What can developers learn from a tech company that connects patients with medical specialists from around the world? During a recent episode of Dynamic Developer, I spoke with Julian Flannery, founder and CEO of Summus Global, about just that.
Summus is a “virtual specialist platform” that connects patients with doctors who provide personalized care. The service is not designed to replace the care provided by a patient’s primary care doctor, but provide virtual access to medical specialists who can offer “information and insights elated to preventive care, recent diagnoses, ongoing health management, surgery considerations and identifying the best courses of treatment.”
On Tuesday, the company announce it raised an additional $22 million in funding, in a strategic investment round led by Danaher Corporation’s Mitchell Rales and the Glenstone Foundation. In 2021, Summus raised $50 million in a Series B funding round.
When asked about how the company intends to use the additional funds, Flannery told me that the money will be used to invest in both growing their position within in the healthcare navigation and second-opinion category and further developing their proprietary platform around technologies like AI and data analytics. The platform of the future is multidimensional and will provide diverse ways for patients and doctors to interact, including peer-to-peer, direct-to-patient and multiple-party communication, Flannery said. “All our wonderful engineers are continuing to build and enhance our platform as we go forward.”
During our conversation, Flannery gave me a rundown on what the company does, the tech that powers its platform, and how its development process has changed over time. The following is a transcript of the interview, edited for readability.
Bill Detwiler: Before we talk about the tech behind Summus Global, for viewers and listeners who don’t have experience with the company, give us a rundown on what you offer, what you do.
Julian Flannery: Summus is a virtual care platform. We do focus on specialty care. So when you think about, sort of historically, virtual care companies were either primary care or urgent care. We focus where the complexity lives and the cost lives, which is specialized care. And that could be anything from allergies to weight management all the way up to cancer and ALS. So, we’ve really designed a platform that has attracted the best in medicine across the United States. Today we have a curated network across 50 top hospitals, over 4,000 leading specialists can access our platform. And then we use a sort of a proprietary marketplace model that allows us to deploy them within hours and days across any health question.
So, the value, we sell mostly to employers, large groups where people are facing health issues every single day. And we help them make better medical decisions within their plan. And for the employer we drive better outcomes, better cost-efficient outcomes for the employer. That’s really what we do. I would characterize us as kind of virtual care 2.0 where we’re sort of on the cutting edge of kind of where the market is going and in my opinion, and just an exciting time for virtual care, certainly, that it’s been driven in a lot of ways by COVID, but more importantly just the adoption and appetite both on the medical consumer side and the physician side. So, we’re excited about the future.
Bill Detwiler: I know that myself I’ve had experience with telemedicine, virtual care, whether it’s sort of chat with a nurse provided through your corporation as you were talking about your employer, or even my GP has experimented with, especially during COVID, doing video consultations. So, how do members of your service incorporate the care and advice they may get from the specialists that they connect with through Summus, with that, that they get from a primary care physician maybe that they’ve been seeing for years? How do they typically make that work?
Julian Flannery: Yeah. Well look, I mean, primary care doctors are an integral part of the healthcare system. The relationship with families, with members. Primary care is really about familiarity and convenience, right? But specialty care is a little bit different, right? It’s all about speed and access. How can I most quickly get to a high-quality doctor who can help me understand what to do with my condition, right? Can be allergies, can be sinusitis, could be cancer. Really depends. The way I like to talk about it is if you have high-quality primary care plus Summus, you have the best healthcare in the world, right?
Because we sit on top of the sort of specialized healthcare world, and really help you access really high-quality expertise, really high-quality doctors that can help you understand what treatment paths you need to go down, what tests you might need, and ultimately help you make much better decisions in healthcare. I think in terms of the integration, there’s a lot of complexity, as you know, in healthcare from the EHR to the primary care doctor, the specialist. That primary care can participate in that, that doctor can participate in the consultation, in the journey with you through Summus, right? There’s really no boundaries between the care continuum. We want to be sort of a holistic solution for our members, but we solve for that kind of complexity of specialized care and the speed to access those doctors.
Summus wants to help patients bridge their primary and specialist care
Bill Detwiler: Yeah, no, that’s something that has definitely been a frustration. Myself and family members. You mentioned allergies, and I think it’s really hard to find an allergy appointment even where I live, which is in Kentucky if folks can’t tell from my accent. And so having access to people outside the region to break down those geographic barriers, it can be really helpful. I like what you were talking about there with integrating the care with your primary care physician with this. How does that work? You talked about electronic health records, you talked about sharing information.
