Accuray is an established player in the radiosurgery world, having gone from disruptive newcomer to serious concern, but Thomson found himself drawn back to the startup game.
"I think Accuray is a great company, it’s a great team, but for me, personally, the disruptive element was diminishing and it was becoming more of an operational type of play," Thomson told MassDevice.com in an in-depth interview at the headquarters of AliveCor, where he took over as interim-ish CEO late last year. "I’m most energized by thinking about disruptive plays in healthcare and how to make big changes and big differences. I think, to a large extent, that period for Accuray has passed."
Thomson resigned from Accuray October 2012 and joined Khosla Ventures, where he played guidance counselor to a handful of different companies. One of those was AliveCor, maker of the FDA-cleared smartphone ECG reader. He took the corner office at AliveCor in August and now spends several days each week in San Francisco’s South Bay to be near the company’s headquarters.
Aside from the energy of the city, which he likened to growing up in London, Thomson was drawn to the novelty and agility of the digital health scene. AliveCor has faced FDA hurdles the same way many other device makers have, but the company lives in a space between traditional medical technology and mobile consumer electronics.
Everything from AliveCor’s website to its headquarters is unmistakably colored in shades of the San Francisco startup scene. AliveCor’s office, unlike the stout and neutral-toned office-park digs of Accuray, is a colorful space shared with a couple of other local startups, sitting on top of a trendy retail housewares shop in the heart of the city’s financial district. AliveCor’s corporate bios include as much about favorite hobbies as they do about professional backgrounds. Chief medical officer Dr. Dave Albert’s bio, rather than drawing attention to his role as inventor of the mobile ECG, says simply that he is "just trying to fulfill his mission of saving lives 1 invention at a time… 36 patents and counting."
"There is a lot of energy in San Francisco itself. I’ve noticed it working with other startups in San Francisco in the same field," Thomson observed. "There’s just a feeling of critical mass and the feeling of pioneering."
Thomson plans to steer AliveCor toward new frontiers in data-driven medicine, turning the device-seller into a data aggregator and analyzer that can provide a nuanced look into the habits of the heart.
"Even though this is essentially a device company, the added value that we can bring is almost certainly through the massive data flow that we’ve developed and the insights that that can lead us to," he said. "This is a field that has to happen. It has to develop and it has to happen now because we have access to data today that we never dreamed we can have access to even 5 years ago."
In an in-depth interview, edited below for clarity, Thomson told us how AliveCor will evolve, why he chases disruption and how to turn big data into meaningful insight.
MassDevice: Why did you leave Accuray and take on digital health with AliveCor?
Euan Thomson: Actually, I have 2 roles. My primary job is I work at Khosla Ventures, the venture capital firm.
I moved from Accuray into the digital health field to do something really disruptive. The field of digital health and analytics is something that I feel will change medicine completely over the next 10 years.
AliveCor is 1 of the companies that I look after. I’m working with about 8 to 10 companies in the digital health field. I joined the board of AliveCor in February of last year. They were just going through a CEO transition and I was interim CEO for about a month, just to help figure things out and start looking at the strategy. I then handed it over to somebody else. For various reasons, that person in that position didn’t last as long as we thought he might. So I agreed to step in again back in August of last year.
I’m here right now as interim CEO of AliveCor, but it’s a longer-term, more permanent type of interim than classic board-type interims. My particular role here is to formulate strategy for the company, think about how this product should be positioned in the market. Then once that’s established and we’re on a good track, at that point start thinking about recruiting a CEO.
MassDevice: Everything from the website to the office-space looks are very different at AliveCor. Tell me a little bit about some of the cultural differences you saw as you changed roles.
Euan Thomson: When I started at Accuray, it was a similar type of environment to this, the early-stage feel of a company, which I find really exciting and interesting. Things tend to move very fast here, on a relative scale. Things are moving very quickly when companies are smaller. You’re solving problems in a very hands-on way. Everybody is multitasking and involved, and it makes for a very exciting environment. I would liken AliveCor now to the way that Accuray was when I first started in Accuray, and, of course, that was a big success. Putting AliveCor on the track to do the same thing would also be very exciting and rewarding.
