An artificial intelligence model could improve echocardiogram assessment of the heart’s right ventricle by estimating its size and identifying reduced function. Called DROID-RV, it estimated right ventricular diameter within about 3 millimeters on average. Its predictions were associated with later heart failure, atrial fibrillation and mortality risk.
Study findings published in October in Nature Cardiovascular Research suggest the system could help clinicians flag abnormal exams and identify patients who may need additional testing. It is not currently part of routine echocardiograms, and the results show associations rather than proof that the model causes or prevents outcomes.
How DROID-RV Could Improve Echocardiogram Assessment
Researchers trained DROID-RV using more than 380,000 echocardiogram video clips from around 17,000 patients. They tested it on 9,100 primary care patients in their health system and 831 patients from a Hungarian hospital.
They compared predictions with health records to assess right ventricular function and associations with later outcomes. The model identified reduced function with what researchers described as high accuracy. Its estimate of right ventricular diameter was within about 3 millimeters on average.
Lead study authors Mostafa Al-Alusi, MD, MS, and Patrick Ellinor, MD, PhD, said preliminary measurements and alerts could help echocardiographers read exams faster and focus on abnormal results. The tool could also help identify patients who may benefit from more expensive MRI scans, they said.
Cardiac MRI is costly and not widely available, according to Paul Drury, MD, a cardiologist interviewed by Medical News Today. Echocardiography is less expensive and widely available, but Drury said the right ventricle can be difficult to image because of its crescent-like shape and position in the chest.
Predictions Were Associated With Later Heart Outcomes
DROID-RV predictions linked larger right ventricular size or worse function with heart failure and atrial fibrillation. Researchers also reported an association with mortality. Patients whose right ventricles were predicted to be dilated had twice the five-year mortality risk of those with predicted normal size, according to the study authors.
After accounting for age, sex and left ventricular function, researchers continued to find an association between predictions and future cardiovascular disease. The study does not establish that a larger ventricle or a model prediction caused a later outcome. Drury noted that chronic conditions can be associated with right ventricular enlargement or failure.
Drury called 3-millimeter accuracy excellent for measuring cardiac chamber sizes on an echocardiogram and said it was acceptable for clinical practice. He said he would want to see findings reproduced across cardiology sites, echocardiographers and patients, including on poor-quality images, against cardiac MRI.
Clinical Use Still Needs Further Evaluation
Study authors said tools like DROID-RV may eventually predict future cardiovascular disease as well as recreate echocardiogram measurements. Drury said automated measurements could alert cardiologists to potential right ventricular issues and lead to further evaluation. These are possible future benefits, not outcomes established by the study.
Other research on AI systems communicating with patients and devices includes a Penn patient-device study. It is separate from DROID-RV, but offers a related example of AI research involving health information and clinical settings.
Testing is also a concern for AI-enabled medical devices. The FDA’s proposed device testing approach addresses generative AI medical devices; the regulatory discussion does not assess DROID-RV.
Hospitals are also investing in digital systems and clinical research, as described in coverage of AIIMS Jodhpur’s health technology infrastructure. That effort is distinct from the DROID-RV study. Wider testing against cardiac MRI and across clinics remains a key step before routine use can be assessed.
