Developing a wearable biosensor: A doctor's story
For this post, I caught up with Dr Brennan Spiegel, to hear in more detail about his journey to get a wearable biosensor from concept to clinic. In the interview, we discuss how an idea for a sensor was borne out of an unmet clinical need, how the sensor was prototyped, tested, and subjected to clinical research, and how it was finally FDA approved in December of 2015. Throughout, we learn about the challenges of developing a wearable biosensor, the importance of working with patients, doctors, and nurses to get it right, and how to conduct rigorous research to justify regulatory approval of a device. The interview ends with seven suggestions from Dr. Spiegel for other inventors seeking to develop wearable biosensors.
1. What is AbStats?
AbStats is wearable sensor that non-invasively measures your intestinal activity – it's like a gut speedometer. The sensor is disposable, about the size of a large coin, sticks on the external abdominal wall, and has a small microphone inside that dutifully listens to your bowel churn away as it digests food. A specialized computer analyzes the results and presents a value we call the "intestinal rate," which is like a new vital sign for the gut. We've all heard of the heart rate or respiratory rate; AbStats measures the intestinal rate. The sensor tells the patient and doctor how much the intestines are moving, measured in "events per minute." If the intestinal rate is very high, like 30 or 40 events per minute, then it means the gut is revved up and active. If it's very low, like 1 or 2 per minute, then it means the gut is asleep or, possibly, even dysfunctional depending on the clinical situation.
2. What existing problem(s) does it solve?
AbStats was specifically designed, from the start, to solve for a real problem we face in the clinical trenches.
We focused first on patients undergoing surgery. Almost everyone has at least temporary bowel paralysis after an operation. When your body undergoes an operation, whether on your intestines or on your toe (or anywhere in-between), it's under a great deal of stress and tends to shut down non-vital systems. The gastrointestinal (GI) tract is one of those systems – it can take a hit and shut down for a while. Normally, the GI system wakes up quickly. But in some cases the GI tract is slow to come back online. This is a condition we call postoperative ileus, or POI, which occurs in up to 25% of patients undergoing abdominal surgeries.
The issue is that it's hard to know when to confidently feed patients after surgery. Surgeons are under great pressure by administrators to feed their patients quickly and discharge them as soon as possible. But feeding too soon can cause serious problems, from nausea and vomiting, to aspiration, pneumonia, or even death. On the other hand, feeding too late can lead to infections, prolong length of stay, and cost money. As a whole, POI costs the US healthcare system around $1.5 billion because of uncertainties about whether and when to feed patients. It's a very practical and unglamorous problem – exactly the type of issue doctors, nurses, and patients care about.
Now, you might ask how we currently decide when to feed patients. Here's the state of the art: we ask patients if they've farted or not. We literally ask them, practically all day long, "have you passed gas yet?" No joke. Or, we'll look at their belly and determine if it looks overly distended. We might use our stethoscope to listen to the bowels for 15 seconds at a time, and then make a call about whether to feed. It's nonsense. Data reveals that we do a bad job of determining whether someone is fit to eat. We blow it in both directions – sometimes we overcall, and sometimes we under call. We figured, in this fantastical age of digital health, there had to be a better way than asking people about their flatus! So we invented AbStats.
3. What prompted you to embark upon this journey?
One day, about 4 years ago, I was watching Eric Topol give a TED talk about wearable biosensors and the "future of medicine." As I watched the video, I noticed that virtually very part of the human body had a corresponding wearable, from the heart, to the lungs, to the brain, and so forth. But, sitting there in the middle was this entire body cavity – the abdominal cavity – that had absolutely zero sensor solutions. As a gastroenterologist, I thought this must be an oversight. We have all manner of medieval devices to get inside the GI system, and I'm skilled at inserting those things to investigate GI problems. But typical procedures like colonoscopies, enteroscopies, capsule endoscopies, and motility catheters are all invasive, expensive, and carry risks. There had to be a way to non-inavsively monitor the digestive engine. So, I thought, what do we have available to us as doctors? That's easy: bowel sounds. We listen to bowel sounds all the time with a stethoscope, but it's highly inefficient and inaccurate. It makes no sense to sit there with a stethoscope for 20 minutes at a time, much less even 1 whole minute. But the GI system is not like the heart, where we can make accurate diagnoses in short order, over seconds of listening. The GI system is slow, plodding, and somewhat erratic. We needed something that can stand guard, vigilantly, and literally detect signal in the noise. That's when AbStats was borne. It was an idea in my head, and then, about 4 years later, became an FDA-approved device.