I know the times when I’ve had family members who needed kind of coordinated care, it was best when all the doctors were in the same room. So it was a cardiologist and it was their GP. And then it might be someone dealing with their kidneys or it might be whatever it is. So, how do you make that happen? Is it, their GP gets on Summus’ platform? Is it, they get reports? Is it the patient kind of sharing the information with their GP? Is it all of the above? How does that work? Because I think it’s fascinating.
Julian Flannery: It can be all the above. The reality is most patients don’t necessarily involve their primary care doctor in all the specialized care decisions, right? They do want them to know about it. They want them to be able to access the records, the data or the sort of the advice that’s been given to them in that platform. But again, the primary care doctor, if somebody feels very strongly about the relationship with their primary care doctor, they can participate in a consultation. They can hear everything that the specialist is saying. And then what Summus does is we actually tailor the journey to condition-specific elements, right? Allergies is one example. A knee surgery is another one.
It might be a little bit more transactional. It’s sort of like, “OK, am I going to get surgery on my knee or am I…” 45 years old, and it’s probably not worth it because it’s a partial tear, right? But cancer’s a whole lot different. Cardiology’s a whole lot different. To your question, you need, sort of, a group of people around that individual to help them make good decisions in that specialty. So Summus is really going to be the platform for that, right? We have a network of what we call Summus MDs who are internal medicine, primary care- and ER-trained doctors depending on the situation. They really serve as the access point to Summus.
They understand the journey, they’re with the member throughout the whole journey. And then we can drop in specialists at certain points in that journey to help people understand where they are, what they need to do, the forks in the road, so on and so forth. So, you can imagine a cancer journey being radically different than just knee surgery, right? Cancer can be over six months, 12 months. There’s chemo, there’s radiation, there’s clinical trials. So, you really need that kind of holistic perspective from multiple data points, multiple people. Summus can provide that.
Building the health platform of tomorrow, today
Bill Detwiler: Let’s talk a little bit about that “how” part that we were talking about, that tech. The tech behind how Summus connects patients with those specialists. Without divulging any kind of trade secrets about what’s happening behind the scenes, give us a rundown on the platform. What does it look like for your members, and maybe what does it look like for the provider as well?
Julian Flannery: Yeah, it’s a great question. We do use a marketplace model. If you think about the platform, it’s really a three-sided platform. One is the member side. It can also be the caregivers or anybody in that ecosystem. The other is the physician side. And then there’s the third which is the admin side, right? Where the team at Summus is working. There’s all of these, everything from open table calendaring or UBER-esque type scheduling and that kind of thing to records and DICOM (Digital Imaging and Communications in Medicine) imaging and helping people understand how to offload images off of a disc and upload them to our DICOM viewer, and all that kind of thing.
So, there’s some pretty cool tech. The interesting part about specialty care, it’s not like you and I here where it’s just we’re on Zoom, we’re sitting here together and it’s just a screen between us. Specialty care requires a lot more, right? The physician has to see the medical history. They have to be able to see the images. They have to be able to see the summary of any prior appointments that the patient has had. And then you also want to build a platform such that when the two parties come together, they’re both ready, right? You don’t want to have a place where there’s introductions, and you ask a bunch of questions, and it’s just a waste of both people’s time.
So, you really curate that interaction through tech and through a combination of integrating with different systems where the physician has reviewed the records, the physician understands the medical history. The physician is a specialist that we match to you using algorithms within the platform. And then the member is prepared with questions, right? So, they come into that interaction. You can be the smartest person in the world, but if you don’t know what to ask during a medical appointment, it’s not going to be as valuable. So one of the things we do on the administrative side is help people prepare questions.
What are the 10 questions that we need to run through for this particular appointment viewable to both sides? And then we enable single multi-part video from anywhere in the world. So oftentimes it could be you in Kentucky, your mom in California and the physician down in Texas. It really doesn’t matter. So, we do a lot of multi-part interactions. Caregivers, treating physicians, spouses, that kind of thing, and then the physicians, which we can also do sort of multi-panel type interactions as well depending on the condition. So that’s from a sort of a high-level perspective that that’s sort of the three dimensions of the platform.