MassDevice: Does it make a difference doing business in a place like San Francisco?
Euan Thomson: When I think of the last 10 years or so, the environment has changed. The Silicon Valley startup has become the San Francisco startup. I think the culture is pretty much the same. If you look at the nature of the startups in San Francisco, they are what people imagine it to be. They have a Silicon Valley profile, and I think it’s just an extended Silicon Valley.
MassDevice: There are a lot of difference between Accuray and AliveCor. How did you have to shift your mindset when you started leading AliveCor?
Euan Thomson: It’s certainly a new approach, but I think the common thread there is disruption.
When we started at Accuray the premise was that we had this big robotic device that could destroy brain tumors. The field of destroying brain tumors non-invasively was reasonably well-established. It was quite new but reasonably well-established. The vision for Accuray at the time was whether we could extend that to treating tumors throughout the whole body. That involved massive changes politically, economically, the way in which the money flows through, which doctors are involved.
The whole dynamic was changing. Anywhere in medicine, there’s resistance to that change.
I think we undertook something extremely disruptive and managed to make it real, and, in the process, literally saved thousands of lives and changed the course of cancer treatment for many groups of patients.
For the last year that I was at Accuray I was starting to think about what I wanted to do next. I think Accuray is a great company, it’s a great team, but for me, personally, the disruptive element was diminishing and it was becoming more of an operational type of play. I know that I’m most energized by thinking about disruptive plays in healthcare and how to make big changes and big differences. I think to a large extent, that period for Accuray has passed. It’s much more about operations and good implementation of what’s been established.
That was my thought process over my last year at Accuray. I met, over that time, with a couple of companies and looked at the business plans, and I started to look at the role that data was playing in the way in which their businesses were evolving.
Transitioning now back to AliveCor and why AliveCor is so important in that type of strategy, we have a device which is able to measure ECGs in patients. It differs from conventional ECGs in that it’s attached to a mobile phone, which means it can be owned and carried around by patients. In addition, it has the ability to communicate an ECG. It could send it to the doctor, it can send it to a database, it can be sent into a computer so that it can be processed. Those ECGs are reported in very many different environments.
There are other companies, for examples, that have very large databases of ECGS. For the most part, they’ve been recorded with the patients sitting in a chair, in a hospital, or a clinic in the middle of the day. The ECGs in our database are recorded with patients under many different circumstances. They might have just woken up in the morning. They might have just exercised, they might have just drunk coffee, they might have just drunk alcohol – things which could impact the heart. Our job is to record the circumstances in the context of ECG, and, in a way, we can develop much deeper insights into the way the heart functions than you could if you take an ECG on a patient every 3 months sitting in a hospital.
The device is very important to that. Having such a high-quality ECG recorded, as we have, gives us very high-quality input data.
By mid- to late-summer this year, we will have approximately a million ECGs in our database recorded under many different circumstances. Today, we’re recording about 2,500 ECGS a day, some of them recorded by doctors, but most of them recorded by patients themselves. We’re able to log those in the database. As this database grows, we can start to transition as a company towards performing certain analytics on those ECGs, giving feedback to patients and feedback to doctors and develop insights that nobody has previously developed.
The principle is that even though this is essentially a device company, the added value that we can bring is almost certainly through the massive data flow that we’ve developed and the insights that that can lead us to.
MassDevice: You’re a device company that’s becoming a data company.
Euan Thomson: Correct.
MassDevice: Tell me more about how you hope to use the aggregated patient data.
Euan Thomson: There are 2 phases to building this type of company, and there have been almost no complete company stories in the field of digital and data-driven healthcare. There have been very few companies that have started, gone all the way through to a large exit, for example. There’s no real playbook for how to build a company as there is in the hardcore medical device industry.