4. What was the journey like from initial idea to FDA approval?
When I first invented AbStats, I wasn't thinking about FDA approval. I knew virtually nothing about FDA approval of biomedical devices. I just wanted the thing built, as fast as possible, and rigorously tested in patients. As a research scientists and professor of medicine and public health, this is all I know. I need to see proof – evidence – that something works. AbStats would be no different.
I was on staff at UCLA Medical Center when I first invented the idea for AbStats. I told our office of intellectual property about the idea, and they suggested I speak with Professor William Kaiser at the UCLA Wireless Health Institute. So, I gave him a call.
Dr. Kaiser got his start working for General Motors, where he contributed to inventing the automotive cruise control system. Later, he went to work for the Jet Propulsion Laboratory, where he worked on the Mars Rover project. Then, he came to UCLA and founded the Wireless Health Institute. He is fond of saying that of all the things he's done in his career, from automotive research to spaceships, he believes the largest impact on humanity he's had is in the realm of digital health. He is a real optimist.
So, when I told Professor Kaiser about my idea for AbStats, he immediately got it. He got to work on building the sensor and developed important innovations to enhance the system. For example, he developed a clever way to ensure the device is attached to the body and not pulled off. This is really important, because if AbStats reports that a patient's intestinal rate is zero, then it might mean severe POI, or it might mean the device fell off. AbStats can tell the difference thanks to Professor Kaiser's engineering ingenuity.
Once we developed a minimal viable product, we worked like crazy to test it in the clinics, write papers, and publish our work. At the same time, UCLA licensed the IP to a startup company, called GI Logic, that worked with our teams to submit the FDA documentation. Professor Kaiser's team did the heavy lifting on the engineering and safety side, and we focused on the clinical side. It was a great example of stem-to-stern teamwork, ranging from in-house engineering expertise, to clinical expertise, to regulatory expertise. It all came together very fast.
Importantly, it was my sister who came up with the name "AbStats." I always remember to credit her with that part of the journey!
5. What role did patients play in the design of AbStats?
Patients were critical to our design process. We went through a series of form factors before settling on the current version of AbStats. At first, the system resembled a belt with embedded sensors. Patients told us they hated the belt. They explained that, after undergoing an abdominal surgery, the last thing they wanted was a belt on their abdomen. We tweaked and tweaked, and eventually developed two small sensors that adhere to the abdomen with Tegaderm. Even those are not perfect – it hurts to pull Tegaderm off of skin, for example. And the sensors are high profile, so they are not entirely unobtrusive. We're working on that, too. But patient feedback was key and remains vital to our current and future success with AbStats.
6. How did patients & physicians respond to AbStats during research & development?
It was gratifying that virtually every surgeon, nurse, and patient we spoke with about AbStats immediately "got it." This is not a hard concept to sell. Your bowels make sound. The sound matters. And AbStats can listen to those sounds, make sense of them, and provide feedback to doctors and nurses to drive decisions. The "so what" question was answered. If your belly isn't moving, then we shouldn't feed you. If it's moving a little, we should feed a little. And if it's moving a lot, then we should feed a lot. The surgeons called this the AbStats "stoplight", as in "red light," "yellow light," and "green light." Each is mapped to a very specific action plan. It's not complicated.
We were especially surprised by the engagement of nurses in this process. Nurses are the heart and soul of patient care, especially in surgery. Our nursing colleagues told us that feeding decisions come up in nearly every discussion with post-operative patients. They said they have virtually no objective parameter to follow, and saw AbStats as a way to engage patients in ways they previously could not. This was surprising. For example, the nurses pointed out that many patients are on narcotics for pain control, and that can slow their bowels even further. By having an objective parameter, the nurses can now use AbStats to make conversations more objective and actionable. For example, they can show that every time a patient uses a dose of narcotics, it paralyzes the bowels further. Knowing that, some patients might be willing to reduce their medications, if only by a little, to help expedite feeding decisions. AbStats enables that conversation. It's really gratifying to see how a device can alter the very process of care, to the point of impacting the nature of conversations between patients and their providers. Almost uniformly, the patients in our trials felt the sensors provided value, and so did their nurses.
