Diagnosing the beginning of heart failure could soon require only the electrocardiogram, a simple, quick and economical test, without waiting for the time and costs of a blood sample and an echocardiogram. This is the news that emerged from an Italian-led international study, published in the European Heart Journal.
Researchers have developed an artificial intelligence model capable of reading traces invisible to the human eye within a very common electrocardiogram, capable of indicating with high precision whether there is a high level of NT-proBNT in the patient’s blood, the reference biomarker for the diagnosis of heart failure.
In the study, which saw the collaboration of the University of Calabria, Humanitas University, the Federico II University of Naples, the University and University Health Authority of Trieste, the University of Zurich, Imperial College London, King’s College London and Seoul National University, the researchers “trained” a deep neural network by analyzing 84,895 electrocardiograms and NT-proBNP tests. The algorithm has learned to capture electrical micro-signals that are imperceptible in traditional reading but linked to the presence of the biomarker. Then tested on 679 patients in the centers of Cosenza and Trieste, the algorithm showed an accuracy close to 90% in identifying values at risk.
«Heart failure is one of the great health emergencies of our country – explains Ciro Indolfi, first author of the study and president of the Italian Heart and Circulation Foundation of the Sic (Italian Society of Cardiology) -. In Italy it affects over a million people and around 80,000 new cases every year, especially among the elderly.” Today the diagnosis of heart failure often starts from generic symptoms such as tiredness, breathlessness or swelling in the legs. «To confirm the suspicion, a blood sample with laboratory analysis and a heart ultrasound are needed – declares Gianfranco Sinagra, president of Sic and co-author of the study -. Applying AI to the electrocardiogram allows it to be transformed into a real ‘digital filter’: a tool capable of immediately alerting the doctor when it is really necessary to request laboratory and specialist investigations.