This is a plain English summary of an original research article. The views expressed are those of the author(s) and reviewer(s) at the time of publication.
When heart muscles thicken (hypertrophy), they become less able to pump blood around the body. New research shows that artificial intelligence (AI) techniques can predict thickening of the heart muscle from electrocardiogram (ECG) readings and health data, such as blood pressure.
Researchers developed an AI tool that uses this routine information to detect hypertrophy. Among 37,500 people, it correctly identified 8 out of 10 of those with thickening of the heart muscle on the left side of the heart.
With further, real-world testing, the tool could help clinicians identify people who would benefit from a specialised heart scan (MRI scan), researchers say. This could avoid unnecessary tests for people unlikely to have the condition, and free up clinicians’ time.
For more information on diseases of the heart muscle, see the NHS website.
The issue: can AI use ECG readings to predict heart muscle thickening?
Cardiomyopathy is a general term for diseases that cause heart muscles to thicken, become stiff and less able to pump blood. When this occurs on the left side of the heart, it is known as left ventricular hypertrophy. People with this condition have an increased chance of heart attack and death.
Specialised magnetic resonance imaging (MRI) is the gold standard for diagnosis, but scanners are expensive and not available in every hospital. Only specialists can perform MRIs. An ECG is simpler and less expensive and can indicate that someone is at risk. But they are not as accurate as MRI scans.
AI has shown promise in early research at predicting left ventricular hypertrophy using ECG readings. This study assessed the accuracy of an AI tool that used ECG readings and other health data.
What’s new?
The study was based on the health records of 37,500 people in the UK Biobank (data collected between 2006 – 2010). 578 people had left ventricular hypertrophy and 36,956 did not. Their average age was 64.
As part of the UK Biobank’s data collection, people underwent an ECG and an MRI scan of their heart. The database included information on their cholesterol levels, blood pressure, and smoking and alcohol status. The AI tool used this data to distinguish between people with and without left ventricular hypertrophy.
The AI tool:
- correctly identified left ventricular hypertrophy 8 out of 10 times (80% accuracy)
- missed few people with the condition (sensitivity of 72%) and rarely identified it when it was not there (specificity of 81%)
- was more accurate when it used both ECG readings and health data, compared with health data alone.
The researchers found that specific ECG and blood pressure measurements were the best predictors of left ventricular hypertrophy.
Why is this important?
An AI tool differentiated between people with and without heart muscle thickening using ECG readings and health data. The researchers say the tool could be used to identify those most in need of an MRI scan. This would spare people unlikely to have the condition from having an unnecessary MRI. ECGs can be performed by non-specialists with relatively simple equipment; this could free up specialists to spend more time on other tasks.
The researchers caution that the tool is not ready to be used in clinical practice. In addition, the people represented in the UK Biobank are mostly white and typically healthier than the general population. Further testing in populations more likely to have cardiomyopathy, and in real-world practice, are needed.
What’s next?
In future, the tool could be used to predict early disease and disease progression. This would provide opportunities to give people treatment and lifestyle advice. Some smartwatches can measure electrical signals from the heart (similar to ECGs); in future, these readings could identify people at risk of heart muscle thickening.
You may be interested to read
This is a summary of: Naderi H, and others. Predicting left ventricular hypertrophy from the 12-lead electrocardiogram in the UK Biobank imaging study using machine learning. European Heart Journal – Digital Health 2023; 4: 316 – 324.
A YouTube video created by the author explaining left ventricular hypertrophy.
A video created by the author summarising the research and next steps
Information on cardiomyopathy from the British Heart Foundation.
Information on taking part in NIHR studies regarding cardiomyopathy.
Funding: This study was supported by the NIHR Bart’s Biomedical Research Centre and the NIHR’s Integrated Academic Training programme and the British Heart Foundation Pat Merriman Clinical Research Training Fellowship to Dr Naderi.
Conflicts of Interest: Steffen Petersen provides consultancy to and owns stock of Cardiovascular Imaging Inc.
Disclaimer: Summaries on NIHR Evidence are not a substitute for professional medical advice. They provide information about research which is funded or supported by the NIHR. Please note that the views expressed are those of the author(s) and reviewer(s) at the time of publication. They do not necessarily reflect the views of the NHS, the NIHR or the Department of Health and Social Care.
NIHR Evidence is covered by the creative commons, CC-BY licence. Written content and infographics may be freely reproduced provided that suitable acknowledgement is made. Note, this licence excludes comments and images made by third parties, audiovisual content, and linked content on other websites.
