“Tell me what your voice is and I will tell you if the blood flows smoothly in the arteries or if there are obstructions that can prevent it from flowing, exposing you to the risk of cardiovascular diseases such as heart attack or stroke”. In the future, thanks to technology, even those who are at a distance from a specialized center can be monitored over time thanks to a simple application that records sound impulses and sends them to an Artificial Intelligence system.
This, as a sort of “laboratory”, will examine the characteristics of the voice and can therefore indicate which subjects will be most in danger of suffering a stroke. Science fiction? No. at least according to a study – we are only at the beginning – presented at the Congress ofAmerican College of Cardiologywhich shows the predictive capabilities of the system, currently in the advanced research phase.
The vocal biomarkers of stroke and heart attack
The research, which he sees as the lead author Jaskanwal Deep Singh Sara from the Mayo Clinichas shown that those with a specific elevated vocal marker have a 2.6 times higher risk of experiencing future coronary artery disease outcomes and an even tripled probability of finding themselves with an accumulation of plaques along the vessels confirmed with clinical tests, obviously in comparison to those who do not have “dangerous” vocal characteristics.
Heart door
A blood sample will reveal who is at risk of heart attack (even with normal values)
at Federico Mereta
According to experts, who also hold back on the use of these tools in the clinic pending further evaluations, the simplicity of the test (just a simple registration with the smartphone) could become a very useful weapon to control the situation. The research examined just over a hundred people who had undergone coronary artery screening.
Then everyone was asked to record three different voice messages of about half a minute: in the first you had to read a short text, in the second and in the third a positive and a negative experience was summarized.
How the algorithm that identifies speech parameters works
Thanks to an Artificial Intelligence algorithm developed in Israel on the basis of over 10,000 recordings, we then proceeded to analyze about eighty vocal parameters, from frequency to amplitude, tone and cadence.
Starting from a previous experimental analysis he had identified six scorers correlated with coronary heart disease, the analysis was then carried out on the basis of the scores obtained. In the two years following the check-up, those who had high scores of this vocal “score” in just under six out of ten cases (58.3%) entered the hospital for chest pain or clear signs of acute coronary syndrome, compared to 30. , 6% observed in subjects with low voice biomarker score. How to explain this situation?
The hypothesis: the role of the nervous system
On the scientific front, a role of the autonomic nervous system is hypothesized, which in some way affects both the voice and heart and arterial health parameters, such as heart rate or blood pressure.
This link could help explain the usefulness of the remote voice control in monitoring people at risk of heart attack.
Heart door
Stress increases the risk of heart attack: by two and a half times
at Federico Mereta
“Speech analyzes for the diagnosis of various pathologies are not new – he explains Filippo Molinariordinary of Bioengineering at the Polytechnic of Turin. In particular, neurodegenerative diseases (in particular Parkinson’s) have been analyzed in the past. In that case, the alteration of the sound is directly correlated to an effect of the pathology on the complex muscular compartment that regulates the human voice. In this study the value of vocal analysis is greatly extended to a class of pathologies (cardiovascular ones) which do not involve, at least in the current state of knowledge, a direct involvement of the nerve centers or muscles dedicated to phonation “.
Biological signals, a mine of information still little exploited
In short, we are at the very beginning. “The study, still very preliminary in its results and on a still rather small number of patients, shows that even in the most well-known and studied biological signals (from the electroencephalogram to the electrocardiogram to get to the voice, just to name a few examples) there is still a lot of potentially useful information that is not normally considered and that can be extracted thanks to new artificial intelligence techniques – reports Molinari.
From a purely technical point of view – concludes Molinari -, an approach of this type is still very far from being able to configure a new diagnostic tool, but it emphasizes how artificial intelligence can effectively extract further information from data (diagnostic value ), and to provide new elements of study and clinical investigation (basic research value) “.
Heart door
Heart failure, here is the index that calculates the risk of death for each patient
at Federico Mereta
