Cardisiography: How AI is revolutionizing heart health

Cardiovascular diseases are still the number 1 cause of death in Germany. It accounts for 40 percent of all deaths. Heart diseases are also so insidious because they often go undetected for a long time and progress without symptoms – the heart attack is usually the first and unfortunately often fatal symptom of the disease. Cardiovascular diseases, especially in young people, can often be treated very well if they are recognized in good time.
The most widespread preventive method in heart medicine is the (stress) ECG. This method is more than a hundred years old and therefore well established. The dilemma here is that an ECG is of little value – many heart diseases often go undetected with this standard method. Other methods, such as the cardiac catheter, are much more precise, but their use is also associated with significantly higher effort and costs as well as the disproportionately higher burden for the patient. A new method has now set itself the task of overcoming this dilemma: similar to an ECG, it is uncomplicated, safe and inexpensive to use and yet precise with a high level of informative value. What helps her: artificial intelligence (AI).
This new method is based on the process of vector cardiography. Similar to an ECG, the electrical activity of the heart is recorded – but with one crucial difference: where the ECG only records two dimensions, an additional electrode on the back in vector cardiography ensures data is recorded in three dimensions. This is how the voltage course of three vectors emanating from the heart is recorded:

The P loop indicates the spread of excitation in the atrium

The QRS loop maps the spread of excitation in the ventricle (the cavities of the heart).

The T loop results from the regression of this excitation

This three-dimensional method enables significantly more precise analyzes than a classic ECG, but has almost never been used in the past because the interpretation of a vector cardiogram is very complex.
This is where the algorithm comes into play: in cardiography (also called 3D vector ECG), a further development of vector cardiography by the German startup Cardisio, the artificial intelligence analyzes the recorded heart data. The results are processed within a few minutes and are comprehensible and interpretable for the doctors. The cloud-based algorithm is capable of processing millions of features per acquisition and graphically displaying the heart's condition in a simple PDF. He can also make a clinical recommendation. In order for the AI to be able to do this, it was trained, among other things, with the help of study results obtained from X-ray examinations of the coronary arteries, known as coronary angiography.
And the algorithm continues to be trained, improving its performance and the accuracy of its predictions. For this, the experts at Cardisio rely on what is known as supervised machine learning. Supervised machine learning is about teaching the AI to correctly interpret and classify certain data sets. The aim is for the AI to be able to make predictions for the future as precisely as possible. To do this, it is fed with data that it can correctly assign based on the numerous parameters.
In the case of the Cardisio algorithm, the following applies: It should correctly interpret the values of the heart obtained during cardisiography in order to be able to make the most accurate predictions possible about the condition of the patient's heart. These predictions are already very precise today. In a clinical study, cardiography showed a sensitivity of 97 percent in male and 90 percent in female subjects with coronary artery disease.
The sensitivity indicates the proportion of patients who are correctly identified as such by this method.
The stress ECG, for example, only has a sensitivity of 50 percent, with the patients being exposed to a significantly higher risk.
Supervised machine learning is understood as an alternative to deep learning. With the latter, the AI is fed with such an amount of data that the complexity increases enormously and the way the results come about can ultimately no longer be explained. With supervised machine learning, on the other hand, the AI is only fed with specifically collected data. As a result, the predictions of the algorithm remain comprehensible and explainable. This is the only way doctors can maintain control and treat patients correctly.
For heart care, AI offers unprecedented opportunities and solves many problems. For example, thanks to the 3D recording and the subsequent evaluation by the algorithm, ischemic heart diseases, i.e. circulatory disorders in the heart, can now also be detected. With the classic ECG, this was only possible to a very limited extent or only at a very advanced stage. Structural heart diseases are also detected with this screening method. In addition, a previously unknown level of accuracy is achieved with the help of artificial intelligence.
Similar precision values have so far only been achieved by extremely complex and cost-intensive procedures, such as cardiac MRT or cardiac catheters. However, cardisiography remains non-invasive and therefore risk-free. Thanks to artificial intelligence, heart diseases can be detected quickly and precisely in the future and treated in a targeted manner before they endanger life.
Cardisiography thus fits well with the objective of modern medicine, which is to focus more on the prevention of diseases. Many people are becoming aware that they should continuously do something for their health in order to stay fit and resilient for as long as possible. For heart health, there has recently been a good contact point: the website. (mp/hk)

Related Posts

Leave a Reply

%d bloggers like this: