A comparison of three software platforms for automated ECG analysis

Abstract

Introduction Current regulatory guidelines attempt to standardize the models used to assess the safety aspects of new test compounds. However, they do not address the means of deriving the critical data. With the increased data volume that is a result of more extensive safety scrutiny and continuous data assessment, especially in cardiovascular telemetry studies, there is a clear need to assess the automated ECG analysis tools currently available on the market. Methods Cardiovascular studies were conducted using implanted beagle dogs following the oral administration of E-4031 (0 and 1 mg/kg) and the animals were monitored for 22–24 h post-dose. The raw ECG data trace was converted into file formats accessible by Data Sciences International (DSI) Ponemah with pattern recognition option (PRO), EMKA ecgAuto and Notocord HEM. Results Validation using a reference signal generator showed comparable performance by the applications being evaluated. Significant increases in QT/QTcV (25–40 ms) were noted following treatment with 1 mg/kg E-4031 ( T max ≤ 0.5 h). All software applications showed similar trends in QT/QTcV analyses. Discussion The choice of analysis applications is dependent on a number of variables, e.g., animal model used, the need for real-time analysis and visualization, flexibility, learning curve, etc. In our study, similar trends were observed for all three ECG software applications, but pattern recognition approaches were more amenable to adapting to ECG morphology changes over the course of 24 hour telemetry study.