Bill Detwiler: When you were building the platform, what was that development process like? I’m just curious because there are a lot of components there, right? We’re talking about video, we’re talking about being able to view records. We’re talking about being able to have a chat. And I know those things exist in a lot of platforms that we’re all familiar with such as the one we’re using to record this video right now, but bringing it all together in a holistic way, I imagine, was fairly challenging. And to do it in a way that fits a medical setting, right? Because you don’t necessarily want the same features in a platform you’re having a meeting with for say a performance evaluation or just a team meeting or a stand up, something like that, that you need in a meeting with a physician and especially specialized care. So, what was that development process like? How did you iterate? How did you pick those right features or how did they evolve over time?
Julian Flannery: No, it’s a great question. And by the way, it’s got to be HIPAA compliant and GDPR compliant, right? And it’s got to pass all the SOC2. And if you’re selling to employers, there’s an OPSEC review you have to go through, an InfoSec review you have to go through, which I’m sure all your viewers are familiar with. Look, I think to some extent, I used to be on the management team of a company that was the largest marketplace for expertise in the world, right? Did millions of transactions to the platform. So, I was somewhat familiar with how to sort of structure a marketplace. I mean, we truly are a marketplace for expertise and in sort of medical care, right? That has evolved to a virtual specialty care platform.
I think it started with that foundational base of really just thinking about all the minute details that go into curating interactions between people or multiple parties in a way that’s valuable for them, right? Everything from the onboarding process, the data you need to capture, how that’s presented to the other side of the market to the physician side, to the alerting mechanisms, right? So not everybody, every doctor has got the latest, fanciest, greatest software. So how do you make sure that there’s troubleshooting, that there’s automated things that can go into the platform, that if their Wi-Fi isn’t great, or they’re not running the latest version of iOS?
You learn through doing it so many times that there’s … it’s just you have to obsess over the small details of an interaction and then build automation into those interactions that makes it sort of infinitely scalable, if you will. So, I would say we started–and you just have to learn from every interaction. The doctor feedback, getting it to an MVP, to really sort of iterate from there, dealing with bad feedback. What happened? Was it technical? Was it service? Was it the physician side, the consumer side? How do you sort of do that? And then what pieces do you integrate into the platform? We use Braintree for payments all around the world. We use Twilio for single multi-party video. We use a bunch of other, sort of, plug-in services.
And then right now we’re thinking a lot about de-identified data and how do we leverage data to make it predictive and allow us to engage people in the right ways, and help them understand what they need to do in healthcare? So, a long winded way of saying it’s just a constant iteration process and it’s really obsessing over the details of how do you put people together to handle sort of complex issues and be informed and have a really good conversation around those issues, right? That’s kind of what we do.
Using algorithms to match each patient with the perfect specialist
Bill Detwiler: We live in an enterprise, everything is a service sort of world these days. Do you think it would’ve been possible to build a platform like Summus without all those parts that are sort of that you can bring together and use to create this platform? I’m not taking anything away from in-house dev teams, but I’m curious to hear your thoughts as someone who’s done this multiple times, that it wouldn’t be possible to do it, especially not as quickly, perhaps. Certainly you could build everything from the ground up. But because we now have all these services and APIs and data sources, and everything can be brought together. Did that allow Summus to develop its platform more quickly?
Julian Flannery: Totally. And not only that, but less expensively too, right? I mean, it’s not speed. It’s also … we wouldn’t know how to build a payments platform. We’re in the virtual care business, but we’re not in the sort of video, bandwidth business, right? That’s not what we do. AWS. I mean, what company is not using AWS these days? I mean, just a hugely economical way to scale software around the world. I would say it’s an absolutely true statement. I mean, we work with Okta to integrate with HR. I mean, there’s APIs and plugins that we use all over the place. But as a founder, you’re also always focused on building enterprise value, right? You don’t want to be purely a solution or you’re sort of stitching everything together and nothing’s really proprietary, right? So, the proprietary stuff is really around the complexity and the sort of, the algorithms tied to the search and the network and all that kind of stuff. But it all really, to your point, it comes together and is accelerated by incredible companies that are building sort of niche software products that can just sort of plug into your platform.
Bill Detwiler: And that’s the secret sauce, right? That’s the value-add for Summus, is those algorithms around matching you with the specialists. So I guess, again, we talked about that a little bit earlier, but is there without, again, sharing anything in trade secrets or in the IP, what is it that you were kind of looking for? Maybe in general terms, what is it that you’re using to kind of match a patient with a specialist? Is it a holistic kind of look at all the records? Is it geographic? Like I said, you don’t have to tell me the exact data points it’s using, but I think a lot of people who are really interested in machine learning and artificial intelligence and looking at the predictions and how they can produce especially better health outcomes are really just kind of curious about what’s happening in that space. So, if there’s anything you can share there, I’d love to hear it.
