What happened
Insertable cardiac monitors record the heart's rhythm for years and flag possible arrhythmias. Many of the alerts are false, and clinicians must review each one. Avalia-t, the scientific advice unit of Galicia's health knowledge agency, assessed AI algorithms that re-read these recordings for Spain's network of assessment agencies in 2025. It searched the literature up to 31 March 2024 and found four diagnostic studies with 445 patients. In three studies, the algorithms removed between 60.3% and 98% of the monitors' false positives. Alerts that the Cardiologs algorithm confirmed were true 74.5% of the time, against 53.9% for the Medtronic monitor alone. IDOVEN's algorithm removed 1,065 of 1,087 false positives from Abbott's Confirm Rx. The algorithms missed between 1% and 2.9% of true events, mostly atrial fibrillation or atrial tachycardia. Avalia-t rated the studies at high risk of bias, and employees or advisers of the makers co-wrote each of them. It found no data on clinicians' time saved and could draw no conclusion on clinical impact.
Sources
-
Assessment report, by authors with no tie to the system.
Unidad de Asesoramiento Científico-técnico, Avalia-t, Agencia Gallega para la Gestión del Conocimiento en Salud (ACIS), for the Red Española de Agencias de Evaluación de Tecnologías Sanitarias y Prestaciones del SNS. López Loureiro I, Puñal-Rioboo J, Mejuto Martí T. Evaluación del rendimiento diagnóstico y la seguridad de la inteligencia artificial en la monitorización cardíaca con dispositivos insertables. Informes de Evaluación de Tecnologías Sanitarias. Madrid: Ministerio de Sanidad, 2025. NIPO 133-25-043-1. Resumen, Conclusiones.
Los algoritmos de IA muestran un VPP entre el 74,5% y el 89,74%. Así mismo, producen una reducción de FP entre el 60,3% y el 98%.
AI algorithms demonstrated PPV values between 74,5% and 89,74%, reducing FP rates by 60,3% to 98%.
Details
AI system
- Workflow
-
The algorithms re-read the ECG episodes that insertable monitors had stored as possible arrhythmias. Electrophysiology experts classified each episode as the reference standard.
Follow-up
- Conditions for replication
-
The monitor sends its recordings to a remote platform where the algorithm can analyze them. Each algorithm is checked on the monitors that it reads, since performance varied with the device and its software.
Entry details
- Sector
-
Healthcare (cardiology).
- Purpose
-
Idoven, Cardiologs and Implicity built Willem, the Cardiologs algorithm and ILR-ECG-A to re-read insertable cardiac monitor recordings and screen out false arrhythmia alerts.
30 more questions not yet checked
model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.
Cite this record
Peter H. Vartanian. (2026). Willem, Cardiologs and ILR-ECG-A cut false heart-monitor alerts (AIBD-0926). Artificial Intelligence Benefits Database, edition of 1 November 2026. https://petervartanian.xyz/aibd/records/AIBD-0926.html (archived at https://doi.org/10.7910/DVN/AADSII, version 7.0)