With Apple days away from potentially unveiling its health-tracking iWatch, all eyes are on the mobile health field. One medical device company that’s been working in the space for years is AliveCor; they introduced a smartphone case that doubles as an electrocardiogram, or EKG, back in 2012.
Their mobile EKG could save the day for someone suffering from heart disease or if they are using a pacemaker. But it’s still a reactive solution–alerts arrive only after a problem begins. Now, a team at AliveCor has found a way predict whether you’re likely to suffer a stroke or even a heart attack.
Up until now, the company relied on human doctors to analyze the various EKG readings that are gathered from users of the device. Soon, however, AliveCor plans to turn much of this analysis work over to algorithms, a process for which the FDA recently granted approval.
“Having achieved clearance, we will work to incorporate the algorithm in our app and plan to make this available to customers during September,” says Euan Thomson, an operating partner at Khosla Ventures who has been serving as the CEO of the company since last summer.
AliveCor’s heart monitor works by sensing the series of electrical signals that keep your heart beating. Because the human body is essentially an electrical conducting system, it’s possible to pick up the heart’s electrical pattern at certain points in the body. One of these areas is the fingertips. When a user rests their fingers on the AliveCor heart monitor, the electrical impulses are turned into ultrasound signals transmitted to the attached smartphone’s microphone.
Historically the only EKGs that were recorded came from patients sitting in a doctor’s clinic, with electrodes stuck on their chest. This meant the measurements were only taken in a select few locations, under certain conditions. Since irregular heartbeats may be infrequent, it also meant that patients were not necessarily suffering from symptoms when they came in to see their doctor for a scheduled appointment, making atrial fibrillation diagnosis a challenge.
“The way we had to treat it was to tell patients that if they suffer an arrhythmia they better run to the doctor’s office so we can capture and document the episode,” says Dr. Richard Wong, a cardiologist with Cardiology Consultants Medical Group of the Valley near Los Angeles, who has been using the monitor with patients for more than six months. “That seems so archaic now, but the reality is that there are just certain arrhythmia that only recur once every two or three months.”
Last February, the company launched a so-called “interpretation service” which uploads the EKGs of users so they can be analyzed by qualified teams of doctors. But right from the start the challenges of scaling the product were enormous.
“Currently we have well over one million EKGs in our database,” says Euan Thomson. “At the moment they’re coming in at a rate that is a little bit shy of 100,000 EKGs each month, from thousands of users who use this device on a regular basis.”
What was needed was an algorithm to do the job of analysis, and this is where AliveCor believes its data-driven approach will make a major impact. Algorithms have been used by cardiologists in hospitals for many years, but the one developed by AliveCor is capable of distinguishing between between normal and abnormal EKGs without the need for human intervention. Not only does their proprietary algorithm mean that AliveCor can serve more users faster, but their data can help make it even more effective.
To Thomson, what separates his device from dozens of other health-tracking apps and products available on the market is one thing: context.
“A lot of people confuse mobile health with the wellness apps that are around today,” he says. “These really don’t give any health insights. The counting of steps, for example, is all well and good–but nobody can tell you definitively whether 10,000 steps per day is going to stop you from getting sick, or will make you live longer. It’s not related to actual health care. These wellness products are really potential health care products without an outcome. As mobile health companies, we need to think about outcomes.”
What gives AliveCor’s heart monitor context, or makes it actionable in terms of patient outcomes, is the way it can be used to help doctors fine-tune patients’ treatment. In short, mobile health devices like this will make health care more personal. As it currently is, advice given by GPs and cardiologists around arrhythmia can be be incredibly broad-based: suggestions like cutting down on stress, getting a good night’s sleep, or avoiding coffee. All of these are good suggestions (like walking 10,000 steps each day) but none of them allow for any difference between patients, or aim to discover specific triggers for atrial fibrillation.
“One of my challenges as a cardiologist is to correlate heart rhythm with symptoms, and that can only happen when I’m able to look at what’s going on with a patient at a time when they’re symptomatic,” says Dr. Kevin R. Campbell, a cardiologist with North Carolina Heart and Vascular, at the UNC Chapel Hill School of Medicine. “Once you’ve done that it is then possible to modify these inciting factors to ensure that their impact is reduced.”
There’s a lucrative data science opportunity in determining those inciting factors. “For the first time, because of the presence of lifestyle trackers embedded into mobile devices–combined with electronic medical records and genetic analysis–we will be able to figure this out. It’s not even an ‘if’ so much as a ‘when,’” says Thomson. “The reality is that some combination of things are causing people to go into A-Fib. It’s an unknown currently only because we don’t have enough data.”
Mobile health devices will soon offer scientists and doctors the ability to tap into contextual information on patients. It’s not just about being able to take an EKG reading at any time of the day, but being able to marry this information with knowledge of what that person was doing at that exact time. Data from apps like Jawbone’s UP Coffee, for instance, would allow AliveCor to analyze how caffeine affects user heart rates: perhaps ruling out coffee as a risk for some patients, and alerting others of its significance. Similar developments are possible with the geo-location tracking made possible by today’s smartphones, to find out how specific locations or movement correlate to A-Fib. Or to use natural language processing to analyze Twitter feeds and cross-reference heart rate with stress (indicated through the use of “stressful” words used on social media channels).
“As someone who works with arrhythmia disorders this is really the future of medicine,” says Dr. Leslie Saxon, founder of the Center for Body Computing and chief cardiologist at USC’s Keck School of Medicine. “As a physician I want to be able to build and draw on large databases, and see artificial intelligence and sophisticated analytics applied to data in a way that’s really going to help people live longer, live safer, perform better, and stay safer. That’s going to be better for both the patient and the physician; it’s absolutely the future.”