Original Title: Evaluación del rendimiento diagnóstico y la seguridad de la inteligencia artificial en la monitorización cardíaca con dispositivos insertables

Authors' objectives: to assess the effectiveness of AI-based algorithms for arrhythmia detection in patients with implantable devices in terms of diagnostic performance, clinical and technical safety (FN and FN), clinician satisfaction, and economic, organizational, ethical, legal, and technical aspects.

Authors' results and conclusions: the evidence analyzed included four observational diagnostic studies and one qualitative study. The diagnostic studies involved a total of 445 patients generating 6113 recordings, of which 3761 were analyzed. The qualitative study included 1136 comments from 905 healthcare professionals. Three diagnostic studies were prospective, and one was retrospective, with patient overlap in two prospective studies. Regarding evidence quality, diagnostic studies exhibited a high risk of bias, while the qualitative study presented high quality. Two studies reported a negative predictive value (NPV) between 98,1% and 98,6%, while three described FN rates ranging from 1% to 2,9%. Positive predictive value (PPV) ranged from 74,5% to 89,74% across three studies, with a 60,3% to 98% reduction in FPs due to AI. For AF detection, sensitivity exceeded 96% in four studies, with specificity ranging from 60,3% to 95,4% in three studies. Sensitivity values for other arrhythmias came from three studies, while specificity data was described in two studies. For ventricular tachycardia (VT), sensitivity ranged from 77,78% to 100%, and specificity from 67,4% to 99,73%. For bradycardia, sensitivity ranged from 83,3% to 100%, with specificity between 73,3% and 99,95%. For asystole, sensitivity ranged from 93,3% to 100%, and specificity from 85,7% to 100%. Regarding technology acceptability, the qualitative study indicated clinicians held neutral attitudes toward AI in cardiac monitoring and ECG interpretation. Insufficient evidence was found to address ethical, social, legal, technical, or economic aspects.

Authors' methods: A systematic review of scientific literature up to March 31, 2024, was conducted using databases specialized in systematic reviews and health technology assessment reports (Cochrane, Epistemonikos, International Health Technology Assessment database, and Prospero), general databases (Medline, Embase, Web of Knowledge), and databases of ongoing research projects (clinicaltrials.org, Cochrane). Manual searches in metasearch engines and websites of national and international scientific societies and organizations, as well as a manual review of the references in selected studies, complemented the process. Search results were managed using EndNote 20 to remove duplicates and facilitate document management. Two researchers independently and blindly reviewed and selected the studies using the Covidence platform. Systematic reviews, comparative, observational, and economic evaluation studies analyzing the use of AI for interpreting ECGs from implantable devices were included. Narrative reviews, case studies, conference abstracts, and studies where algorithms did not involve AI were excluded. Screening was initially based on titles and abstracts, followed by full-text reviews. Discrepancies were resolved by consensus. Study validity and risk of bias were assessed using the QUADAS-2 scale for diagnostic tests and the CASPe scale for qualitative studies. Extracted results were qualitatively analyzed, with safety evaluated from clinical and technical perspectives. Diagnostic performance was assessed both for specific arrhythmias and overall.

Project Status: Completed

Year Published: 2025

URL for published report: https://runa.sergas.gal/xmlui/handle/20.500.11940/19961

Requestor: Ministry of Health

English language abstract: An English language summary is available

Publication Type: Full HTA

Country: Spain

Organisation Name: Scientific Advice Unit, avalia-t; The Galician Health Knowledge Agency (ACIS)

Contact Address: Conselleria de Sanidade, Xunta de Galicia, San Lazaro s/n 15781 Santiago de Compostela, Spain. Tel: 34 981 541831; Fax: 34 981 542854;

Contact Name: avalia_t.acis@sergas.es

Contact Email: avalia_t.acis@sergas.es

Copyright: Scientific Advice Unit, Avalia-t; The Galician Health Knowledge Agency (ACIS)