[00:01:15] Aleksandra Zuraw : Welcome, everyone, to the podcast. Today, my guest is Dan Lambert, and Dan Lambert is the CEO of PathologyWatch. PathologyWatch is a little bit of a different company that I usually have guests on the podcast, because it’s a CLIA lab. Welcome, Dan. How are you today?
[00:01:34] Dan Lambert: Good. Thanks, Aleksandra, for having me.
[00:01:36] Aleksandra : And this CLIA lab, pathology lab, is providing pathology services for dermatologists incorporating digital pathologies. So talking about that, but let’s start with you, Dan. How are you today? And tell us about yourself.
[00:01:51] Dan: I’m doing great. I’m a computer engineer by training and then went to business school and graduated 10 years ago. I’ve been doing entrepreneurship for the last 10, 15 years, so this is what I do. I start companies. And PathologyWatch, I started this company about five years ago. We’ve raised about 40 million in venture funding and we’re growing really fast right now, so it’s a exciting time to be leading a company.
[00:02:11] Aleksandra : So you started this lab five years ago. Did you have any medicine, pathology experience? Why did you decide to go into pathology?
[00:02:21] Dan: Yeah, so my last company was actually in medical education. We did a lot of board prep and continuing education. And as I was going through that experience, we saw a lot of other fields like radiology and ophthalmology moving forward with AI, but pathology tended to a little bit slower, and also the industry needed to go digital. And so we thought that we could provide a solution that took over the laboratory process and the digital viewer, and then eventually, we’ll have AI integrated as well. But we wanted to be the first end-to-end solution in pathology and thought we had a real opportunity to break in here.
[00:02:58] Aleksandra : I think this is unique, because you are the lab, you’re providing the lab services, but digital pathology’s incorporated there. So what is your mission at PathologyWatch?
[00:03:10] Dan: Yeah, before we started the company, we sat down and interviewed about 10 different dermatology clinics, and they said over and over again, we do not want just a piece of the solution. We do not want just the lab. We do not want just a viewer. We do not want just an AI. We want a system that integrates end to end, where you pick up the sample for us, you handle the generation of the slides, the courier system, the insurance billing, all the necessary laboratory processes, and then provide the image back to us through a digital viewer, for viewing or for kind of a secondary opinion.
[00:03:44] And they all said the same kind of things over and over again. You need to solve all those problems, but you also need to make sure that you’re integrated with our EMR and that we’re getting the results back, and also that you’re saving our office a lot of time that they’re having to deal with on the pathology side. And so we took that feedback and designed a true end-to-end solution that was built around the dermatologist. And I think this is the first time that this has existed.
[00:04:06] Aleksandra : Yeah, I’m very enthusiastic about taking pathology into a different medical group, to dermatologists, but they are, how do I say, how they are connected? They’re medically connected to the pathologists, but I have never seen digital pathology as a service offering for other medical specialties where this is not only the report and the digital pathology is supporting the path pathologist doing the work, but here, also, the pathology, digital pathology, is supporting the dermatologist. So ,this is really fascinating. Like you said, it’s the first company on the market that’s doing this. I have not heard about anyone doing this either. So tell me about your services. Elaborate a little bit on what is this digital pathology in your lab, and how does this empower or give value to dermatologists?
[00:05:00] Dan: Yeah, absolutely. The way that the process works is that the dermatologist does the biopsy like usual. We pick up their sample, either via courier or FedEx. They send to one of our five lab locations across the country. And we do the normal lab processing, the grossing and the stain. And then it starts going a little bit differently, where we use our own laboratory information system to track the information of everything that happens inside the lab. We also provide the scanning service, so we have scanners in all of our laboratories. We digitize the image, and then we make that image available via web access, both the pathologist, as well as the individual dermatologist can see the image. And that opens up a whole new realm of possibilities, where the dermatologist can communicate with the derm-path. They can talk about the same case.
[00:05:52] For the first time, the dermatologist is not only getting just a fax report over. They’re actually getting to see the case and see the area that the pathologist has highlighted. And now, also for the first time, a lot of our dermatologists are showing the case to the patient, showing why there’s going to be a procedure. And there are a number of benefits to going digital. I would say the first one is that there can potentially be financial upside here, in that if the dermatologist is reading the case, it may increase the complexity of the visit, so there’s some potential there. I think the second compelling factor for dermatologists is that because our system is integrated with their EMR, they’re typically saving, on average, about 25 hours per month, per clinic, not having to deal with organizing patient cases.
