the Patient Attendance Predictor, one may get a different notification, message, or call than others. It depends on how the AI perceives your possible action.

I’m honored that you would say that. I would also challenge those others that are perhaps using their own technical language to describe something to ask themselves, “Are you able to communicate this to somebody who’s not from your field?” I could use the learning and development language and the academia of it but it’s pretty useless in this conversation.

Since this is labeled an AI show, let’s jump into a little bit of the technology of it. How does the technology serve your user clients and your business as a result?

At a high level, Edify is a platform that allows an engineering manager or an engineering leader to develop their own onboarding plan and to build it. We lead them through a set of best practices on doing that without interacting with another person, although we have services to help if they need it. Once they’ve developed that, we deliver that information to a new hire via Slack. We have a bot called eddy. Eddy has a little bit of a personality based on a couple of different factors around that new hire when they’re starting like what time zone and things like that.

Eddy is delivering certain amounts of information and types of information every day based on that new hire’s needs. Let’s say that the new hire is you, Rich. You’d get a message from Eddy saying, “Rich, welcome. It’s your first day. Here are the things that you need to do,” and it lists out your tasks. At the end of the day, eddy would come back to you and say, “Were you able to get these things done?” It can then lead you through a process of getting help or delaying those tasks if you need more time.

Eddy has some intelligence to be able to know who can it connect you with to help you solve certain problems. All of this is going to continue for your first 30 days. You then transition to being able to use the product differently as a team member, search and find information, and coordinate better with your team members.

To a degree, can I call eddy an AI bot?

You may technically call it that. It’s definitely a bot. It lives inside of Slack for a new hire’s experience and for a manager. It has got a web dashboard that they log into, which is so interesting. Speaking of human behavior, when we set out to build this prototype initially, our prototype first off was a human as a bot. I didn’t have any money, so I was trading with my Director of Product. She was running her own business and I was helping her with that, so she was helping me with product management.

She acted like a bot in a customer’s Slack environment. The poor thing was running things on Belfast in New York time for several weeks to just prove out that people would use “a bot.” In that process, we learned that the managers did not like using Slack. Slack was something that they had to use for their job because their teams wanted to do it. They didn’t love it but new hires love it. Individual contributors like it. We adapted to that with our product design.

If you want to be a good team member, you need to understand how your teammates like getting feedback or talking to you.

You heard me use the word introverts earlier, describing some of the population that you aim at. I’ve got this long-term belief that, for example, in the diversity and inclusion movement, one of the most underserved populations in tech if you’re looking for a bias, the bias is against highly technical employees who are introverts by nature. People make assumptions about them regarding the fact that their being introverted implies X, Y and Z. Nine times out of ten, or 99 times out of 100, those are not true. Yet we do know diversity or inclusion work around the introverted members of technology. How do you sense the ability and the capability of people who are introverted, who are joining into a new software engineering group to be able to work with your environment and with your toolset?

I would put this under the umbrella of accessibility, the idea of whether or not your company, technology or building is accessible to people. Neurodiversity is something that we are pretty uncomfortable talking about. It’s much easier for us as a workforce to talk about gender or race. While those things do need to be discussed and it’s certainly valid to have questions about power structures around that, especially when we think about hiring and training, we get overly comfortable with the idea that we’re going to get certain kinds of feedback.

We're going to understand that people enjoy training or that they interact with us a certain way. The reality is that when you think about a neurodiverse spectrum, people who may just be introverted or on the Asperger’s or autism spectrum may not give you the feedback or behave in a conversation the same way that you might expect somebody who is not on that spectrum to behave.

For example, this can have impacts on the eye contact that people make, how exhausted they might be from video calls or in-person meetings. They might have a hard time reading body language and the way that we interact. They may prefer to take in information differently. They might want to act on information differently. I was having a conversation with somebody about the complexity of something which sounds very simple, and in pop culture has become very simple or simplified. Unfortunately, it isn’t that simple, which is the concept of learning styles.

