Wearable Patch ECG AF Detection Algorithms 2026
Wearable Patch ECG Atrial Fibrillation Detection Algorithms
Wearable ECG patch algorithms for AF detection have evolved from classical RR interval analysis to predictive deep learning pipelines across 60+ patent and literature records spanning 2006–2026. Key players include CardiOVU, West Affum Holdings DAC, Topia Life Sciences, and Google LLC.
From Disposable ECG Patches to Predictive AI: The AF Detection Landscape
Wearable ECG patch technology for AF detection spans two principal signal acquisition paradigms: direct ECG electrode patches adhered to the chest or wrist, and optical PPG sensors embedded in wristbands, smartwatches, or rings. Algorithms range from classical RR interval statistical methods to deep learning architectures including CNNs, LSTMs, and Bayesian neural networks.
The innovation timeline divides into four phases. The foundational period (2006–2015) was anchored by CardiOVU, Inc.’s disposable ECG patch patent family. The algorithmic diversification period (2016–2020) introduced Poincaré plot methods, entropy measures, and the Apple Heart Study enrolling 419,093 participants for PPG-based AF screening validation.
The clinical validation and AI integration period (2021–2023) is the most densely populated in this dataset, with more than 25 literature records. Studies validated algorithms on specific devices including Withings ScanWatch, Apple Watch, Fitbit, CardioTracker ring, Spyder patch, and mHealth patches, covering prospective cohorts and multi-center trials.
The predictive intelligence and multi-modal period (2024–2026) is characterized by a shift from detection to prediction, with Google LLC’s WO patent on future AF recurrence and Topia Life Sciences’ multi-jurisdictional pending family. In this dataset, India accounts for all four new filings in 2025–2026, signaling a new geographic entrant to this patent space.
Detection Algorithm Clusters and Jurisdiction Distribution
The dataset reveals four distinct algorithm clusters spanning classical statistical methods to predictive AI, with filings concentrated in the US and an emerging surge from India in 2025–2026.
Algorithm Cluster Distribution by Patent and Literature Record Count (Dataset Snapshot)
Deep learning on raw or time-frequency signals is the most represented algorithm cluster in this dataset, appearing across 10+ records from 2018 onward and dominating filings from 2020 onward.
↗ Click bars to explorePatent Filings by Jurisdiction — Wearable ECG AF Detection (Dataset Snapshot)
The US holds the largest share of patents in this dataset with 8 records; India represents the fastest-growing new filing jurisdiction with 4 records all dated 2025–2026 in retrieved records.
↗ Click bars to exploreClinical and Consumer Deployment Zones for Wearable AF Detection
The dataset covers five principal deployment contexts, from post-cryptogenic stroke screening to consumer wellness, each characterized by distinct device types, patient populations, and algorithm validation requirements.
Cryptogenic Stroke Screening
The CANDLE-AF trial (2022) is a multicenter RCT comparing 72-hour single-patch monitoring against standard strategy specifically in post-cryptogenic stroke patients. The Spyder device study (2021) enrolled patients for both post-ablation AF recurrence detection and post-cryptogenic stroke AF screening, representing a focused clinical validation of patch-based AF detection in this high-risk population.
Clinical TrialPost-Cardiac Surgery Monitoring
A 2022 study developed an Apple Watch PPG algorithm for immediate post-operative AF detection in 80 cardiac surgery patients. A parallel 2021 study validated an automated algorithm for new-onset AF in critically ill sepsis patients using stored ICU continuous ECG data, demonstrating the value of wearable-grade algorithms in inpatient environments.
Post-Operative CareConsumer Population Screening
The Fitbit Heart Study (2022) used overlapping 5-minute tachograms to flag irregular heart rhythm in adults 22+ years across compatible Fitbit devices. The Apple Heart Study (2019) enrolled 419,093 participants using PPG pulse irregularity notification on Apple Watch, marking the largest prospective consumer AF screening study in this dataset. Google LLC’s 2024 WO patent targets predictive AF recurrence in this same consumer base.
