Algorithm That Diagnoses Heart Arrhythmias

Cardiac arrhythmia is in most cases harmless extras without the disease. Both a slowing of the heart rate (bradycardia) and a speeding up of the heart rate (tachycardia) can also become life threatening. Nearly every person is affected by cardiac arrhythmia during his life.


These can be harmless or dangerous. Often they are only subjectively unpleasant ("stumbling-heart") and do not need treatment. However, untreated pre-fibrillation is accompanied by a marked increase in stroke risk. This makes it all the more important to have an early and reliable algorithm diagnosis. However, they can also lead to dangerous incidents and subsequent episodes.

But how can pre-fibrillation be detected when it occurs irregularly and the patient himself often does not notice it? An ECG - that is, the measurement of cardiac currents - is the gold standard in the algorithm diagnosis of atrial fibrillation. In that case, it can help measure cardiac currents over a longer period of time and keep the heart in view all the time. It is important that even small signals and short episodes can be perceived.

Definition and types of cardiac arrhythmia

The normal heart rate is 50 to 100 beats per minute. The heart rate is generally higher in young women than in men, presumably because in men the rest frequency decreases by frequent sports. A heart rate <50 a="" bradycardia="" called="" frequency="" is="" min="" slowed=""> 100 / min at rest is too fast (tachycardia). Both manifestations of cardiac arrhythmia can be life-threatening. They then usually occur as a result of severe structural heart disease, e.g. Heart attack, cardiac insufficiency, or heart valve defect, or in the case of congenital or severe hypertension caused by hypertension.

Cardiac arrhythmia can be divided into disturbances of stimulation and disorders of the excitatory conduction. Causes of ectopic stimulus formation can be increased automatization, abnormal automatism, and triggered activity. Excitation line disturbances can lead to arrhythmias in linear closed conduction pathways or also in the spatial whole cellular network.

Thanks to the computer scientists at Stanford, who developed an algorithm that diagnoses 14 types of Heart defects. These group of Stanford scientists led by Andrew Ng has developed this deep learning algorithm. With the collaboration with the iRhythm – the heartbeat monitoring company, roll up a solid data set that they used to teach a profound neural network model.

The new heart monitor has now been developed for this purpose. The special feature of this heart monitor, which can also be used on an outpatient basis, is its intelligent design. Its not only records the ECG, it also analyzes it and sends the most important episodes on the so-called home monitoring - a technology built into the implant - to the treating physician every day.

Thanks to this close-meshed observation of the cardiac rhythm, can be detected reliably. This is not only crucial for a safe diagnosis at the beginning of the therapy for cardiac arrhythmia- it may also be necessary to keep a close eye on the heart during treatment.

For example an ablation, that is, an intervention for the obliteration of the diseased cardiac muscle tissue, the doctor sees thanks to home monitoring, whether the procedure was successful or if again atrial fibrillation occurs. The physician can control the therapy better, and the patient gains safety.