[00:06:40] And they really value that time savings. We have a couple of case studies that highlight exactly how much time we save in the office. And then number three is simply better patient care. You’re less likely to have errors if the dermatologist can see the image and they can make sure it’s the right patient and the right case, the pathologist can look at it as well. And the reason that we started with dermatology is because dermatologists are actually board certified to look at these cases and provide opinions, and so it makes sense that digital is more compelling for them than any other area of medicine right now. And so we’re really focused on derm-path, but then eventually, we hope to get into the other verticals like GI and prostate.
[00:07:20] Aleksandra: So your current customers are dermatologists. You said you interviewed many dermatologists before embarking on the journey, but did they know about digital pathology? Did they want it, or did you have to educate them? How was that process? Because you are the first provider like that, so I assume they didn’t know they could ask for something like that, or tell me about that a little bit more.
[00:07:48] Dan: Yeah. That’s a really good point. So the very first interviews that we did, they knew that the digital pathology existed generally, or that the digitization of slides existed. Many of them actually, especially if they’re younger, they experienced digital pathology during their training program. So this was not completely new, but they didn’t know that it was actually available in clinical medicine. And so, as we’ve grown to sell more, we’re in about 65 dermatology clinics now, sometimes we do actually have to educate that this service is available and that they can pull up this image on their tablet for the very first time. And that’s actually a big deal in a field that doesn’t change very often, like pathology. So in general, I’d say they’re really excited. They also get super excited out the EMR integration that we provide, too.
[00:08:35] Aleksandra: Yeah, that would be a gain of its own. So even if they wouldn’t see the image, that would already be something of value for them. But now, seeing the image is also something additional. And like you say, they can show this to the patient, which is another level, I don’t know, improved patient care and educating your patients as well, because this is their data and that’s their tissue.
[00:08:59] Dan: Absolutely. Yeah. As a patient myself, I have to get checked every six months because I’ve had some skin conditions in the past, and being able to see the image of your case, it’s probably the first time that patients got to see. It’s a new experience, but one that I think that trend continue going.
[00:09:15] Aleksandra: And my husband is a physician and I have experienced that the U.S. patients are very educated. He’s working for a urology lab, and they are discussing cases with them on a very high level. Seeing that for the first time is something new for the patient, but you are the person that comes every six months, and after several visits, you will be able to recognize that this image looks different. So, that’s another level. That’s fantastic. So what platform does your service run on?
[00:09:49] Dan: So our platform is built entirely in-house. So we built the LIS and we built the digital viewer, and we use those resources internally. What we do integrate with, though, so we’re integrated with Modernizing Medicine, EZDERM, and eClinicalWorks, which make up together approximately 90% of the EMR space in dermatology. In almost all cases, we’ve been able to support clients that want that data pushed in. And why that really matters is that since we push the report into the EMR system, the dermatologist can easily click on a link inside of the EMR report. It automatically launches our viewer, automatically authenticates them, and allows the physician to very easily navigate between the EMR and our viewer. And that kind of integrated seamless experience means a much faster interaction with systems. They’re not having to juggle multiple independent systems like an LIS, for example, like they have in the past. A true end-to-end software integration. We are the first one that we know of in this industry.
[00:10:50] Aleksandra: So basically, they are doing their workflow, reading their reports as they were in the past, but they can click on links, and it takes them immediately to your viewer or whatever other information they want to know from this report. They don’t need to switch software. They don’t need to switch the computer or anything like that?
[00:11:11] Dan: That’s correct. Yeah, same device, automated authentication. It’s really a seamless experience. And our dermatologists tend to really rave about this, because that ability to quickly link out to a case has never existed before.
[00:11:26] Aleksandra : It’s so intuitive in the normal PC experience, but in the medical world, where half of the records are still written down, and I have experience with my kids that I have to send a letter to the provider so that they can fax or send the paper records. Yeah, this is a big advancement.
[00:11:50] Dan: Yeah. Our tagline is software eats pathology. So historically, this has been an industry that is mostly microscopes and fax machines, and we are replacing that entire process with software, which probably should have happened 20 years ago. But here we are, and trying to fix it now as best as we can. And the difference is truly astounding.