Consumer WellnessRemote Cardiology & Telemedicine
A 2022 ultralow-power ECG patch and cloud platform design enables 14-day uninterrupted monitoring with ANN+KNN algorithm accuracy of 98.7%. A self-reporting clinical trial platform (2022) integrates AWS-VPC cloud services with wearable ECG data and patient self-reporting for clinical trial infrastructure, supporting distributed remote cardiology workflows.
Remote MonitoringLeading Patent Assignees in Wearable ECG AF Detection — Dataset Snapshot
In this dataset, CardiOVU, Inc. is the most prolific filer with 5 records spanning 2006–2014, while West Affum Holdings DAC holds 3 active or pending US patents filed 2022–2025, making it the most active current ECG patch AF detection filer in the US jurisdiction in retrieved records.
Top Patent Assignees by Filing Count — Wearable ECG AF Detection (Dataset Snapshot)
↗ Click bars to exploreCardiOVU, Inc.
CardiOVU, Inc. is the most prolific patent filer in this dataset with 5 records spanning 2006–2014 across US, WO, and EP jurisdictions. Their filings establish the core hardware template of a thin, flexible, battery-powered patch with embedded amplifier, processor, and on-board software for arrhythmia classification, including the Programmable ECG Sensor Patch (US, 2006 and 2014). The EP filing (2007) is noted as inactive in the dataset; the US filings include both granted and programmable clinical parameterization variants.
United StatesWest Affum Holdings DAC
West Affum Holdings DAC holds 3 active or pending US patents in this dataset, with filings in 2022 (active legal status) and 2025 (granted active and pending continuation). Their 2022 US patent covers dry ECG electrode wearable AF detection with on-board or remote algorithms, noise detection, and AF burden characterization — a clinically significant metric beyond binary AF/NSR classification. The 2025 continuation further extends this dry-electrode AF detection claim set, making them the most active current ECG patch AF detection filer in the US jurisdiction in retrieved records.
Ireland (DAC)Next-Generation Trends in Wearable AF Detection (2024–2026)
The most recent records in this dataset (2024–2026) show a clear shift from binary AF detection toward prediction, spatial cardiac mapping, and adaptive AI noise filtering, with new geographic entrants accelerating the pace of patent activity.
Predictive AF Onset: From Detection to Anticipation
Google LLC’s 2024 WO patent explicitly claims predicting future AF recurrence from temporal patterns of biometric sensor data after an initial detected event. A 2022 deep learning model from literature achieved AF onset prediction an average of 30.8 minutes in advance with 83% accuracy and 85% F1-score. Only one patent in this dataset explicitly claims future AF event prediction, representing a defensible IP white space for teams with longitudinal AF event data.
Electrocardiographic Imaging on Flexible Patches
The 2026 Indian patent by Mihir Harishbhai Rajyaguru describes a dense dry-electrode array on a flexible patch solving the cardiac inverse problem to reconstruct high-resolution spatio-temporal cardiac activation maps — previously a technology confined to clinical electrophysiology labs. On-board ML handles rhythm classification from the reconstructed maps. This ECGI-on-patch architecture represents a step-change in diagnostic depth compared to single-lead systems in the dataset.
ECG Patch vs. PPG Wristband: AF Detection Algorithm Approaches
Click any row to explore further.
