An artificial intelligence algorithm applied to standard electrocardiograms can accurately identify early signs of impaired heart function, according to a study led by researchers at UT Southwestern Medical Center. The findings suggest that AI-enhanced ECG analysis could become a scalable and affordable tool for detecting people at risk of heart failure, particularly in countries where advanced cardiac imaging is limited.
The study focused on patients in Kenya and examined whether AI could detect left ventricular systolic dysfunction, or LVSD. This condition, in which the heart’s left ventricle no longer pumps blood effectively, is a major precursor to heart failure. Detecting LVSD early is important because treatment and monitoring can begin before patients develop severe symptoms. The study was recently published in JAMA Cardiology.
Diagnostic gap
Heart failure is increasing worldwide, but the burden is especially high in sub-Saharan Africa. Patients in the region often develop heart failure at younger ages and experience poorer outcomes, while health systems face shortages of equipment, specialists and diagnostic infrastructure.
Echocardiography is the clinical standard for diagnosing LVSD because it provides ultrasound images of the heart’s structure and function. Yet access to echocardiography remains limited in many lower-resource settings. The equipment is expensive, trained staff are scarce, and large-scale screening is often not feasible.
ECGs, by contrast, are relatively inexpensive, widely available and easier to perform. Traditionally, they measure the heart’s electrical activity rather than directly visualising heart function. By adding AI analysis, researchers aimed to determine whether ECGs could reveal patterns linked to hidden cardiac dysfunction.
Strong performance
The research team recruited nearly 6,000 patients receiving routine care across eight healthcare facilities in Kenya. All participants underwent AI-ECG screening. A subgroup of 1,444 patients also received echocardiograms, allowing researchers to compare the AI results with the gold standard.
Among patients who underwent both tests, the algorithm identified LVSD in 14.1 percent. The AI-ECG showed a negative predictive value of 99.1 percent, meaning that almost all patients classified as not having LVSD were confirmed negative by echocardiography.
The tool also demonstrated high sensitivity, correctly detecting 95.6 percent of patients with LVSD, and specificity of 79.4 percent, correctly identifying many patients without the condition. Positive AI-ECG results were also strongly associated with other signs of adverse cardiac remodelling, including left ventricular hypertrophy and diastolic dysfunction.
These results suggest that AI-ECG could be especially useful as a rule-out tool: helping clinicians identify which patients are unlikely to need echocardiography, while prioritising imaging for those at higher risk.
Scalable cardiovascular prevention
The study’s significance lies not only in its technical performance, but in its potential impact on healthcare delivery. In settings where echocardiography cannot be offered systematically, AI-ECG could provide an intermediate layer of risk detection between basic clinical assessment and specialist cardiac imaging.
According to the researchers, this approach could help close a critical gap in global cardiovascular care by making early heart failure screening more accessible. It may also support better triage, earlier treatment and more efficient use of scarce diagnostic resources.
However, AI-ECG should not be seen as a replacement for echocardiography. Instead, it could function as a practical screening tool that identifies patients who need further evaluation. Broader implementation will require validation in additional populations, integration into clinical workflows, and attention to data quality, equity and regulatory oversight.
For health systems under pressure, the study points to a powerful principle: when AI is embedded in simple, affordable diagnostics, advanced cardiovascular prevention may become available far beyond specialist centres.
AI-ECG
Last year, an AI-model, developed at Mayo Clinic, enabled routine electrocardiograms (ECGs) to be used to detect advanced chronic liver disease at an early stage. By analysing subtle changes in cardiac electrical signals, the system identified approximately twice as many cases compared to standard clinical practice, often in patients without symptoms.
The approach leverages the physiological link between liver dysfunction and cardiovascular changes, which are difficult for clinicians to detect but recognizable through AI. Trained on data from over 11,000 patients, the model was validated using imaging and blood tests. A clinical trial involving 248 clinicians showed the technology can be integrated into routine care, enabling earlier intervention and potentially improving patient outcomes.