[00:12:09] Aleksandra: So this unique value proposition already differentiates you from the competition, but when you want to stay ahead of competition, what do you do? How do you keep differentiating and keep innovating?
[00:12:24] Dan: Yeah, I think at this point, in terms of technological integration, the companies in a few years ahead, we haven’t seen any others be able to provide this kind of offering yet it. But where we need to go, which should be not a surprise, is that we’re deep into research on the AI front of trying to figure out how do you detect different tumors and skin conditions, and how do you make it really accurate so that it can be a true assistive device, or an assistive tool, for both dermatologists and dermatopathologists? And that takes a lot of time, and it takes a lot of data, and we’re still in the process of aggregating a lot of data and organizing that data so that we can build algorithms in the future.
[00:13:09] Aleksandra : So you mentioned, you are an end-to-end solution. So you have the wet lab and everything up to the reporting. The pathologists are working within your network, right?
[00:13:21] Dan: Yes. Correct. We have the labs and the insurance contracts. We have the software that manages the laboratory, the digital viewer that the dermatologist can view on, and we have all the integrations with the major EMRs in dermatology right now.
[00:13:35] Aleksandra: So you have the ideal conditions to have homogenous slide data sets for developing your AI algorithms. Is that intended, or do you want to keep it like that? Do you want to, I don’t know, get data, get images from other sources? What is your approach to your AI model development?
[00:13:59] Dan: Yeah, that’s a really good question. And it touches on something is very tricky in the AI development space right now, which is that generalizability is a very hard problem. The ability for the AI to recognize cases that may have bad stains or that may be smudged, or that have other real world conditions, or techniques used in a laboratory that the AI is not as inherently, not very adaptable to real world conditions. And so you can either work on generalizability and include many data from many different labs, or you can go heavily supervised learning, and focus on the conditions that exist within your own footprint of laboratories. And so in our environment, we can very clearly ensure that the quality and the standardization of the techniques used to generate a slide are consistent, and that means ultimately better outcomes for the AI and, specifically, amazing specificity and sensitivity when you are using completely controlled conditions, and not dealing with really other bad conditions that can get introduced into the process.
[00:15:03] Aleksandra : So the AI algorithm development happens at the end of the pipeline, so to say. Is the whole pipeline optimized for AI or with AI in mind, including the people that are doing the previous steps and controlling the conditions? How do you go about this in the company?
[00:15:23] Dan: Yeah. Yeah, that’s correct. We wanted to, from the get go, before even using the AI, the important steps were to standardize all of those conditions across the laboratories so that the slides came out perfectly and could be used both for AI training, and then eventually for clinical use, or clinical interpretation by the AI. And that means some very tight controls. We’ve introduced some new staining techniques that aren’t used generally in other laboratories. We’ve also switched to fully automated equipment, and we’re not doing a lot of processes by hand that introduce variability. And that tight control of the process means true consistency. And then also a number of QA/QC processes that exist in our lab that do not exist in other laboratories, to make sure that those inputs are the best that they possibly can.
[00:16:09] Aleksandra : So you had that all planned. That was your value proposition from the beginning?
[00:16:15] Dan: Yeah. We planned to own the lab and be able to control that environment very well, in order to succeed in the long run.
[00:16:22] Aleksandra : And was there anything that failed? Did you have a failure that set you up for success? What didn’t go according to plan?
[00:16:30] Dan: Yeah, that’s a good question. In part, we needed to be able to detect the edges of the tissue so that we can eventually do things like measurements. And so we introduced this green stain that would highlight the edge of the tissue, and it took us a couple of months of trying a bunch of different colors to try to capture this special staining. And the ink would get on other parts of the tissue, other colors would get washed out. We had complications that the color could be the same as a tattoo ink, for example. And so working through some of those actual laboratory changes were very important, but now, we figured some of those things out along the way, but these are not one week or two week problems. Sometimes they take several months of trial and error before you find something interesting that works and something that really-
[00:17:17] Aleksandra: Did you find something that worked?
