Cardiology · Ambulatory ECG analysis (deep-neural-network arrhythmia detection)
Cardiologs
Philips (Cardiologs Technologies)
One of the earliest deep-learning ECG platforms to clear the FDA, Cardiologs has done the harder thing than most catalogue-mates: earned independent evidence. A Minnesota emergency-medicine group found its network read 12-lead ECGs more accurately than the conventional algorithm it was tested against, and external validators in Belgium have scrutinised its atrial-fibrillation model. Now a Philips product, it is a decision aid whose reports a clinician must still sign — a point one adverse-event report drives home.
Performance Metrics
Clinical Evidence
Cardiologs has an unusually independent evidence base for a catalogue graduation. Its foundational validation is external, not internal: Smith and colleagues at Hennepin County Medical Center and the University of Minnesota (Journal of Electrocardiology, 2019) compared the deep neural network against the conventional Veritas algorithm on emergency-department 12-lead ECGs. The network interpreted 72.0 percent of ECGs accurately against 59.8 percent for the conventional algorithm, with a similar sensitivity for any major abnormality (69.6 versus 68.3 percent, not a significant difference) but a markedly higher positive predictive value (74.0 versus 56.5 percent). The headline is a gain in specificity and accuracy rather than in raw detection. The largest routine-practice study is manufacturer-affiliated and should be read as such. Fiorina and colleagues (Journal of the American Heart Association, 2022) analysed 1,000 twenty-four-hour Holter recordings from three tertiary hospitals and reported the AI-based platform non-inferior to the conventional analysis across five predefined abnormalities — pauses, ventricular tachycardia, atrial fibrillation or flutter, high-grade atrioventricular block, and a high burden of premature ventricular contractions. Non-inferiority, not superiority, is the claim the data support. The most consequential external work concerns the harder task of predicting paroxysmal atrial fibrillation from an ECG recorded in sinus rhythm. Gruwez and colleagues (JACC: Clinical Electrophysiology, 2023) independently validated that approach in a Belgian cohort. It confirms the direction of the manufacturer's own claims for this feature while underlining that sinus-rhythm prediction of future atrial fibrillation remains an emerging capability rather than a settled one. The consistent reading across the literature is of a competent, clinician-supervised interpretation aid whose strongest evidence is independent, not of an autonomous diagnostician.
| Study | Design | n | Sensitivity | Specificity | AUC | Published |
|---|---|---|---|---|---|---|
| Smith SW, Walsh B, Grauer K, Wang K, Rapin J, Li J, Fennell W, Taboulet P (independent, Hennepin / University of Minnesota) | ProspectiveProspective | 1,500 | 69.6% (any major abnormality; vs 68.3% conventional, ns) | Higher than conventional (accuracy 72.0% vs 59.8%) | — | J Electrocardiol 2019;52:88–95 (PMID 30476648). Independent comparison of the Cardiologs deep neural network against the conventional Veritas algorithm on emergency-department 12-lead ECGs; PPV 74.0% vs 56.5%. |
| Fiorina L, Maupain C, Gardella C, et al. (manufacturer-affiliated) | ProspectiveProspective | 1,000 | — | — | — | J Am Heart Assoc 2022;11(18):e026196 (DOI 10.1161/JAHA.122.026196; PMC9683671). 1,000 24-hour Holter recordings from 3 tertiary hospitals; AI platform non-inferior to conventional analysis across 5 predefined abnormalities. Authors affiliated with Cardiologs. |
| Gruwez H, Barthels M, Haemers P, Verbrugge FH, et al. (independent, Belgium) | RetrospectiveRetrospective | 0 | — | — | — | JACC Clin Electrophysiol 2023;9(7 Pt 2):1771–1782 (PMID 37354171). Independent external validation of the AI approach to identifying underlying paroxysmal atrial fibrillation from a 12-lead ECG in sinus rhythm; cohort size not asserted here pending a primary-source read. |
Clinical Pulse
Verified clinician sentiment
Verified clinician reviews launching soon · Apply to contribute →
Reviews are signed by clinicians verified against their regulator's public register.
Inside the algorithm
Editorial featureHow Cardiologs reaches a report.
Five stages — from raw input to verdict — drawn from manufacturer documentation and the public regulatory record.
- INGEST
- NORMALISE
- DETECT
- LOCALISE
- VERDICT
Stage 01 · INGEST
An ambulatory recording, from almost any device.
Cardiologs works on ambulatory ECG uploaded to a cloud service. Unlike a bedside monitor tied to one hardware line, it is device-agnostic by design: it accepts recordings from Holter monitors, patches, event recorders and other single- or multi-lead systems, and the manufacturer describes support down to consumer smartwatch traces. The recording is the raw material; the network does the reading.
Input
Ambulatory ECG recording
Inference
Cloud-based, per recording
Inside the Auris+ Listing
Five more sections complete this device’s Auris+ Listing.
Decision Ledger
ProPro unlocks a private, cross-vendor log of every case you read on this device — what the AI called, what you concluded, and a one-line reason if you overrode it. Ready for the EU AI Act's deployer logging obligations when they land in 2028.
Clinical Evidence Deep Dive
ProPro unlocks the structured clinical-evidence summary — study count, target patient population, and a tabular accuracy-metrics view drawn from peer-reviewed sources.
Peer-Reviewed Publications
ProPro unlocks the curated peer-reviewed publication list with PubMed cross-links — the citation backbone of every editorial verdict.
Post-Market & Regulatory Conditions
ProPro unlocks the post-market surveillance summary, recall record, and the conditions of approval that bound real-world use.
AI Algorithm Version History
ProPro unlocks the chronological record of algorithm version changes — what changed when, drawn from manufacturer changelogs and regulatory filings.
Regulatory Approvals
Safety Record
A MAUDE report associated with the Cardiologs Holter Platform (product code DPS) describes a patient death during ECG recording, with a customer alleging the platform missed a pause or ventricular fibrillation. Ventricular-fibrillation detection is outside the platform's cleared indications; a manufacturer review of ~400,000 recordings reported pause-detection performance at or above expected levels. Recorded as a single adverse-event report, not a recall or FDA safety communication.
One adverse-event report frames the device's risk boundary precisely. A MAUDE report associated with the Cardiologs Holter Platform describes a patient who died while wearing a recording device, with a customer alleging the platform missed a pause or ventricular fibrillation. The report notes that detection of ventricular fibrillation is not within the platform's indications, and that a manufacturer review of roughly 400,000 recordings found pause-detection performance at or above expected levels. The episode illustrates the cleared design: the software is an aid whose reports a clinician must validate, and it makes no claim to catch every rhythm. No recall or FDA safety communication for the platform was located in indexed public sources as of July 2026. MAUDE could not be queried directly from this environment, so the absence of further reports is "none found in public reporting" rather than an exhaustive audit.
Intended Use & Indications
Cardiologs is prescription software for use by qualified healthcare professionals for the assessment of arrhythmias using ECG data. It accepts recordings from Holter monitors, event recorders, patch and other ambulatory ECG devices, processes them with a deep-neural-network algorithm to detect heartbeats and rhythm abnormalities, and returns an editable draft report. The original 2017 clearance covered subjects over 18; a 2021 clearance extended the indicated population to include paediatric patients. The interpretation results are advisory and are not intended to be the sole means of diagnosis; a clinician reviews, edits and validates each report before it is issued.