Researchers Using AI and Apple Watch ECGs to Detect Weak Heart Pump

[MacRumors]

Researchers at the Mayo Clinic have created an artificial intelligence algorithm that is able to use single-lead ECGs taken by the Apple Watch to find patients who have a weak heart pump, according to research data that was shared this week at the Heart Rhythm Society conference.

A weak heart pump, or left ventricular dysfunction, is an issue that affects two to three percent of people around the world, and up to nine percent of people over age 60. Like atrial fibrillation, another heart issue the Apple Watch can detect, a weak heart pump can have no symptoms. It can also be accompanied by symptoms that include racing heart beats or shortness of breath.

Paul Friedman, chair of the Department of Cardiovascular Medicine at the Mayo Clinic in Rochester, Minnesota, said that it is "absolutely remarkable" that AI is able to use a consumer watch ECG to detect the condition, as it is normally something that requires an echocardiogram, CT scan, or MRI to identify.

The ECG feature on the Apple Watch is a single-lead ECG, and it requires users to place a finger on the Apple Watch digital crown for 30 seconds. Results are uploaded to the Apple Health app and can be shared with medical professionals. The ECGs are designed to help detect atrial fibrillation, but ECG functionality and other Apple Watch features are also being studied in the detection of other conditions.

A standard ECG uses 12 electrode leads that are placed on a person's chest, arms, and legs to monitor the heart's electrical signals. To use the single-lead ECG results from the Apple Watch, researchers modified an existing 12-lead algorithm that is known to be able to detect a weak heart muscle.

The study included 125,610 ECGs collected over a period of six months from 46 states and 11 countries. Each person submitted many ECGs, and the cleanest readings were used for the algorithm. Several hundred of the participants had clinical testing to measure pump strength, and that data was used to determine whether the Apple Watch could detect an issue.

Approximately 420 patients had a watch ECG recorded within 30 days of a clinically ordered echocardiogram, or ultrasound of the heart, a standard test to measure pump strength. We took advantage of those data to see whether we could identify a weak heart pump with AI analysis of the watch ECG. While our data are early, the test had an area under the curve of 0.88, meaning it is as good as or slightly better than a medical treadmill test. AI analysis of the watch ECG is a powerful test to identify a weak heart pump.

Researchers plan to launch global prospective studies to do further testing in more diverse populations to demonstrate the benefit of the single-lead ECG feature in the Apple Watch.

"This is what the transformation of medicine looks like: inexpensively diagnosing serious disease from your sofa," said Friedman.

Article Link: Researchers Using AI and Apple Watch ECGs to Detect Weak Heart Pump

 

[jav6454]

This means, that a feature like this for the Apple Watch could come standard after 2 years taking into account internal testing and FDA approval.

Apple was smart to drive independent research using the Apple Watch. It offloads R&D and very likely yields better ideas on the Apple Watch’s capabilities.

 

[ethanwa79]

"meaning it is as good as or slightly better than a medical treadmill test"

This is actually quite amazing!

 

[diddl14]

would share all my medical data for these studies in a heartbeat

 

[TheYayAreaLiving 🎗️]

So much research is going behind this. Wow, this is for a good cause. Saving lives! ❤️

 

[jz0309]

great results, more studies planned.
Big question: will Apple include this "AI" in watchOS/iOS? and if so, when?

 

[_Spinn_]

This is fantastic! I really think there is so much untapped potential from the Apple Watch. So much data is currently collected but I feel like there's not much to do with it at the moment.

 

[DocMultimedia]

Would be great if they let anyone upload their data to get an unofficial result (there would of course be fifty legal warnings about the unofficial part).

Says a sixty year old who ran a 6:20 mile yesterday... I'm always worried about my heart cause it sure beats a lot during races.

 

[Corsig]

Apple Watch is an amazing product

 

[wigby]

This is fantastic! I really think there is so much untapped potential from the Apple Watch. So much data is currently collected but I feel like there's not much to do with it at the moment.