[00:17:19] Dan: Yeah. We finally got a staining technique that works consistently on probably about 98% of tissue. Eventually, you do get to a solution. I think also for those getting into digital pathology in real volume, we’re processing about 1,000 cases a day, you start experiencing things like what do you do when the scanner arm breaks or, yeah, an internet connection goes down and you need to upload these files? There are a lot of things that get introduced when you’re developing a completely new process of doing things. But I think we’ve managed to solve most of those problems, but it has taken a good couple of years to work through all of the real clinical workflow hangups. And also, it’s solving the problem of just massive data storage. We only get paid once for the case, but we need to hold that image for seven years. So moving quickly from providing production level images to rapidly moving a whole bunch of data to localized, long-term, low-cost storage, and being able to do that really quickly, is super, super important, and allows us to actually make digital pathology profitable.
[00:18:20] Aleksandra : That is a very good point, because making digital pathology profitable, it’s tricky, because you don’t immediately see the return on investment, and your value proposition incorporated that. I think the mindset that you start to endeavor with is very important. And then you can tackle those. Some of the problems you can tackle in advance because you are thinking already with this in mind. Some of them are going to come as you go. But you already have the problem-solving mindset to make it happen, to make it work, whereas in institutions that are switching from analog to digital, it’s not always possible. And I think this is one of big growth.
[00:19:09] Dan: And we have a lot more freedom to try things than I think other organizations do. And I think that’s why a startup has to be the one to solve this problem first. And just for those who are listening in on how we solve the cost problem, what we ended up doing is keeping the image, these massive 300 megabyte images, on high availability storage for just a period of two weeks, so that the dermatologist could view the case, and then rapidly moving, taking those images, getting them off of high availability storage, moving them to local boxes, but leaving behind the top layers of the case, because those are relatively small images. And so in case anyone wanted to come back and do a light reference later, they could see up to 10X zoom, or sometimes a little bit more, depending on the case. And then if they wanted to, there’s a button that they could request the case be moved automatically back into high availability storage. So it took a lot of creativity and again, six months of navigating this storage problem, to really solve all these-
[00:20:12] Aleksandra : So basically, you’re keeping part of the pyramid, part of the digital image, host light image pyramid, up to a certain zoom, and the rest is not available for the immediate access.
[00:20:26] Dan: Yeah. But it can still be requested. And that way, we can also keep the raw image file in storage for long term. And that allows us to simultaneously be compliant, allow easy viewing, and then allow request cycles. If you need to see down to 40X, we can enable that as well. It’s just then you have to wait 24 to 48 hours for us to restore that image.
[00:20:49] Aleksandra : This is interesting. I didn’t know you could do that. I didn’t know you could just save part of the pyramid, because whole slide images as are stored as a magnification pyramid. Could you do that also for just to have the high zoom, and not the low? That would not-
[00:21:07] Dan: The problem is, yeah, the high zoom is that bottom of the pyramid, where there’s just a million images.
[00:21:13] Aleksandra : Okay. So it doesn’t really help you.
[00:21:15] Dan: Yeah. Yeah. And actually, I think this is important for anyone to understand coming into digital pathologies, that in order to even display these, we have to take these images and actually break them up into about 100,000 tiles that are delivered to the browser in order to make it really fast, because pathologists, they are not going to spend more than a second on these things. They go really fast, and you need to have dynamic delivery of tiles based on the area that you’re zooming into. And so it works a little bit like Google Maps in that we’re delivering tiles around whatever focus point you’re on, and that way, if you’re scrolling around, we’re able to really load those images very quickly. And that actually took a lot of software development and advances to cull.
[00:21:58] Aleksandra : So I don’t think we emphasize that, so this is all web-based. Everybody can just access through the browser, right?
[00:22:05] Dan: Yeah. And of course, with significant layers of security, but yeah, it’s designed to be accessible anywhere. We do ship the slide to the pathologist as well, but the image is made available via the web browser to both the dermatologist and the dermatopathologist.
[00:22:19] Aleksandra : Okay. This is really interesting. Thanks for being my guest today. Is there anything else that I forgot to ask, or something that you would like to talk about that we didn’t cover yet?
[00:22:32] Dan: No. I think this was a great interview. Thanks for asking great questions, and appreciate you taking the time.
[00:22:36] Aleksandra : Thank you for being a great guest, and have a great day.
[00:22:40] Dan: You too. Bye.
