The Cardiologs AI-based ECG analysis platform detected all arrhythmias identified on cardiologist-reviewed conventional telemetry, corresponding to a sensitivity of 100%.
Observational (n=30)
Blinded reviewers
Does an AI-based ECG analysis platform accurately detect arrhythmias compared to cardiologist-reviewed conventional telemetry in cardiac intensive care patients?
An AI-based ECG analysis platform demonstrated 100% sensitivity for arrhythmia detection compared to cardiologist review in cardiac ICU patients, highlighting its potential for automated ECG interpretation.
Abstract Background Continuous ECG monitoring via telemetry is a diagnostic cornerstone for inpatients' cardiological care. Philips IntelliVue, supported by the ST/AR arrhythmia algorithm and the PIC iX platform, provides real-time rhythm monitoring and arrhythmia detection. However, confirmation by trained staff is required for the clinical interpretation of telemetry data. Cardiologs, an AI-based ECG analysis platform validated for ambulatory ePatch recordings, offers the possibility of obtaining AI-based analysis and documentation of a large quantity of continuous ECG data. However, the feasibility of data export to Cardiologs and the quality of the AI-generated data analysis have not been assessed in inpatient settings. Purpose The primary goal of this pilot project was to evaluate the feasibility of integrating the Cardiologs platform into cardiac care workflows and the technical feasibility of using the platform to collect, structure, and process continuous ECG data from hospitalized patients monitored via standard telemetry. The secondary goal was to evaluate whether the AI-based Cardiologs platform could detect arrhythmias with an accuracy comparable to cardiologist-reviewed conventional telemetry. Methods Thirty patients in a cardiac intensive care unit underwent simultaneous monitoring with Philips IntelliVue telemetry (MX40/PIC iX) and a Philips ePatch for approximately 48 hours. The telemetry data were sent for processing and analysis to the Cardiologs platform, and the AI-generated report was compared to the manual ECG data review by blinded, experienced cardiologists. Results Cardiologs detected all arrhythmias identified on cardiologist-reviewed telemetry, corresponding to a sensitivity of 100%. In one case, Cardiologs identified an arrhythmia episode (Type 2 second-degree AV block) that had not been identified during telemetry monitoring or by the reviewing cardiologists. Conclusion This pilot project demonstrates strong agreement between Cardiologs and cardiologist-reviewed conventional telemetry for arrhythmia detection in hospitalized cardiac intensive care patients. The AI platform shows potential to support workflow optimization, standardized reporting, and automated ECG interpretation. Larger-scale evaluations are warranted to confirm these findings.
Stojadinovic et al. (Wed,) conducted a observational in Cardiac intensive care patients (n=30). Cardiologs AI-based ECG analysis vs. Cardiologist-reviewed conventional telemetry was evaluated on Arrhythmia detection. The Cardiologs AI-based ECG analysis platform detected all arrhythmias identified on cardiologist-reviewed conventional telemetry, corresponding to a sensitivity of 100%.