The DeepRhythmAI (DRAI) algorithm, developed by the Polish medtech company Medicalgorithmics, significantly outperforms ECG technicians in detecting arrhythmias relevant to clinical indications for ambulatory ECG monitoring. These new findings from the DRAI MARTINI study show a 75 to114-fold reduction in the risk of missing clinically meaningful diagnoses. These breakthrough results were presented on November 8 during the American Heart Association (AHA) 2026 meeting.
The DRAI MARTINI study is a large-scale evaluation of the DeepRhythmAI algorithm for autonomous analysis of ambulatory ECGs that involved 14,606 patients and >200,000 days of continuous ambulatory ECG monitoring, and compared DeepRhythmAI only analysis to ECG technicians in real-world clinical practice. Building on prior results showing DeepRhythmAI’s superior sensitivity (14-fold) for detection of critical arrhythmias, this analysis differentiates between findings that the patient was monitored to detect, and that therefore have direct clinical relevance, and incidental findings.
For patients monitored for suspicion of tachyarrhythmias—such as palpitations, atrial fibrillation, or stroke—DeepRhythmAI had a sensitivity of 99.5% (95% CI: 98.8-100.0%) for indication-relevant arrhythmias while the ECG technicians sensitivity was 67.9% (95% CI: 62.9-71.8%). With DeepRhythmAI only analysis, the risk of a missed indication-relevant diagnosis in this group of patients was 75 times lower than with technician analysis.
In patients monitored for syncope, dizziness, or bradycardia, DeepRhythmAI’s sensitivity for relevant findings was 99.4%, compared to 54.6% (95% CI: 45.3-61.9%) for technicians. This translates to a 114 times lower risk of a missed indication-relevant diagnosis.
At the same time, DeepRhythmAI also excelled in identifying incidental findings, with sensitivities of 98.5% (95% CI: 96.3-100%) in the tachyarrhythmia group and 99.7% (95% CI: 99.3-99.9%) in the syncope/bradycardia group, far surpassing technicians’ rates of 49.2% and 77.2%, respectively.
“These new results show that DeepRhythmAI leads to substantially increased detection rates of arrhythmias relevant to patient care, while also delivering comprehensive detection of incidental issues that could impact long-term patient care”, said the lead author Dr. Linda Johnson (Lund University, Sweden).
“The presented study results once again confirm that our advanced algorithms significantly enhance both the speed and quality of diagnosis. Greater effectiveness in patient monitoring translates directly into improved patient care while simultaneously reducing healthcare costs. The technologies developed during the creation of DeepRhythmAI form the foundation for future products that we will be releasing to further advance cardiac diagnostics,” commented Przemysław Tadla, CTO of Medicalgorithmics.
Medicalgorithmics is a global provider of state-of-the-art software solutions for non-invasive cardiac diagnostics. The company develops proprietary artificial intelligence algorithms for the analysis of ECG signals and cardiac computed tomography (CT) imaging. Its latest medtech innovations – recently certified by the FDA – rank among the most advanced AI-driven diagnostic tools worldwide, offering superior accuracy (14 times fewer missed diagnoses compared to traditional analysis performed by qualified ECG technicians), faster performance, and greater user convenience.
The Polish medtech company is also developing a breakthrough technology known as VCAST – AI-powered software for segmentation and reconstruction of coronary vessels based on widely available and cost-effective cardiac CT data. This enables non-invasive diagnosis of coronary artery disease (CAD) – one of the leading causes of myocardial infarction. Compared with invasive and costly procedures such as coronary angiography (exceeding USD 10,000 per case), VCAST represents a faster, more affordable, and accessible alternative. Currently, only a few companies worldwide (including HeartFlow Inc. and Cleerly Inc.) offer comparable solutions. The VCAST technology developed by Medicalgorithmics already holds CE certification and is now in the process of commercialization.