In Nature Digital Medicine, a study (with an “Italian” signature) on over 450 thousand patients. Early diagnosis of this form of arrhythmia can reduce the risk of stroke and heart failure
A new artificial intelligence (AI) model designed by scientists at Scripps Research in La Jolla (San Diego, California) could help doctors better detect atrial fibrillation (AF), the most common form of arrhythmia, a heart disorder in patients. not lethal in itself but which can cause serious complications starting from cerebral stroke: approximately 30% of cases are associated with atrial fibrillation. It is estimated that approximately 2% of the population suffers from itthe most affected are the elderly: 1 in 10 after the age of 80.
The AI model detects small changes in a person's normal heartbeat that are indicators of atrial fibrillation risk, which standard screening tests cannot detect. The study, published on npj Digital Medicinebears the signature of two Italians: Matteo Gadaletaas first author, e Giorgio Quer, as senior leading author , both from Scripps Research Translational Institute. The researchers used data on almost half a million people who had each worn an electrocardiogram (ECG) patch to record their heart rhythm for two weeks, a routine screening test for atrial fibrillation and other heart conditions.
Risk stratification
An artificial intelligence algorithm then analyzed this data to find patterns, other than atrial fibrillation itself, that distinguished people with atrial fibrillation from those without. «The model integrates the ECG with demographic and heart rhythm characteristics for predict the presence of AF in the following weeks – explains Giorgio Quer, director of Artificial Intelligence at the Scripps Research Translational Institute and university researcher in Digital Medicine at Scripps Research -. Looking at a 1-day ECG (without fibrillation), achieves an accuracy (AUC) of 0.80, significantly outperforming predictions obtained with standard techniques (AUC of 0.67). This algorithm allows you to stratify the risk of individuals who have undergone a first screeningsuggesting prolonged monitoring to those most at risk, and allowing AF to be identified and treatment to begin as soon as possible.”
«In the long term, this can help direct the right resources to the right people and potentially to reduce the incidence of stroke and heart failure
».
Difficult diagnosis
The irregular heartbeat due to atrial fibrillation increases the chance of blood clots forming in the heart, which can then travel to the brain, blocking an artery and causing a stroke. Atrial fibrillation is also associated with an increased risk of heart failure or death. To prevent these complications in people with known atrial fibrillation, doctors often prescribe anticoagulants — drugs that prevent blood clots from forming — as well as other lifestyle and medical therapies.
However, Diagnosing atrial fibrillation can be tricky because many with this condition have only occasional attacks or few symptoms. In some people, atrial fibrillation causes heart palpitations (so-called palpitations), dizziness, shortness of breath, and chest pain. For patients experiencing these symptoms, cardiologists usually record heart rhythms for about ten seconds, using a ten-electrode ECG.
The “patch” for the electrocardiogram
If nothing abnormal appears immediately, specialists recommend a Continuous monitoring at home for one to two weeks with an ECG patch simpler and wearable that has only one electrode. But even over the course of two weeks, people with very occasional atrial fibrillation could do not experience an episode detected by this device. That's why Quer, in collaboration with the manufacturer of a wearable electrocardiogram patch, set out to find other patterns in the ECG data of people with atrial fibrillation.
«We believe that the electrical activity of the heart is slightly different for people who suffer from atrial fibrillation, but the differences are so subtle that cardiologists cannot identify these differences,” says Quer. The team developed a machine learning model to analyze the data the company had collected out of 459,889 people who wore the ECG patch at the company's home for two weeks.
The data used
From each ECG, the model used data from a day that reported no episodes of atrial fibrillation, but was still able to distinguish people who subsequently had AF from those who did not. Even when researchers have integrated all known atrial fibrillation risk factors into its manual models, including demographic data and ECG measures such as variability between different heartbeats, the machine learning model was more accurate at predicting the risk of atrial fibrillation. “There was a gap between what we could achieve with any known ECG feature and what the model could achieve,” says Matteo Gadaleta, scientific collaborator at the Translational Institute. “It was definitely better.”
A “useful” model also for people under 55
The model was confirmed accurate both with regard to the older populationwho is at higher risk of atrial fibrillation, and for people under the age of 55, which present a much lower risk and are usually excluded from general atrial fibrillation screening. Although the model is not intended for the diagnosis of atrial fibrillation, the authors consider it a first step towards designing a screening test for people at increased risk of atrial fibrillation or showing symptoms. That way, they might wear an ECG patch for just one day to determine whether longer tests are advisable. Alternatively, the model could analyze one or two weeks' worth of ECG data to identify patients who, even without any atrial fibrillation in that time interval, should undergo a repeat test.
“Patients with frequent episodes of atrial fibrillation can be easily identified with an ECG recorded over at least a week,” says Quer. «But this artificial intelligence model it could really help people who have very rare episodes of atrial fibrillation, but who could still benefit from diagnosis and intervention.” Quer and his colleagues hope to plan a prospective studyas well as integrate other data sources, such as electronic health records, into their models to improve them even further.
Corriere della Sera is also on Whatsapp. It's enough click here to subscribe to the channel and always be updated.
December 21, 2023 (changed December 21, 2023 | 08:10)
© ALL RIGHTS RESERVED
#Atrial #fibrillation #algorithm #understand #sooner #risk #treated