Eko Analysis Software

The Eko Analysis Software is a cloud-based API utilizing signal processing and neural network algorithms to analyze synchronized ECG and heart sound data from adult patients. Its purpose is to support physicians by detecting suspected murmurs, atrial fibrillation, or normal sinus rhythm, and calculating cardiac intervals such as heart rate, QRS duration, and EMAT, with its performance validated through retrospective clinical data analysis.

Eko Analysis Software (EAS) is a cloud-based software API designed to assist physicians in evaluating adult patients' cardiac health by analyzing synchronized ECG and phonocardiogram (PCG) data. The software employs signal processing and artificial neural network algorithms to interpret these physiological signals, which are acquired using Eko DUO or Eko CORE devices and uploaded via compatible mobile applications. EAS is intended as a decision support tool, providing interpretations that require physician over-read and are not a sole means of diagnosis. Its core functionalities include the detection of suspected heart murmurs from PCG, and from ECG, the detection of atrial fibrillation or normal sinus rhythm, alongside calculations of heart rate, QRS duration, and the electromechanical activation time (EMAT). Clinical performance validation using retrospective analyses on public and proprietary datasets demonstrated specific efficacy metrics: rhythm detection (atrial fibrillation/normal sinus) achieved 100% sensitivity and 96.2% specificity on classifiable ECGs (74.3% of recordings); murmur detection yielded 87.6% sensitivity and 87.8% specificity. Heart rate calculation error was 1.14%, with high sensitivity and specificity for bradycardia (Sens 94.7%, Spec 99.7%) and tachycardia (Sens 93.6%, Spec 99.0%). QRS duration calculation had an absolute mean error of 9.25ms, and EMAT calculation an absolute error of 1.68%. The software explicitly does not distinguish between different types of murmurs (e.g., innocent vs. pathological) and does not identify arrhythmias other than atrial fibrillation and normal sinus rhythm, emphasizing its role as an adjunctive tool dependent on physician oversight.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. How does the Eko Analysis Software's murmur detection algorithm performance (Sensitivity 87.6%, Specificity 87.8%) compare to expert human auscultation in the same datasets, and what are the clinical implications of its false positive/negative rates for screening purposes?; Given that the rhythm detection algorithm successfully classified 74.3% of ECG recordings as either Normal Sinus Rhythm or Atrial Fibrillation, what were the primary characteristics or quality issues of the remaining 25.7% of recordings that rendered them unclassifiable, and what strategies are in place to improve this classification rate?; The software does not distinguish between different kinds of murmurs. What is the potential roadmap, if any, for developing capabilities to differentiate between innocent and pathological murmurs, and what are the anticipated challenges in achieving clinically reliable differentiation?; Considering the EMAT interval algorithm's reliance on Q peak detection and S1 envelope detection, how robust is its performance in patient populations with significant arrhythmias (e.g., frequent PVCs, non-AFib tachycardias) or conduction abnormalities that might distort these ECG and PCG landmarks?; The QRS duration is calculated from a single-channel ECG. How does the accuracy of this measurement (Absolute Mean Error 9.25ms) compare to QRS duration derived from a standard 12-lead ECG, particularly in patients with known bundle branch blocks or ventricular hypertrophy where precise QRS measurement is critical?