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]

Enter your email address to get notified by email every time I publish a new post:

Delivered by FeedBurner

The Tyranny of The Should

Many of us lament at the paternalistic attitudes and behaviours, we both witness and experience, when dealing with healthcare systems. We hope that this is now the era when we can really transform healthcare, wearables may play a role in that transformation, or perhaps not. Now, one of the aspects of Digital Health I find most fascinating is the potential to change our behaviour, both at an individual level and a population level. So many firms have already launched some form of wearable technology which can monitor activity levels, heart rate and sleep. Many more are about to jump in this year. There are increasing levels of interest, excitement and expectation from these devices. Can simply tracking how much we walk, run or sleep really be the panacea to all our problems in healthcare? I don’t believe so, but some do. “The future of the NHS is in Apple’s hands”, is the headline of a post, by David Claire. He believes that the soon to be launched Apple watch will have a fundamental impact on the way the NHS runs. He even goes as far as to predict, “If self-awareness is the first step to kicking bad habits then the preventative care factor alone of the Apple Watch and similar devices will save the NHS billions.”

During 2014, I personally tried so many different pieces of wearable technology, hoping I could gain new insights about myself and use them with their apps to support my desire to live a healthier life. The truth is that none of them really worked for me. None of them engaged me for more than a few weeks, and also I just didn’t find anything that truly integrated with my life.

Two of my most recent purchases, in December 2014, have been the Samsung Gear S, and the Basis Peak. I had high hopes about the Basis Peak, as it can automatically detect when you fall asleep and when you wake up. I stopped using it to track my sleep after 3 nights. Why? I just didn’t feel comfortable wearing a watch whilst I slept. Apparently, we “should” be walking at least 10,000 steps a day. So I set my daily goal with the Samsung Gear S at 10,000 steps. A message pops up when you’ve achieved 50% of your daily goal, and how much time you’ve got to achieve your goal. Samsung’s website says the Gear S provides a “smart wearable experience.” Surely, if was a genuinely smart wearable, it would know that it’s about to going to rain all afternoon, and I’m unlikely to meet my goal?”

The Samsung Gear S highlights how much time I have left to complete my goal of 10,000 steps

The Samsung Gear S highlights how much time I have left to complete my goal of 10,000 steps

There have been headlines recently about “sitting disease”, and how being inactive is dangerous to our health. Hence, I’ve programmed my Gear S to bleep if I’ve been inactive for 30 minutes. Not particularly useful when you’re in the middle of eating lunch with a friend, despite the device itself knowing I was at lunch because it popped up a calendar reminder 10 minutes before the lunch meeting. Still a way to go before we can truly regard these devices as smart.

What also bothers me about these devices is that currently we’re simply digitising flawed messages. Does simply being told to be more active, or eat less, or sleep more actually make a difference? There is now scrutiny on the recommendations of 150 minutes a week of physical activity. I “should” walk 10,000 steps a day, and when my watch informs me that I haven’t, then I don’t feel that good. We hear lots of talk about “Gamification”, the idea that gamifying health can make a difference. I was chatting with Dr Pritpal S Tamber a few months ago, and he mentioned the term, “Shamification.” Is making us feel shame or guilty about our behaviour really the way to improve health, both of individuals and populations?

I’m not surprised that research shows wearables being are not being used after 6 months. You get called into a series of meetings, or your child is sick, and suddenly, life gets in the way of maintaining a healthy lifestyle. At present, the device or app won’t take that into context, it will still ominously display that you haven’t met your goal.

In the US, wearables are getting on people’s wrists from employer wellness programs. In the UK, a survey found 55% of business leaders would look at the health data from an employee’s smartwatch or health app, if that employee called in sick. Employees calling their boss and pretending to be sick could become a thing of the past. However, in the same survey, just 6% of employees would want their boss to have access to the health data from the watch or phone.

You might believe that it’s easy to deceive the system, because you could put the device on your dog, cat, or even another person. Yes, that’s true right now, but that could be impossible after 2018, if research in Australia is successful. Researchers at the University of New South Wales are embarking on a 3 year study to develop an algorithm that can verify who generated the data, so that the data could be fed back into mainstream health systems.

So many are hoping that these emerging technologies can integrate into our lives and provide the “Digital Nudges” to that will empower us to live healthier lives. The Internet of Things promises smart homes with all our devices connected to each other. Mitsubishi of Japan have showed their concept of a smart home. In their vision of the future, your fridge would connect the data from the sensors on your body with the data on the contents of the fridge, and display recipes on the fridge door. Is this concept dystopia or utopia?

Today, childhood obesity is of growing concern, around the world. A fascinating use of technology may be tested in schools in Dubai this year. Using a payment system linked to the child’s phone and an app that contains information on the nutritional content of food in the school cafeteria. So if the child tries to buy a burger at school, and that burger would take them beyond their daily nutritional allowance, the payment is declined, and the app suggests something healthier. Now that system is under the control of parents, but is this the route to dealing with obesity with adults too? In an increasingly connected world, do you want machines and algorithms limiting your choices, or would you prefer the freedom and autonomy to eat what you want, when you want?

What if the government offered you a way of paying for public transport using wearable technology that would also monitor your health status, provide suggestions, and even reward you if you decided to walk than take public transport? Such a scheme just launched in Beijing, with a plan to roll out to 400 cities across China. 

In the UK, it appears the NHS will have a “huge rollout” of wearable technology as part of a “revolution in self care”. Being able to monitor patients remotely, especially those with a chronic condition, is admirable. If entities in healthcare will be able to monitor us remotely, surely that’s always going to be a good thing? Perhaps not. Given the huge financial pressures facing the NHS over the next 20 years, we may have to ration access to care. In the future, could all this data collected about our behaviour be used to ration or even deny care? I’m not the only one who is asking that question.

In a great article by Hamza Shaban examining the impact of sensors collecting data about our health on the pricing of health insurance, one sentence stands out, “Imagine a pricing scheme that would punish sleep-deprived single parents or the dietary habits of the working poor.” A world where our health insurance premiums decline when we behave within the guidelines, and rise when we deviate from the guidelines. Will you avoid watching late night movies on TV because sleeping less would increase your health insurance premium next month?

Should doctors “Google” their patients? A really fascinating BBC podcast highlights the potential for the impact on the trust between patient and medical provider. There is a huge need to understand what happens to you in between visits to the healthcare system. In this new world of monitoring, if your healthcare provider has recommended you cut down on alcohol consumption, but can then obtain data on how much alcohol you’re continuing to drink, is that too intrusive? Today, the National Institutes of Health announced it's searching for a wearable or otherwise discreet device capable of measuring blood alcohol level in real time.

There is a fine line between “Digital Nudges” and “The tyranny of the Should” – and it’s not clear to me that we’re having the right conversations in the right places. Will we be nudged into living healthier lifestyles because we want to or because we should? Is everyone a winner or will there be losers? Is there really a place for autonomous decision making within the context of improving our health?

Hugo Campos, a patient in the USA, concludes his recent interview with a critical question, "Will WE have autonomy over ourselves and the data WE create?" Progress of technology can be easy to predict, what we can’t predict very well are the consequences of technologies. In the realm of Digital Health, some believe that power is moving away from the healthcare systems into the hands of consumers and patients, but over the long term, is the power going to move full circle? This year celebrates the 800th anniversary of the Magna Carta, an ancient symbol of justice over tyranny and protecting individual liberties. In 2015 and beyond, do we need a Magna Carta for Digital Health?

[Disclosure: I have no commercial ties to any of the individuals or organisations mentioned in the post]

Enter your email address to get notified by email every time I publish a new post:

Delivered by FeedBurner