The research on learning styles is that there is evidence to suggest that there are different modalities where somebody might prefer to go out, physically. Let’s take learning a sport like basketball, for example. Somebody might prefer to get out there with no instruction and start trying to learn how to dribble. Some people might prefer to sit down, watch a video, listen and watch somebody else do it. Those are just two examples of many types of learning styles.

The interesting part of that is that there’s not much research to suggest that outcomes are very different if you have somebody learn in a style that is not their preferred style. What changes is their motivation to learn and to be engaged. That activity is different. What we leave behind, which we ought not to do, is the idea that people and communication are complex. We cannot simplify it even with an AI tool. With eddy, our product, we’re even trying to think about how do we ask learners, how they might want to receive information, interact and deliver feedback.

Something simple that we do in our company in our team now is something we jokingly called the user manual. When you buy a new appliance, you get the manual. We write our own manuals. For example, “I like to receive feedback in this way. I prefer getting praise, not in public or I do prefer getting praise in public.” It gets more complicated than that, but it allows us all to understand how we want to interact with the world and how we want others to interact with us.

Now there’s no excuse. Even if you don’t spend a lot of time with this person on the other side of the business, you can go read their user manual if you want to be a good team member. You can understand how they’d like to get some feedback, talk to you or be talked with. Hopefully, that’s an interesting or helpful answer to that question.

Patient Attendance Predictor: Practices that see high no-shows and late/cancels rates would look for other solutions that actually meant offers.

A lot of these things we can integrate back into the practice management system and create that automation for the staff, they can focus more on care and less on maybe calling patients to remind them or doing other tasks that the AI could communicate intelligently like, “This person is not coming in. Let's get them off the schedule,” and then we can get them off the schedule automatically. There's going to be a lot of value that technology will bring to help people get the care they need faster. We're hoping to lead that initiative forward.

One thing I could be sure of, Matt, and as you can guess we can talk to a lot of different companies about AI. Sometimes, the discussions get a little esoteric because of the nature of the application of AI in all kinds of industries. This is a conversation that our readers, as a whole, there's not a single one that won't be able to identify with it because they can identify with both sides of the equation, what the problems are, how you're trying to solve that and how you're using technology to do that. That's going to make for a great story and one that'll be memorable and helpful in our series of shows. I appreciate you coming on to share that with us. It's meaningful. It says a lot about you and what your team is trying to do.

We appreciate that. We're all going to be a patient maybe a few times or maybe more. Even doctors are going to be patients. What drives and motivates the team is that healthcare is an area where we can make an impact. We want to use AI and other technologies to help improve things for people. Thanks for inviting us. It was awesome to be on the show.

This has been another spectacular and thoughtful discussion on the show. This is the show that specializes in nothing but AI and is all about AI. Thanks for reading. Matt, thanks for joining us. We'll be on to the next episode. Have a great day.

Important Links:

About Matt McBride

Matt McBride is a digital thought leader with deep expertise in cybersecurity, digital transformation, and in building nimble technology teams using agile and scrum. He has led multiple technology transformations from the ground up in organizations such as Raytheon, Bank of America, Tyco, and ADT Security.

He comes to Modo after serving four years as Executive Vice President at Genesis10. In this role, Matt regularly consulted with executives and boards of Fortune 500 companies to evaluate and optimize their technology stack, and to architect tailored enterprise solutions to drive cybersecurity, agile, and digital maturity.

He previously served as the Chief Information Officer for 1st Global Resources, a nationwide broker dealer and investment advisory firm. Matt is also an Adjunct Professor teaching graduate cybersecurity, leadership, architecture, and software design courses for Southern Methodist University in Dallas. He is a regular contributor on cybersecurity, agile, and digital transformation topics for CIO.COM, InformationWeek, and the Enterprisers Project.

He speaks on these topics nationwide and is also the author of Leadership Patterns for Software and Technology Professionals (ISBN 1508634408) which is in wide release worldwide, and available from sources such as Amazon.

Previous
Previous

How To Onboard Software Engineers The Easy Way With Kristen Buchanan

Next
Next

Is AI The Real Deal? Using Artificial Intelligence To Optimize Sales With Patrick Moorhead