| Dimension | ECG Electrode Patch | PPG Wristband / Smartwatch |
|---|---|---|
| Signal Type | Direct cardiac electrical signal (single or multi-lead ECG) | Optical pulse waveform (photoplethysmography) |
| Primary Algorithm Input | Raw ECG waveform or R-R interval time series | Pulse interval tachogram or PPG waveform |
| Representative Algorithm Methods | CNN on time-frequency ECG, LSTM on raw ECG, dry-electrode AF burden characterization (West Affum Holdings DAC, 2022) | RR interval irregularity, Poincaré plots, autocorrelation morphology, LSTM on raw PPG; Apple Heart Study (419,093 participants) |
| Motion Artifact Sensitivity | Lower for chest patches; addressed by AI dynamic noise filtering (GITAM, 2026) | Higher; addressed by autocorrelation morphology features and multi-feature predictors (2021 wristband study) |
| Clinical Validation Scale | 178 patients (mHealth patch, 2022); 102 hospitalized patients (2020 inpatient study, 95.2% sensitivity) | 419,093 participants (Apple Heart Study, 2019); large Fitbit Heart Study (adults 22+, 2022) |
| Diagnostic Depth | AF/NSR classification, AF burden quantification, ECGI spatio-temporal maps (2026 ECGI patch) | Irregular rhythm notification; confirmed by follow-up ECG patch or 12-lead ECG |
| Wear Duration | Up to 14 days continuous (ultralow-power ECG patch, 2022); disposable single-use formats | Continuous use dependent on battery; typically days to weeks |
| Hybrid Approach Evidence | DoubleCheck-AF (2022): combined PPG + 6-lead ECG outperforms either modality alone, especially against ectopic beat confounders | DoubleCheck-AF (2022): PPG continuous monitoring triggers on-demand multi-lead ECG confirmation |
Frequently Asked Questions: Wearable Patch ECG AF Detection Algorithms
The dataset identifies four main clusters: (1) classical RR interval statistical analysis using metrics such as RMSSD, SDNN, sample entropy, and Poincaré plot ratios; (2) classical machine learning on engineered features including SVM, random forests, CatBoost (F1=0.92 on a 7,270-segment test set), and ANN; (3) deep learning on raw or time-frequency signals using 1D/2D CNNs, LSTMs, and Bayesian neural networks; and (4) predictive and multi-lead spatial approaches including ECGI patch reconstruction and temporal AF recurrence prediction.
In this dataset, CardiOVU, Inc. (US) is the most prolific filer with 5 records spanning 2006–2014 across US, WO, and EP jurisdictions. West Affum Holdings DAC holds 3 active or pending US patents filed 2022–2025. Topia Life Sciences Limited holds a 4-jurisdiction pending family (WO, GB, AU, US — 2024–2025). Google LLC filed a WO predictive AF patent in 2024.
ECG patches provide direct cardiac electrical signals enabling AF/NSR classification, AF burden quantification, and — in the 2026 ECGI patch — spatio-temporal cardiac maps. PPG wristbands use optical pulse waveforms, with algorithms based on pulse interval irregularity, Poincaré plots, and autocorrelation morphology. The DoubleCheck-AF study (2022) demonstrated that combining continuous PPG monitoring with on-demand 6-lead ECG outperforms either modality alone, especially against ectopic beat confounders.
Key studies include: the Apple Heart Study (2019, 419,093 participants, PPG pulse irregularity on Apple Watch); the Fitbit Heart Study (2022, adults 22+ with compatible Fitbit devices); the CANDLE-AF trial (2022, multicenter RCT, 72-hour single-patch post-cryptogenic stroke); a 2022 mHealth patch study on 178 patients; a 2020 inpatient study on 102 hospitalized patients achieving 95.2% sensitivity and 92.5% specificity with deep neural network algorithms on 5-minute PPG periods; and a 2022 post-cardiac surgery Apple Watch study on 80 patients.
Based on records in this dataset from 2024–2026: (1) Google LLC’s 2024 WO patent predicts future AF recurrence from temporal biometric patterns; (2) a 2026 Indian patent by Rajyaguru describes ECGI dense dry-electrode flexible patches reconstructing spatio-temporal cardiac maps; (3) a 2026 GITAM University patent claims AI-driven dynamic noise filtering integrated into the patch; (4) West Affum Holdings DAC’s 2022 and 2025 US patents claim AF burden characterization; and (5) Topia Life Sciences holds a multi-jurisdictional AI-ML pipeline ECG patch family (WO, GB, AU, US).
Among patents with identified jurisdictions in this dataset: US has 8 records (CardiOVU ×3, West Affum Holdings ×3, Jun Moon-Seog ×1, Urja S. Kadam ×1); India has 4 records (all filed 2025–2026, all pending); WO (PCT) has 3 records (CardiOVU, Topia Life Sciences, Google LLC); GB has 1 record (Topia Life Sciences); AU has 1 record (Topia Life Sciences); EP has 1 record (CardiOVU, inactive). India is the most active new filing jurisdiction in 2025–2026 within this dataset.
Data and insights on this page are based on a limited patent and literature dataset and are for reference only. Figures may not represent the complete technology landscape.