Is Google's new 'loss of pulse' feature useful?

Earlier this week, Google announced its forthcoming Pixel Watch 3 will include a new "loss of pulse detection" feature meant for people who go into cardiac arrest when no one is around to help them.

How does it work? The algorithm is designed to use signals from the smartwatch's heart rate and motion sensors to detect loss of pulse in under a minute, including a "check in" prompt period which lasts about 15 seconds while the feature continues to look for signs of pulse or motion. If the user doesn't respond to the check-in, a 20 second countdown starts after which the device uses built-in cellular or a connected smartphone to contact emergency services with context and location.

How was it developed? Google representatives said it would be impossible to design and test the feature prospectively by handing out devices and waiting for events to happen. So the company captured simulated loss of pulse events in several ways to develop and test the feature.

How well does it work? Google did not share any data about the performance of the feature saying it was working on a paper to submit to a peer-reviewed journal. The feature, I was told, was developed and tested on a diverse population.

Who is it for? The feature is meant for the general population, in which loss of pulse might be caused by cardiac arrest or overdose. Google specifies that the feature is not to be used by people who have preexisting heart conditions. And Google told me the feature is not designed to be a monitor, so people who need one should consult their doctor about using the right device.

Will it make a difference? When someone's heart stops, time is the enemy. For people to survive, CPR needs to be started basically immediately, and without it, brain damage and death follow within minutes. Even if the feature is technically feasible, it remains to be seen whether it realistically helps people get help fast enough to make a difference. Ted Melnick, an associate professor of emergency medicine at Yale School of Medicine told me in an email that "practically speaking, this feature would only benefit users in areas with very rapid EMS response times — such as densely populated areas near hospitals."

I can imagine all kinds of scenarios where maybe this helps and others where it doesn't. If you have a thought, don't hold back: mario.aguilar@statnews.com

Medical devices

Studying how people really use brain interfaces

Brain computer interfaces, or BCIs, promise to greatly expand communication for people who have been paralyzed. For the five million-plus people in the United States living with paralysis, including those diagnosed with neurodegenerative diseases like ALS, this could extend how long they're able to communicate with their loved ones.

To get a sense for the state of the art, STAT's Timmy Broderick spoke with neuroscientist Mariska Vansteensel, the president of the international BCI Society. Vansteensel has spent most of the last decade experimenting with BCI at-home use, rather than in a laboratory. And, in a new paper in the New England Journal of Medicine, her team chronicled an ALS patient's use of the device for over seven years. They found that the patient's BCI usage increased as time passed and her paralysis increased. The technology could prove fruitful for ALS patients as common communication techniques for people with paralysis that use eye-gaze lose utility as eye muscle control wanes.

"Most of the research is still in the lab. We need to learn what happens in settings of daily living. What are the things that you have never thought about? Answering those questions is important, if we want to clinically implement this technology."

With AI regs brewing, researchers look for lessons in existing systems

As the debate about how to regulate artificial intelligence in health care continues, two new viewpoints on the matter were published in JAMA. In one, MedStar's Raj Ratwani, the University of Utah's David Classen, and UCSD's Christopher Longhurst pointed out flaws in the current health IT certification system used by the Office of the National Coordinator, including the conflicts of interest that arise when tech developers pay the certifying bodies to certify their products — and suggested improvements for an eventual health AI certification process.

The other, from Duke health AI leaders including CHAI board member Michael Pencina, proposed a federal registry that would be a "clinicaltrials.gov for health AI." The registry entries could use standards such as those currently being formulated by the industry body CHAI, the authors noted.

Until we have a federal registry, don't forget to check out STAT's Generative AI tracker to stay abreast of how health systems are using AI tools.

Medical devices

Medtronic-backed AI-polyp detection beats unaided exams in study

Colonoscopies performed with the help of an AI system outperformed unaided exams in newly published randomized trial. The system, called GI Genius, highlights potential areas of interest in live colonoscopy feeds and was cleared by the Food and Drug Administration in 2021. The study, published this week, was conducted in ordinary clinical practice among 2,000 people receiving colonoscopies. Polyps were detected in 56% of participants who were examined with the help of the system and 48% of those receiving an unaided colonoscopy. An increased detection rate has been associated with improved colorectal cancer outcomes, and the study helps build the case that more practices ought to use AI systems to aid their exams.

The study was funded by Medtronic, which distributes GI Genius and has invested in the tool's developer Cosmo Pharmaceuticals.

More around STAT

Check out more exclusive coverage with a STAT+ subscription

Read premium in-depth biotech, pharma, policy, and life science coverage and analysis with all of our STAT+ articles.

What we're reading

Surgeons Cut a Giant Tumor Out of My Head. Is There a Better Way?, Bloomberg
Genetic variation impacts drug efficacy. Could testing help?, Undark
A.L.S. stole his voice. A.I. retrieved it, New York Times
Kaiser Permanente rolls out Abridge's gen AI clinical tech across 40-hospital system, Fierce Healthcare