One assignee leads this dataset with 41 records, far ahead of the rest of the ranked field, where fifth place holds only 3 records and tenth place holds 2. That gap indicates a single company built a broad early portfolio around implantable-device arrhythmia detection, while most other filers hold only a handful of related patents. Anyone assessing competitive risk in this space should treat that leader's portfolio, particularly its highly-cited R-wave and episode-profiling filings, as the primary block to clear.

Filing peaked in 2018 at 15 records in this dataset and has pulled back since; the clearest recent comparison, 2021 against 2024, shows a 57% decline. That decline reflects a field where the foundational detection logic — R-wave and premature ventricular contraction detection, episode profiling — was largely claimed by the mid-2010s. Because publication lags filing by around 18 months, 2025 and 2026 counts in any dataset will still be filling in, so a slower apparent pace in those years should not be read as confirmed further decline.

The dominant classes are A61B (diagnosis and surgery), present in 69.5% of the 105 records in this dataset, and A61N (electrotherapy and radiation therapy), present in 45.7%. Healthcare informatics under G16H appears in 12.4% of records, while pure software classes such as G06N for AI models sit at only 1.0%. This mix shows most detection-algorithm claims are drafted as part of a diagnostic or therapy device rather than as standalone software methods, which matters for how broadly a software-only classifier claim can be drafted.

Based on the class and claim distribution in this dataset, photoplethysmography-derived RR interval fusion, on-device false-positive suppression and deep learning classifiers built for wearable input carry visibly lighter claim density than the core implantable-device detection logic. Alert-threshold auto-calibration and cross-device clinical validation workflows show similar patterns. These are not guarantees of patentability, but they are areas where the dense prior art sitting on R-wave and episode-profiling claims does not directly reach.

US20220304612A1, assigned to Pacesetter, Inc. and published 2022-09-29, covers a machine learning model trained on cardiac activity signals from an implantable medical device to predict a future arrhythmia risk before the monitored characteristic shows overtly abnormal behaviour. It sits within the same assignee family responsible for several of the most-cited records in this dataset, including R-wave and premature ventricular contraction detection filings. Anyone building a predictive, ML-based arrhythmia risk classifier on implantable-device signal data should review this filing's specific claim language on training data and risk-factor identification before finalising claim scope.