Artificial intelligence, increasingly, is introduced in medicine. Be like A tool to determine treatment, indicate diseases or design drugs. But also as a predictor of different pathologies.
Now, A new AI model proves to be much better than humans to identify patients likely to suffer cardiac arrest. The key lies in the system’s ability to analyze infrautilized cardiac images for a long time, together with a complete spectrum of medical records, to reveal information about the patient’s heart health.
The work, led by scientists from Johns Hopkins University, PHe hated saving many lives and also avoiding many unnecessary medical interventions, such as defibrillators implementation.
“Currently, we have patients who die in the flower of life due to lack of protection, and others who support defibrillators for the rest of their lives without obtaining any benefit,” says Natalia Trayanova, leader of the study and specialized in the use of artificial intelligence in cardiology -. We can predict with great precision if a patient has a very high risk of sudden cardiac death or not. ”
The findings have been published in Nature Cardiovascular Research. Hypertrophic myocardiopathy is one of the most common hereditary heart disease, than It affects one in 200 to 500 people worldwide And it is one of the main causes of sudden cardiac death in young people and athletes.
Many patients with hypertrophic myocardiopathy lead a normal life, but A percentage presents a significantly higher risk of sudden cardiac death. It has been almost impossible for doctors to determine who these patients are.
Current clinical guidelines used to identify patients with the highest risk of mortal infarcts have A probability of approximately 50 % of identifying correct patients.The team model significantly exceeded clinical guides in all demographic groups.
Multimodal AI for the stratification of the risk of ventricular arrhythmias (maars) Predicted the risk of sudden cardiac death of each patient through the analysis of various medical data and records And, for the first time, exploring all the information contained in the magnetic resonance images with contrast to the patient’s heart.
People with hypertrophic cardiomyopathy develop fibrosis, or heart healing, and it is this healing that increases their risk of sudden cardiac death. Yes ok Doctors have not been able to interpret the unprocessed magnetic resonance imagesthe AI model focused directly on critical healing patterns.
“Deep learning has not been used in those images – adds Trayanova -. We can extract this hidden information in the images that are not normally taken into account. ”
The team tested the model with real patients treated with traditional clinical guides and, compared to these that showed being precise in 50% of cases, the AI model had An 89 % precision in all patients and, crucially, 93 % in people from 40 to 60 yearsthe population with the greatest risk of sudden cardiac death among patients with hypertrophic myocardiopathy.
The AI model too You can describe the high risk of patientsso that doctors can design a custom medical plan for their specific needs.
“Our study shows that the model of IA significantly improves our ability to predict people with greater risk Compared to our current algorithms and, therefore, has the power to transform clinical attention, ”says co -author Jonathan Crispin, a cardiologist at Johns Hopkins.
In 2022, the Trayanova team created a different multimodal AI model than offered a personalized evaluation of survival for patients with heart attacks and predicted if someone was going to die for cardiac arrest and when.
The next goal is to test the new model in more patients and Expand the new algorithm for use with other types of heart disease, such as cardiac sarcoidosis and arrhythmogenic myocardiopathy of the right ventricle.