14x Fewer Missed Diagnoses of critical arrhythmias. A clear path to direct-to-physician reporting of ambulatory ECG results, reveals a recent Nature Medicine publication.
Warsaw, Poland – [10.02.2024] -Medicalgorithmics, a pioneer in AI-driven cardiac diagnostics, proudly announces the publication of groundbreaking independent research in Nature Medicine demonstrating the superior performance of its DeepRhythmAI (DRAI) model in diagnosing cardiac arrhythmias.
The development of DeepRhythmAI marks a paradigm shift in ECG analysis, promising faster diagnoses, improved patient outcomes, and streamlined workflows for healthcare providers. The global shortage of ECG technicians and increasing patient wait times create a critical need for innovative solutions in heart care. Analyzing human heartbeats for potentially life-threatening arrhythmias is a complex and often error-prone process. DRAI addresses this challenge head-on, offering a solution that substantially outperforms traditional analysis by human technicians.
The “DRAI MARTINI” study was led by Dr. Linda S. Johnson, Associate Professor at Lund University, Sweden, and Dr. Jeffrey S. Healey, Professor of Cardiology at McMaster University, Canada, and tested DRAI on recordings that had been analyzed by licensed ECG technicians in clinical practice in the U.S. This is the world’s largest study of AI for cardiac arrhythmia diagnosis to date, including >200,000 days of ECG data from >14,000 adult patients. Both the AI and the technician analyses were compared to diagnoses by panels of three cardiologists independently reviewed 5,000 randomly selected arrhythmias to establish a definitive “gold standard” for diagnosis, ensuring an unbiased evaluation. In total 17 panels of three cardiologists each, from leading institutions world-wide, participated in the study, providing extremely high quality beat-to-beat diagnoses of >5,000 rhythm events.
“We tested what would happen if technicians were replaced by DeepRhythmAI, for arrhythmia diagnosis on ambulatory ECGs. The study was rigorous – more than 200,000 days of ECG were used, and 50 independent experts were involved in providing an extremely high-quality benchmark to which both the AI and the technicians were compared. Frankly, the results surprised us. The AI network has a major advantage in terms of patient safety – with 14 times fewer missed diagnoses of critical arrhythmias – a result of a much better sensitivity. The implications for patient care are large. DeepRhythmAI is good enough to be used to report directly to physicians, improving access to care, shortening time to diagnosis, and hopefully improving patient outcomes as well”, says Dr. Linda S. Johnson, Associate Professor at Lund University
The results are striking. DRAI demonstrated substantially higher sensitivity for detection of critical arrhythmias, including atrial fibrillation, complete heart block, pauses, supraventricular and ventricular tachycardia. The rate of false negative findings with DRAI was low, 3.2 per 1000 patients, compared to 44.3 per 1000 patients with technician analysis – reducing by 14 times the rate of missed diagnoses compared to technician analysis. This dramatic improvement in accuracy underscores DRAI’s superiority in terms of patient safety. This translates to an impressive 99.9% negative predictive value for DRAI as a direct-to-physician reporting tool to rule out the presence of critical arrhythmias on an ambulatory ECG.
DRAI’s remarkable performance is powered by both CNN and transformer models and an unparalleled training dataset of 250 billion annotated heartbeats. This vast library enables the AI to learn and identify subtle patterns indicative of arrhythmias with exceptional efficiency. DRAI, as a 100% cloud-based solution, has the ability to provide direct-to-physician reporting of ambulatory ECG results and can streamline clinical workflows by up to 65%, leading to lower healthcare costs, increased access to monitoring, and faster diagnoses, regardless of the number of consecutive sessions.
The results of this independent study validate what we at Medicalgorithmics have been working towards—leveraging AI to transform cardiac diagnostics. DeepRhythmAI’s ability to reduce missed arrhythmia diagnoses by 14 times compared to human analysis is a breakthrough that directly impacts patient safety and clinical efficiency. With an accuracy level that ensures doctors can confidently rule out critical arrhythmias, DRAI is setting a new standard for direct-to-physician ECG reporting, reducing workloads for healthcare providers while ensuring faster, more reliable diagnoses. This is the future of cardiac care—more precise, scalable, and accessible to patients worldwide,” says Przemyslaw Tadla, CTO of Medicalgorithmics S.A.
By leveraging the power of the only FDA-cleared AI-powered integration solution with third-party software, DRAI offers a solution that surpasses traditional healthcare standards, significantly reducing the risk of missed diagnoses and improving the speed and accuracy of cardiac care. This groundbreaking technology represents a significant advancement in the fight against heart disease.
Link to article: https://www.nature.com/articles/s41591-025-03516-x