Filip Brož Heart rhythm disorders can be a very unpleasant disease, as you often do not have to recognize and record such a problem at all. These are disorders that occur very sporadically, but if you do not have your heart examined with an EKG, you may not find out about them at all. Therefore, the developers of the watch application cardiogram created an AI-based algorithm that can detect atrial fibrillation with 97% accuracy.
If you have an Apple Watch with the Cardiogram app on your wrist, there is a high chance that if you have a heart rhythm problem, you will detect it. "Imagine a world where your heart can be monitored 24/7 using a device you buy at an electronics store or online," he says on the Cardiogram blog software engineer Avesh Singh, adding that their app's algorithms can transform raw heart data from your Apple Watch into specific diagnoses.
"These can then be automatically sent to your doctor, who is alerted to everything in a timely manner," continues Singh. For example, a cardiogram can warn of an impending stroke or heart attack.
The developers teamed up with the UCSF Cardiology Clinic in San Francisco more than a year ago to launch the mRhythm study involving 6 users using the Cardiogram app. Most of them had normal ECG results, but 158 participants were diagnosed with paroxysmal atrial fibrillation. The engineers then applied the aforementioned algorithm to the measured cardiovascular data and trained deep neural networks to recognize abnormal heart rhythms.
With this combination of cardiovascular data and deep neural networks, the engineers were finally able to achieve a high 97% success rate in detecting atrial fibrillation, which is not easy to detect otherwise.
Atrial fibrillation affects 1% of the population
Atrial fibrillation, or atrial fibrillation, is the most common heart rhythm disorder in adults. More than 4,5 million people in Europe suffer from it. The name itself comes from the fibrillation (shaking) of the heart muscles in the atria. This condition leads to a fast, slow or irregular heartbeat. Atrial fibrillation is caused by a malfunction in the transmission of electrical signals that control the contraction of the heart.
The disorder puts a person at risk by impairing the heart muscle's ability to pump blood, thereby causing blood clots to form in the heart chamber. The risk of atrial fibrillation increases with age and affects one percent of the adult human population worldwide. One in four adults over the age of 55 suffers from this disease.
Of course, lifestyle and other pathological diseases, such as diabetes, obesity, high blood pressure, lung cancer or excessive alcohol consumption, also influence the disease. However, most people with atrial fibrillation have no symptoms, especially if their heart is not beating very fast. The most significant symptoms are then excessive heart palpitations, dizziness, chest pain or shortness of breath. Early detection of this disease can prevent a stroke or heart attack. Treatment takes place either with drugs or with a minor surgical procedure, so-called catheterization.
It was the second method of treatment that I underwent twice in my childhood. During a random check-up at the pediatrician, I was diagnosed with a heart rhythm disorder. At that time, I was a top athlete and I was told that in extreme cases and with enormous physical activity, a cardiac arrest could occur, which is not unusual. Unfortunately, many athletes have already died in a similar way, for example, when they suddenly fell to the ground during a football match.
A big step into the future
"The most promising finding of our study is the evidence that wearable electronics can be used to detect disease. The future is bright here, and there are several research directions that are particularly interesting to us," says Singh. I more than agree with this statement. I'm honestly excited about their research, as I've always envisioned this direction of collaboration between app developers and Apple described several times.
Cardiogram's developers want to continue to pursue deep learning to deliver personalized care. “Suppose an app notifies you of a panic attack. Combined with the measured data and our algorithm, the user receives simple advice such as take three deep breaths and exhale,” Singh exemplifies.
"In the future, we don't just want to detect the disease, but also treat it directly in the sense: the application has detected abnormal heart activity - do you want to contact your cardiologist or call an ambulance?" calculates the developer of Cardiogram. After connecting with the doctor, the developers want to continue to monitor the progress of the patient's treatment and its effects. They also want to implement the heart rate measurement algorithm in other human activities, such as sleep, driving a car or sports. The result is the early detection of the disease with the help of smart devices and the initiation of the necessary treatment.
In connection with health and the Apple Watch, there is also something else being talked about in recent weeks. Although the operation of Cardiogram pushes "mobile healthcare" somewhere further, Apple is said to be working on even more revolutionary matters. According to CNBC Apple boss Tim Cook himself tested a prototype device that pairs with the Watch and can non-invasively measure blood sugar levels.
This would mean a fundamental breakthrough in the treatment of diabetes, because it is currently not possible to measure the blood sugar level, which diabetics need to know, non-invasively. Current sensors on the market have to go under the skin. For now, it is not clear what phase Apple is in testing, but at least the prototype should be out in the world. It is not even clear whether Apple will be able to integrate the device directly into the Watch, but even if it was initially supposed to be a separate non-invasive glucose meter, the Californian company would start another revolution.
Good day. I installed Cardiogram on my iPhone, played with the watch. After starting the alication in the watch, the measurement takes place, but the synchronization in the iPhone does not take place. I am asking for detailed instructions on how to use the application. thank you Pavel Wasik
Hello, detailed instructions can be found at http://cardiogram.helpscoutdocs.com/
I installed it and will probably only use it as a "sender" of data for the study, otherwise I find it terribly confusing...