Julian Flannery: Yeah. I mean, so first of all, you have to gather the data on the physician. So, our 4,000 plus physicians you have to be able to onboard them in a very efficient way that allows you to capture data on them, right? And then there’s obviously publicly available data. And then you really build the algorithms around a combination of multiple variants, right? I mean, you have… Variants is probably not the right word. Variables. I have variants on the mind right now.
Bill Detwiler: Don’t we all?
Julian Flannery: Exactly, exactly. A combination of training of areas of expertise, of research, of outcomes in clinical data, importantly procedure and treatment volume, right? So if you’re going to get a knee surgery, you don’t want to go to somebody who’s done it twice in a year. You want to do somebody who does it 200 to 300 times a year, right? And has practiced at a high-quality institution and all that kind of stuff. And then you have to take into account geography and gender and history of present illness, and sort of the dynamics of acumen and whatnot.
Bill Detwiler: And is all that customizable? So from a member’s perspective, when you come in, can you make selections to say, “I would like a specialist to do this.” So, it’s taking the AI, but it’s also because healthcare is so personal that you’re layering on those personal needs or choices. Is that possible as well?
Julian Flannery: You are, but we’re not completely taking it out, right? So, if we go on Amazon and we search for a product, in healthcare you can’t afford to make mistakes, right? So you have to put a clinical layer, a human layer on top of the algorithms to make sure that you’re doing it in the right way. So I wouldn’t want you going into our platform and saying, “OK. I got this type of allergies. I don’t know if it’s a room or it’s an allergy spell. I don’t know what it is.” And then you end up talking with a doctor that’s completely the wrong person, right? So, we want to inject, or put a clinical layer on top of that, where our doctors are speaking to you, helping understand your history, what your diagnosis is because sometimes people who self-diagnose or they have a diagnosis, it’s not true.
Bill Detwiler: Oh, it’s doctor internet.
Julian Flannery: Exactly. So, I think we’re not in a place where it’s self-service yet because I think it’s specialty care. It’s a lot different. In primary care you go on Zocdoc. “OK. It’s close to me. OK. It’s got five stars. OK. I like where they went to school. OK. I’ll go try them out,” right? Specialty care is a lot different. It’s much more nuanced, much more niche. The subspecialties are so much more precise in terms of the different types of conditions. And so maybe we’ll get there one day, but the underlying technology to make our clinical decisions rapid and high quality is all built. We would never want to take out that clinical layer at this point in time.
Success with artificial intelligence comes from know when and where to include a human-touch
Bill Detwiler: I think that makes sense because I’ve talked to a lot of people about how we can successfully use AI right now. And some people talk about it as a replacement for that kind of human interaction. But most people that I’m talking to right now are talking more around augmentations, right? So it’s a way to blend the two together. And especially as you’re describing in clinical and specialty medical care, that’s definitely important.
So I guess my last question is really kind of to follow up on what you were talking about a little bit earlier on the lessons that you learned and how built the platform, which is what advice would you give to other organizations, other founders, other executives who are looking to kind of build similar platforms or working to sort of augment a current sort of human task with AI? What advice would you give folks?
Julian Flannery: I would say it really depends on the market, right? So, healthcare at its core is human, right? It is, you have to keep… You can do robotic surgery, you can do AI driven reads, radiology and whatnot, but it all is … Maybe the surgery is not human necessarily all the time, but you have to decide what level of human you want in your business to some extent, right? And then the software needs to be built in a way that respects that, right? I see a lot of companies out there, “Oh, we’re AI for healthcare. We take the human being completely out of it.” It makes me nervous a little bit because at the end of the day there are challenges with AI, there’s challenges with machine learning.
And if you completely take out that judgment or the context or the clinical perspective that comes with decades of experience for doctors who’ve trained very hard, it can be a little tricky. So, I would say just, the advice is just really think hard about building a platform that can scale how much human do you really want to take out, right? And how do you sort of build in tech to kind of augment those interactions in a way that’s powerful. And then I think secondarily is just I mean, this is no secret, but just constantly iterate, constantly get feedback. We’re iterating every single day, and have a development cycle and a team that is very responsive. And ultimately your software becomes just increasingly better and better and better, the more people are using it and all that kind of stuff.