There are two different approaches that keep this from moving faster. 1. Regulations - medical devices (and their algorithms) require years of testing in order to provide useful applications that are also safe. 2. Capitalism - Instead of working together, companies like Google and Apple generally fight it out to see who dominates in the marketplace. Once a winner (usually revenue-based) is declared, manufacturing partners can move in and accelerate the process. We are lucky enough to have both of those systems working here in the U.S. but they also both have their own pros and cons.

 

[hasanahmad]

Jony Ive was against Apple Watch being used for this. If he was in charge, fashion analysts would be researching using Apple Watch by models on the next runway show

 

[Whathappened]

Would be great if they let anyone upload their data to get an unofficial result (there would of course be fifty legal warnings about the unofficial part).

Says a sixty year old who ran a 6:20 mile yesterday... I'm always worried about my heart cause it sure beats a lot during races.

I don’t think you have to worry if you have no previous illness. Consult your doctor to assure you and keep up the good work! ?

 

[DrV]

Medicine is going to change in a very radical way. This study is an example of the transformation.

Throughout history, medicine has been about making decisions with too little information. Almost any measurement has been a single point in time. This has applied to body temperature, heart rate, ECG, blood pressure, blood oxygen, physical activity, or glucose levels. All these parameters — and a lot more — can be measured continuously and in real time. The scarcity of information becomes a flood of information.

Until now, all interpretation has been carried out by the physician. Even ECGs (simple, homogeneous one-dimensional signal) have been interpreted visually. With the abundance of data, this is no longer possible. Meaningful handling of the data requires sophisticated algorithms.

This sounds very good, and few people disagree with the basics. However, there are several psychological obstacles we need to pass. At least:

1. The data needs to be collected and stored so that it is interchangeable between different parties. But, if we create a large data storage with all electronic health records, we face a risk of severe privacy violations. That data can be used in many malicious ways. Just think of blackmailing or nation-state cyberattacks.
2. We do not know what the data means. There are huge amounts of data collected by, e.g., sports watches, but very little of that data can be used in medicine, as we do not understand the data. There are all sorts of obstacles from regulations to inability to figure out the business model hindering research in the field.
3. The only known way to handle a lot of the data is to use machine learning (often neural networks). The algorithms may produce complete black boxes with no visibility into their inner workings. While this is the way people operate, it is not the way we expect medical information systems to operate. On the other hand, if we decide to exclude black-box algorithms, we'll most probably toss away a lot of life-saving information.
4. In some cases, it is possible (and has actually been shown) that data-based algorithms may give better results than human doctors. Errare humanum est, unfortunately, does not hold true, and people will die because of misdiagnoses by algorithms, if we choose to use them. Fewer people than by human errors, but still decisions by computers will kill people. Are we ready? (We can, of course, try to duck the responsibility by saying that "computers recommend, doctors decide", but that is an unfair approach.)
5. If algorithms yield better results than human physicians, phycisians will become hands and eyes for computers. That is a very big change in mindset, and has a profound impact on hierarchies. It may also change the skill set required from medical practitioners.

The change is unavoidable, and it is required if we want to keep the healthcare of western societies at today's level. It will be a painful change, though.

 

[Seoras]

"inexpensively diagnosing serious disease from your sofa"

Yeah, the sofa. Maybe get off the sofa?

 

[Leon Ze Professional]

In my eyes, the health related issue detection abilities of the Apple watch is what sets this device apart from anything else. Already what it can do is "absolutely remarkable".

Tim Cook was right to make this the focus of the Apple watch. The fashion accessory part is certainly a secondary component and contactless payments via the watch is a nice to have feature in Covid times.

Keep up the brilliant R&D on this Apple!

 

[Keymaster]

great results, more studies planned.
Big question: will Apple include this "AI" in watchOS/iOS? and if so, when?

They will probably wait until they get FDA approval, after the research studies are done. That's their best time to make a big splash...a watch that can detect heart problems while you watch TV or take a walk is a huge, huge deal.