In order to generate a data-flow, you need to be providing something useful to patients and doctors today so that they use it. If you don’t do that, then you don’t get the data in place, you don’t have any long-term big data in order to carry out these advanced analytics.
The first phase is you get a high-quality device. That’s so people want to use it and also the quality of data coming in is good and you can actually analyze it. Then you have to put it into a format in which people want to use it. If I take an example, probably the largest group of patients that we’re looking after today or have been looking after today have atrial fibrillation. That’s a sporadic type of symptom that occurs in these patients. It doesn’t occur all the time, every day. It’s not regular.
What we found is that many patients buy our device to figure out when their atrial fibrillation is occurring and what prompts it. We have examples of patients who have figured that out. The doctors gave them a whole range of things, behaviors that they should and shouldn’t do – lose weight, check the blood pressure, avoid alcohol, avoid caffeine, be active. These are really designed to make the heart healthy, but some of them will have a direct impact on atrial fibrillation.
They buy the device to monitor how frequently their afib is occurring, and to figure out when it occurs. For example, we do have patients who have figured out for themselves that drinking coffee does not impact the afib, whereas, drinking alcohol does impact the afib. That has reintroduced coffee into their lives, which is to them is a big deal. Many of our patients are actually trying to track their disease to monitor its progress and to work out what might impact it.
Having identified that is a very important thing. We can then start to build services around those patients and the doctors who are managing those patients. We can, for example, make it easy for the doctors themselves to review their ECGs so that they can confirm, yes, this was afib or it wasn’t, in a timelier manner.
We introduced in November of last year the over-read service which, historically, is a service provided to doctors to help them analyze ECGs. What we did is we’ve made it available to patients so that they can access an over-read service.
It takes advantage of the fact the ECG is on a mobile phone. They record the ECG. If they were experiencing some symptoms, for example, at the time and they’re interested to know whether that was atrial fibrillation, they can send it off to a technician service or to a board certified cardiologist and get the ECG interpretation which will then come back to their cellphone. That helps them with medical-grade interpretation.
MassDevice: And this is the start of AliveCor becoming a data company?
Euan Thomson: Correct. That happens over time. Our plan, in parallel to launching for primary care physicians to call in screening services for patients on their biannual physicals, is to focus very much on a clinical indication such as atrial fibrillation and see it in a new way – see it from the doctor’s perspective, see it from the patient’s perspective, know where the gaps are today.
For example, there is a gap in that doctors and patients really have no idea how frequently patients are experiencing atrial fibrillation symptoms. Sometimes atrial fibrillation can be asymptomatic. Sometimes they can have symptoms but it’s not really atrial fibrillation. This type of device and this type of service can help them to demystify that and ultimately provide better quality treatment and patient management.
MassDevice: What happens next after you’ve got all that data?
Euan Thomson: We have no plans to sell patient data, to use it for marketing purposes. We have a very academic view of this data. Any way that we use it, it will be de-personalized. We’re likely to be looking for clusters of patients with certain symptoms, with certain characteristics and we’re looking for insights that we can gather from studying those clusters of patients.
MassDevice: And how will you turn that data into insight?
Euan Thomson: I’ll give you an example. These are theoretical because we are still in the early stages of this. Right now we actually have an algorithm – it’s not FDA cleared, so it’s not deployed on the app – but we have an algorithm for identifying atrial fibrillation in an ECG. It can automatically detect atrial fibrillation. There are some other algorithms of this sort of around. We believe ours is extremely good. One of our plans is to actually take it through the FDA process and get it used by doctors and their patients.
