Wear OS Weekly
My weekly column focuses on the state of Wear OS, from new developments and updates to the latest apps and features we want to highlight.
Google’s Pixel Watch engineers collaborate with cardiologists and other clinical experts to improve their sensor data, while relying on AI and algorithms to give the next best thing to your doctor’s advice. So I had Google’s senior clinical lead explain how her medical expertise factors into your wearable tech — and where things are headed next.
I spoke with Dr. Jacqueline Shreibati, a practicing cardiologist and adjunct faculty at Stanford University who previously served as Chief Medical Officer at AliveCor, a startup dedicated to AI insights from mobile EKGs.
Aside from serving her patients, she guides the consumer health team consulting on health projects like Loss of Pulse at Google. She is “embedded” in a “shoulder-to-shoulder process” with Google’s engineers when envisioning new health features, while also connected with the external cardiologists and practitioners who test these tools before they ever show up on your Pixel Watch 3.
I asked Dr. Shreibati how they convert ideas like Loss of Pulse into actionable features, how wearables will transition from “wellness” devices to “medical” devices, how Gemini AI factors into the discussion, and what to expect next from Pixel Watches.
The long process to make Loss of Pulse detection possible
Out of the 350,000 out-of-hospital cardiac arrest events in the United States, at least half aren’t witnessed, meaning the chance of survival is “extremely low,” Shreibati explained. The idea was to add a virtual witness to change the equation.
They knew that a Loss of Pulse feature “lined up at a high level” with Pixel Watches’ current capabilities — optical heart rate and emergency SOS — but the “real work” began with figuring out how to catch it accurately without false positives. That’s where Shreibati’s expertise came into play.
Google says it tested Loss of Pulse’s algorithm over “hundreds of thousands of hours,” using tourniquets for artificial pulselessness or putting watches on patients in a cardiac electrophysiology lab who were scheduled to have “induced” no-pulse procedures. This gave Google’s “AI-powered algorithm” the data it needed for accurate Loss of Pulse readings.
Google knows it needs these kinds of external tests but relies on cardiologists like Shreibati to facilitate them, as she can think of methods and scenarios that Google engineers can’t. For example, Shreibati talked about using stunt actors to simulate the falls people might experience during cardiac arrest, giving your Pixel Watch another point of data with its accelerometer.
I asked Shreibati why this feature took so long to arrive in the United States, since Europe got Loss of Pulse back in September 2024. She answered diplomatically that “each regulatory body has its own process,” and that the FDA’s rigorous (or simply slow) testing period is “not within our control.”
So when we see Fitbit patents like detecting arterial stiffness or measuring your blood pressure that are years old, it doesn’t necessarily mean that Google has abandoned the idea. It could simply be a matter of needing years-long testing to make sure the feature is accurate enough to pass FDA scrutiny.
I wrote in a previous column how tech brands are pushing to transition from “wellness” to “medical” devices, going beyond advice on general energy levels and sleep quality to actual diagnoses. So I questioned Dr. Shreibati on Google’s perspective on this trend, as well as her own.
The typical clinical approach, she explained, is to use population-level guidelines to calculate your risk scores for conditions like heart disease; it’s more about extrapolating than knowing, at least until you go through diagnostic tests.
But the Google engineers she works with think that you should know your “personalized risk using data rather than extrapolation,” combining your wearable health data with your medical records for more advanced, preventative insight before you ever step in a doctor’s office.
We’re starting to see hints of this with features like sleep apnea detection, but when it comes to detecting more serious heart health conditions, “We’re not there yet, but that’s the goal,” Shreibati says.
Part of that comes down to using machine learning algorithms to connect the dots between data and conditions, but it’ll require “rigorous validation and verification” before wearables can offer that kind of insight in consumer apps. After all, misdiagnosing someone could lead to serious anxiety and expensive check-ups.
In theory, though, Google’s clinical team seems to believe that Gemini insights will make a difference in terms of accurate medical recommendations in the future.
When I asked Shreibati about the potential for diagnoses from wearables to become a source of anxiety and fatalism, she argued that it’s a matter of determining your “relationship with the data.” Early warnings of potential issues can be “empowering,” but they have to think about the frequency of how often a watch warns you about things before it becomes “distracting or overwhelming.”
Using your watch to get healthier
As someone immersed in both cardiology and the tech world, Dr. Shreibati would have a unique perspective on the whole premise of wearables: To use health and fitness data to guide a healthier lifestyle. So I asked if she had any advice for Pixel Watch owners.
She responded by pointing out how the Fitbit app has customizable goals for a reason: whether you should focus more on weight, sleep stress, VO2 Max, or heart health will vary by person to person.
Without specific data, she can only focus on “the things we know we all have to do.” In her mind as a cardiologist, that’s getting sleep, making time to be physically active throughout the day, eating more fresh fruits and vegetables, and monitoring important stats like your blood pressure and weight.
In terms of future potential Pixel Watch features to help you get healthier, she talked about Fitbit “exercise plans,” which could potentially build on the daily recommendations you get based on your Fitbit cardio load for more long-term guidance.
She also mentioned using Health Connect to increase the amount of data that Fitbit (or Samsung Health) have access to, with blood pressure as her example. It’s much more important to have that data from medical-grade sources for their machine-learning algorithm, rather than wait for smartwatches to start tracking blood pressure, glucose, and other stats accurately.
The one question Dr. Shreibati couldn’t answer was how Google claims to make its heart rate algorithm more accurate year over year while using the same sensor hardware — as that’s not her area of expertise.
I received the following response from a Google spokesperson: “We are able to drive these improvements with a deeply research-based approach, combined with expertise in machine learning, to push the boundaries even further.” This was more marketing than explanation, but does emphasize how much ML work goes into making Pixel Watches more accurate.
Generally speaking, I found the interview quite enlightening. I’ve seen how these companies hire doctors who then attest to the accuracy of these devices, but only now do I have a clearer sense of how they fit into the wearable pipeline. They envision big-picture ideas and work with engineers to figure out what’s “technically feasible,” as Shreibati put it, and then communicate with doctors and patients outside of the tech sphere to see how the ideas work in practice.
My next Google interview will be with a sleep expert on their team, so feel free to drop a comment if you have any questions about sleep tracking — or questions I missed about heart health data that you’d like me to forward along!
Source link
