Physicians’ perspectives on artificial intelligence in electrocardiography in clinical practice: a qualitative study

BACKGROUND: Artificial intelligence-enhanced electrocardiography (AI-ECG) may support ECG interpretation, prioritization, and workflow efficiency. ECGs are widely used, inexpensive, and often obtained with a low clinical threshold, meaning that AI-ECG may affect large numbers of patients and reshape an established clinical workflow. This study explored physician's perspectives on integrating AI-ECG into clinical practice. METHODS: Semi-structured interviews were conducted with 12 physicians in Region Stockholm, Sweden. Participants were selected through purposive and snowball sampling and included both specialists and nonspecialists. The interviews were transcribed and analyzed using thematic analysis. RESULTS: Two main themes were identified: Designing a new clinical landscape and Navigating a new clinical landscape. Participants saw potential for AI-ECG to improve diagnostic accuracy, support prioritization, and reduce workload. At the same time, they raised concerns about overreliance, accountability, and the limitations of human oversight. Participants also emphasized that AI-ECG may generate incidental or prognostic findings beyond the original reason for obtaining an ECG, raising questions about disclosure, consent, and actionability. CONCLUSIONS: Informants viewed AI-ECG as a potentially useful support for ECG interpretation in high-volume clinical workflow, but also identified ethical, practical, and professional aspects that require consideration beyond technical performance alone. Because ECGs are common investigations and are often obtained with a low clinical threshold, AI-ECG should be evaluated with regard to its role as a core component of clinical workflow. Implementation should include real-world validation, attention to alert burden and deskilling, and protocols for incidental or prognostic findings.