Tech News

Artificial intelligence can detect a heart attack with a voice recording

It is said that the face is the mirror of the soul. And it seems that the voice is the mirror of the heart. The latter concludes from research recently presented at the Annual Meeting of the American College of Cardiology, which shows that a artificial intelligence (AI) is able to predict future heart disease with just a voice recording.

[article_mb_code]

Both in the presentation of the event and in a study published in the journal Mayo Clinic Proceedings, it is highlighted the poorly understood relationship between the voice and the obstruction of the coronary arteries.

Logically, this is something that human ears cannot detect. But for a well-trained AI, it becomes a piece of cake. In fact, In the first tests carried out after training, the results were very good.

[article_mb_code]

As the study authors explained in a press release, this artificial intelligence algorithm could be very useful in telemedicine. It is not to replace human doctors. But yes, to give them the tools that make their work easier.

Also, it means more comfort for the patient, given that voice recording is completely non-invasive and non-painful. Logically, the results must be compared with traditional tests, but they can at least provide a good starting point.

[article_mb_code]

How this AI works to detect abnormalities in the heart

This AI algorithm was trained on 10,000 samples of voice recordings collected in Israel. Some belong to patients with good heart health and others belong to people with heart disease.

The machine is responsible for analyzing various audio parameters, such as frequency, amplitude, pitch and rhythm. He then established a relationship between certain values ​​of these parameters and the presence of heart disease.

[article_mb_code]

After training, artificial intelligence was tested with a new group of 108 patients. They all went to the hospital for an angiogram. That is, an X-ray technique that analyzes the arteries of the heart. In addition to the test A voice recording was made in three different cases:

– In the first, they were made to read a specific text.

– In the second, they were asked to speak freely about a positive experience.

– And finally, in the third part, they had to talk about a negative experience.

The three recordings were made with an application called Vocalis Health and analyzed with an artificial intelligence algorithm. The AI ​​scores the voice of each patient with a system from -1 to +1, depending on whether they have a high or low number of biomarkers related to heart disease. After that, each patient was followed up for two years.

The results of the voice recordings

During the two years of the study, 58.3% patients were classified by artificial intelligence with a high number of biomarkers in their voice recordings, and had to go to hospital for chest pain or acute coronary syndrome.

For those classified with few biomarkers, the rate was much lower, at 30.6%. Also, those with high scores were more likely to develop coronary heart disease when the angiography was done later.

The term refers to a group of conditions that include angina attacks, unstable angina attacks, acute myocardial infarction and sudden cardiac death.

Finally, there is a clear correlation between seemingly indistinguishable parameters in voice recordings and heart health. The researchers don’t know what could be causing this. However, they suggest that it may be related to the autonomic nervous system.

And it is that, the branch of the nervous system is in charge of regulating the unconscious functions of the human body, such as the heart rate. It also controls some functions of the larynx in relation to the voice, so there may be a relationship. If one part is not working properly, the other part may also be affected.

It should be noted that this study was conducted primarily in two countries. The AI ​​has been trained in Israel and tested in United States, with voice recordings in English. So it is not yet clear if it will work with other languages ​​or other platforms.

This will be one of the next steps for scientists, who realize that they still have a long way to go to bring their artificial intelligence algorithms to clinics. However, they have made it clear that the technology can do amazing things in the service of medicine. It has been seen many times, but it never ceases to amaze.

Tags
Back to top button