That’s representative of the state of the art today – you can identify atrial fibrillation. What nobody has been able to do up until now is to analyze whether the nature of the atrial fibrillation or the frequency of the atrial fibrillation can be related to certain circumstances, or does it represent a natural progression of this particular illness? Do the characteristics of the atrial fibrillation and that change over time correspond to certain changes in the patient that should be monitored or should impact the type of treatment that they get? As we continue to develop this stream of data, what we will be able to do is to investigate and experiment with that type of concept – can we predict atrial fibrillation? Can we even, for example, in a patient whose ECG or a person or a consumer whose ECG is apparently normal, if we monitor that over a period of time can we spot characteristics that nobody has been able to see before and predict that atrial fibrillation is imminent and be more proactive in the ways in which they are treated and merged? Those are the long-term gains for a company of this sort, developing new insights from very large quantities of data.
To transition back to what I mentioned in the beginning, about existing medical practice, if doctors are managing 100 patients a year or 500 patients a year, we could be monitoring a million patients a year with data that comes from different circumstances. Our ability to develop insights is way larger than the ability of 1 individual practicing physician. Many physicians are realizing this today. For the most part, the doctors that I’m talking to are very much behind companies that have got this big data type of approach.
Ultimately, it could change the way in which doctors manage if computers know more about how to manage a disease than individual practitioners. These companies become a tool for the doctor to use. In many cases, they may also become a tool for the individual patient use.
MassDevice: What does it mean to have 1 million ECGs? What comes next?
Euan Thomson: It’s more of a significant milestone than it is an enabler. If you have 10,000 ECGs there’s not much you can do with them; 100,000 there are some things you can start doing; when you get to half a million, which we got to in a couple of months back, this really starts to become meaningful data. When you get to a million and your data-flow is very strong, as it is now, you can really start to carry out some experiments.
The data-flow is as important as the size of the database, because there may be things which you want to experiment which you didn’t capture in the early days. It may be 1 important premise to figure out, for example, what level of activity is necessary to make a strong heart. If you’re not capturing activity level, you can’t experiment with it. Having an engaging product that generates a strong data-flow means that you can introduce variables and experiment with new variables and answer certain questions.
MassDevice: You’re at an interesting junction between the medical device world, which has very long product timelines, and Silicon Valley technology, which changes every few months. How do you negotiate that tension?
Euan Thomson: I don’t think we can ever lose sight of the fact that this is a new and emerging field, but that’s where the excitement comes from. Ultimately, we still have to remember that we are a medical company that needs to approach things in a very responsible way. I think those 2 things can sit well together; like at Accuray, for example. There has always been disruption inside medicine and that’s where some of the biggest breakthroughs have come from.
In this particular case, this is a field that has to happen. It has to develop and it has to happen now because we have access to data today that we never dreamed we can have access to even 5 years ago. The expansion of electronic medical record system, the expansion of utilization of smartphones into all layers of the population means that we can gather information today that we just couldn’t have had access to even a short number of years ago.
MassDevice: Do you live in the city or you commute in?
Euan Thomson: Actually I live in the South Bay, because this is a serious startup and serious startups need engaged CEOs. I’m staying up in the city 3 to 4 nights a week.
MassDevice: What’s your favorite thing about being around San Francisco?
Euan Thomson: I like the energy. I grew up in London and I love the energy of the city. Having done the last 10 years working in various office parks in Silicon Valley, something just as simple as being able to step out the door and walk up the sidewalk here, which is what I grew up with in London, is different. It’s very different for me to get into a car to go and buy a lunch and getting a car and driving back. I like energy of the city. I like the mood of the city.
There is a lot of energy in San Francisco itself. I’ve noticed it working with other startups in San Francisco in the same field. There’s just a feeling of critical mass and the feeling of pioneering, it’s a very engaging field right now. San Francisco is really at the heart.
At DeviceTalks Boston, Tyler Shultz will give attendees an inside look at Theranos and how he was able to sound the alarm after he realized the company was falling apart. Shultz will take attendees behind the story that everyone is talking about: the rise and fall of Elizabeth Holmes and her diagnostic company, Theranos.
Join Shultz and 1,000+ medical device professionals at the 8th annual DeviceTalks Boston.