7. Would you approach the problem differently if you had to do this again?
Not really. Considering that in 4 years we invented a sensor, iteratively improved its form factor, conducted and published two peer-reviewed clinical trials, submitted an FDA application, and received clearance for the device, it's hard to second guess the approach.
8. What other problems would you like to solve with the use of wearable technology in the future?
AbStats has many other applications beyond POI. We are currently studying its use in an expanding array of applications, including acute pancreatitis, bowel obstructions, irritable bowel syndrome, inflammatory bowel disease, obesity management, and so on. There are more opportunities than there are hours in the day, so we're trying to remain strategic about how best to proceed. Thankfully, we are well aligned with the startup, GI Logic, to move things forward. I am also fortunate to be at Cedars-Sinai Medical Center, my home institution since moving from UCLA, where most of the clinical research on AbStats was conducted. Cedars-Sinai has been extremely supportive of AbStats and our work in digital health. We couldn't do our research without our medical center, patients, administrative support, and technology transfer office. I am immensely grateful to Cedars-Sinai.
More generally, wearable technology and digital health still have a long way to go, in my opinion. I've written about that before, here. AbStats is an example of a now FDA-approved sensor supported by peer-reviewed research. I'd like to see a similar focus on other wearables. There are good examples, like AliveCor for heart arrhythmias, and now Proteus, which is an "ingestible." But, for many applications in healthcare, there is still too little data about how to use wearables.
I believe that digital health, in general, is more of a social and behavioral science than a computer or engineering science. Truth be told, most of the sensors are now trivial. Our sensor is a small microphone in a plastic cap. The real "secret sauce" is in the software, how the results are generated and visualized, how they are formed into predictive algorithms, and, most importantly, how those algorithms change behavior and decision making. Finally, there is the issue of cost and value of care. There are so many hurdles to cross, one wonders whether many sensors will run the gauntlet. AbStats, for example, may be FDA approved, but that doesn't mean we're ready to save money using the device. We need to prove that. We need data. FDA approval is a regulatory hurdle, but it doesn't guarantee a device will save lives, reduce costs, reduce disability, or anything close to it. That only comes from hard-fought science.
9. Are clinically proven medical applications of wearable technology likely to grow in years to come?
Almost certainly, although my caveats, above, indicate this may be slower and more deliberate than some are suggesting in the digital health echo chambers.
10. For those wishing to follow in your footsteps, what would you words of wisdom be?
First, start by addressing an unmet need. Clinical need should drive technology development, not the other way around.
Second, if you're working on patient-facing devices, then I believe you should really have first hand experience with literally putting those devices on patients. If you're not a healthcare provider, then you should at least visit the clinical trenches and watch what happens when sensors go on patients. What happens next can be unexpected and undermine your presuppositions, as I've written about here and here. I do not believe one can truly be a wearable expert without having literally worked with wearables. That's like a pharmacist who has never filled a prescription, or, a cartographer who has never drawn a map. Digital health is, by definition, about healthcare. It's about patients, about their illness and disease, and about figuring out how to insert technology into a complex workflow. The clinical trenches are messy, gray, indistinct, dynamic, and emotional — injecting technology into that environment is exceptionally difficult and requires first-hand experience. Digital health is a hands-on science, so look to the clinical trenches to find the unmet needs, and start working on it, step-by-step, in direct partnership with patients and their providers.
Third, make sure your device provides actionable data. Data should guide specific clinical decisions based on valid and reliable sensor indicators. We're trying to do that with AbStats.
Fourth, make sure your device provides timely data. Data should be delivered at the right time, right place, and with the right visualizations. We spent days just trying to figure out how best to visualize the data from AbStats. And I'm still not sure we've got it right. This stuff takes so much work.
Fifth, if your'e making a device, make sure it's easy to use and has a favorable form factor. It should be simple to hook up the device, it should be unobtrusive, non-invasive, with zero infection risk, comfortable, safe, and preferably disposable. We believe that AbStats meets those standards, although there is always more work to be done.
Sixth, the wearable must be evidence-based. A valuable sensor should be able to replace or supplement gold standard metrics, when relevant, and be supported by well designed, properly powered clinical trials.
Finally, and most importantly, the sensor should provide health economic value to health systems. It should be cost-effective compared to usual care. That is the tallest yet most important hurdle to cross. We're working on that now with AbStats. We think it can save money by shaving time off the hospital stay and reducing readmissions. But we need to prove it.
[Disclosure: I have no commercial ties to any of the individuals or organizations mentioned in this post]
