A Multicenter Pragmatic Implementation Study of ECG-AI-Based Clinical Decision Support Software to Identify Low LVEF

A prospective, cluster-randomized, care-as-usual controlled trial to evaluate the impact of an ECG-based artificial intelligence (ECG-AI) algorithm to detect low left ventricular ejection fraction (LVEF) on diagnosis rates of LVEF ≤ 40% in the outpatient setting. The objective of this study is to evaluate the impacts of an ECG-AI algorithm to detect low LVEF and an associated Medical Device Data System when used during routine outpatient care. The study will be conducted in 2 phases: feasibility assessment phase and clinical impact phase.

The study is a prospective, cluster randomized, care-as-usual controlled trial that will be conducted at 6 sites in the USA. Primary care clinicians and general cardiologists will be invited and consented to participate in the study. For clinicians that accept, practice groups will be randomized to receive access to and education about the Low EF AI-ECG software and encompassing software or to provide care-as-usual in the control group. The study will be conducted in two phases: a feasibility pilot to evaluate integration and usability followed by observational period(s) to evaluate clinical outcomes. Analyses of the primary and secondary endpoints will be conducted on data from patients that meet the inclusion and exclusion criteria. The expected duration of the study is 12 months, including a feasibility phase (estimated 6 weeks) followed by a 3-month initial observation period with rolling observation count monitoring until the target number of patient encounters is reached, followed by a 90-day follow up period. At the completion of the feasibility period, we will evaluate quantitative and qualitative outcomes to inform the following observational period(s). Primary endpoints and exploratory endpoints will be assessed the end of the study.