This means, that a feature like this for the Apple Watch could come standard after 2 years taking into account internal testing and FDA approval.

Apple was smart to drive independent research using the Apple Watch. It offloads R&D and very likely yields better ideas on the Apple Watch’s capabilities.

It's really the best play, simply because they get to run the research through labs that are experienced, know how to do it, and know how the regulatory system works for medical devices. Apple has always seemed to work with universities instead of trying to do every research project in house (like other similar companies), and that always leads to better results as you get the field experts involved in the work who already are set up to do the research. Even if you hire someone good, they still have to create a lab, hire people, and that takes time...plus you miss out on students who are both eager and have fresh eyes.

 

[eifelbube]

There are two different approaches that keep this from moving faster. 1. Regulations - medical devices (and their algorithms) require years of testing in order to provide useful applications that are also safe. 2. Capitalism - Instead of working together, companies like Google and Apple generally fight it out to see who dominates in the marketplace. Once a winner (usually revenue-based) is declared, manufacturing partners can move in and accelerate the process. We are lucky enough to have both of those systems working here in the U.S. but they also both have their own pros and cons.

3. Must not lead to reduced income for cardiologists …

 

[Apple_Robert]

Amazing news. I hope we see the benefit of this in a few years.

 

[jav6454]

It's really the best play, simply because they get to run the research through labs that are experienced, know how to do it, and know how the regulatory system works for medical devices. Apple has always seemed to work with universities instead of trying to do every research project in house (like other similar companies), and that always leads to better results as you get the field experts involved in the work who already are set up to do the research. Even if you hire someone good, they still have to create a lab, hire people, and that takes time...plus you miss out on students who are both eager and have fresh eyes.

Yes that's true. Besides, such development at universities tends to translate to more valuable data for improving or introducing features/algorithms for this sort of thing.

Not only are the true experts there, but they also tend to think outside the box when researching with an advanced tool. Something an engineer with no true knowledge (besides making the algorithm) will be able to perfect.

 

[44267547]

This is a really interesting piece from the article:

“A weak heart pump, or left ventricular dysfunction, is an issue that affects two to three percent of people around the world, and up to nine percent of people over age 60.”

******************

What’s interesting about that, if this feature proves to be reliable and makes its way into the market, that small percentage of 2% to 3% that might have a ‘weakened left ventricle’, could potentially be saved by a smart watch. I don’t think anyone would’ve pondered these advancements when the watch first launched in April 2015.

But like anything with an intricate advancement like this, I’m curious to see how long it actually makes its way into the market and how reliable it really is.

 

[svish]

Apple watch is advancing its health features!!

 

[alien3dx]

This is a really interesting piece from the article:

“A weak heart pump, or left ventricular dysfunction, is an issue that affects two to three percent of people around the world, and up to nine percent of people over age 60.”

******************

What’s interesting about that, if this feature proves to be reliable and makes its way into the market, that small percentage of 2% to 3% that might have a ‘weakened left ventricle’, could potentially be saved by a smart watch. I don’t think anyone would’ve pondered these advancements when the watch first launched in April 2015.

But like anything with an intricate advancement like this, I’m curious to see how long it actually makes its way into the market and how reliable it really is.

maybe mitral valve prolapsed / regutite?

** we do have mitral valve prolapse.

 

[wigby]

3. Must not lead to reduced income for cardiologists …

I went to my cardiologist and he spent 10 minutes talking to me and another 10 minutes scrolling through my Health app. He loved all the data because it gave him a quick insight into my activity and heart rate patterns. Also, he told me about all the false triggers that brought users to his office. I'm sure he is annoyed but billing more patients than ever. Cardiologists love this device.

 

[NobleSwineherd]

Two years ago, a-fib put me in the hospital for two days. If a device like an Apple Watch could accurately detect a-fib and automatically send the data to my cardiologist, that would be a killer app (metaphorically speaking, of course)! I have never worn a watch regularly (or jewelry of any kind) but if Apple's next gen of watch can accurately detect a-fib, I'm getting one.