Get started
Work

Performance of an artificial intelligence-powered smartphone application in the UK clinical settings: ECG automation compared to healthcare professionals

Year: 2025
Type: article
Abstract: BACKGROUND: The electrocardiogram (ECG) is widely used in clinical practice, but accurate interpretation requires significant expertise. Variability in training leads to inconsistent diagnostic accuracy amongst healthcare professionals. Artificial intelligence (AI) applications, such as PMCardio (Powerful Medical, Samorin, Slovakia), can digitise and interpret ECGs. While validated in selected populations, its performance compared to clinicians in UK practice has not been assessed. METHODS: Seventy-six healthcare professionals interpreted eight ECG traces (seven abnormal, one normal). Their performance was compared with the PMCardio application. Accuracy and time were recorded. RESULTS: Healthcare professionals achieved a mean accuracy rate of 67.1% (SD 24.0%), improving with seniority (junior 60%, mid-level 67.5%, senior 80%). PMCardio achieved perfect accuracy on the tested ECGs. Clinicians interpreted faster (median 23.7 s, range 9.1 s) compared to PMCardio (39.0 s, range 8.0 s), noting that the application's timing included operational steps such as loading and capturing ECG images. CONCLUSIONS: PMCardio demonstrated higher diagnostic accuracy than healthcare professionals but required longer interpretation times. Given the small dataset (8 ECGs) and lack of patient context, results should be interpreted cautiously. While AI tools may support clinicians and enhance consistency, they may also introduce uncertainty for less experienced users. Further studies with larger, real-world datasets are needed before widespread adoption.
DOI: 10.1186/s43044-025-00689-1
Authors: Ahmed Kassem, John Folkes, Sahil Mukherjee, James Rosengarten
Language: English
Field: Medicine
Open Access status: diamond
OpenAlex ID: W4415247000
FWCI: 1.297
Cites: 21
Cited by: 2
Related to: