AI can improve heart rhythm disorder screening, study finds

Credit: Unsplash+

Researchers at Scripps Research have developed an artificial intelligence (AI) model that could significantly enhance the screening for atrial fibrillation (AFib), a condition associated with irregular, rapid heartbeats and heightened risk of stroke and heart failure.

The AI model can identify subtle variations in a person’s normal heartbeat, indicative of AFib risk, which may go undetected by standard screening tests.

The study, published in the journal npj Digital Medicine on December 12, 2023, involved the analysis of data from nearly half a million individuals who wore an electrocardiogram (ECG) patch to record their heart rhythms for two weeks.

This routine screening test aimed to identify AFib and other heart conditions. The AI model then examined this data to identify patterns, distinct from AFib itself, that could differentiate individuals with AFib from those without the condition.

AFib poses a considerable health risk, primarily due to the potential for blood clots forming in the heart, which can lead to strokes.

It is also linked to an increased risk of heart failure and death. Diagnosing AFib is challenging, especially as some individuals may experience only occasional irregular heartbeats or exhibit few symptoms.

Traditionally, cardiologists conduct in-office electrocardiograms with ten electrodes for around ten seconds to diagnose AFib in patients with symptoms such as palpitations, lightheadedness, shortness of breath, or chest pain.

If initial tests do not indicate any issues, patients are advised to wear a simpler, single-electrode wearable ECG patch at home for one or two weeks. However, this approach may not capture occasional AFib episodes.

To address this, the research team collaborated with iRhythm Technologies, the maker of the ZioXT wearable ECG patch, to identify alternative patterns in ECG data from individuals with AFib.

The AI model was developed to analyze data from 459,889 people who wore the ECG patch for two weeks. Remarkably, the model could differentiate those who developed AFib from those who did not, even when compared to manual models incorporating known risk factors.

The AI model proved to be more accurate in predicting AFib risk, outperforming traditional ECG features and risk factors.

Importantly, it demonstrated accuracy across different age groups, including older individuals at higher risk and younger individuals typically excluded from general AFib screening.

While the AI model is not intended for AFib diagnosis, it represents a significant step towards designing a screening test for those at elevated AFib risk or showing symptoms. Patients may only need to wear an ECG patch for one day to determine if extended testing is required.

Alternatively, the model can analyze one- or two-week ECG data to identify patients who should undergo repeat testing, even if no AFib episodes occurred during that timeframe.

The researchers plan to conduct prospective studies and integrate additional data sources, such as electronic medical records, to further enhance the AI model’s accuracy and utility in clinical practice.

If you care about heart disease, please read studies about a big cause of heart failure, and common blood test could advance heart failure treatment.

For more information about heart health, please see recent studies about a new way to repair human heart, and results showing drinking coffee may help reduce heart failure risk.

The research findings can be found in npj Digital Medicine.

Copyright © 2023 Knowridge Science Report. All rights reserved.