Systems and methods of arrhythmia detection

Inventors

McCalmont, Stephen A. • MacEachern, Stuart P. • Beck, Ralph L. • Falcone, Anthony

Assignees

Braveheart Wireless Inc

Abstract

Systems and methods of arrhythmia detection and associated apparatus that utilize machine learning techniques that allow for the consideration individual characteristics and the tailoring/personalization of biometric data allow for early detection and treatment, especially of cardiac arrhythmias and other abnormalities.

Core Innovation

The invention provides a detection and classification system that receives a biometric signal from a user and automatically processes the signal for medical-condition related diagnosis and follow-up. The system includes a feature detection module that analyzes the biometric signal for repeating segments, and a signal feature classifier that classifies the biometric signal using information regarding the repeating segments. The system is configured to automatically notify a predetermined third-party of the occurrence of a predetermined classification contained within a reference model, and the predetermined classification indicates the presence of a medical condition.

The reference model initially comprises a biological mechanism model and data corresponding to a general population. The system updates the reference model over time using data corresponding to the biometric signal received by the sensor module, such that the reference model incorporates the user's baseline patterns. The updating process includes a supervised learning phase configured to create an inferred function used by the signal feature classifier to map data corresponding to the biometric signal received subsequent to the updating as normal or abnormal.

In a wearable form, the invention provides a unitary housing configured to be worn by a patient that includes an ECG sensor module and an ECG signal filter. A beat detection module analyzes the processed ECG signal for repeating segments, where the repeating segments are QRS complexes. A beat classifier classifies the ECG signal as normal or abnormal based on a comparison of processed ECG signal information, biometric data from at least one of a plurality of biometric sensors, and a model ECG signal.

The model ECG signal includes a heartbeat mechanism model combined with a population-based model ECG signal integrated with patient data obtained over time to incorporate the patient's baseline patterns. The wearable system assigns a score to a classification based on how closely it matches a particular model and automatically notifies a predetermined third-party of an abnormal classification and its associated score for formal diagnosis and follow-up.

Claims Coverage

The partial content provides two independent claims. Across these claims, the main inventive features focus on repeating-segment feature detection, model-based classification that incorporates patient baseline patterns, automated notification of a predetermined third-party, and assignment of a classification score in the wearable claim.

Across the independent claims, the system detects repeating segments in a biometric signal, including QRS complexes in the wearable claim, classifies normal versus abnormal or a predetermined medical-condition classification using a reference or model ECG signal that incorporates patient baseline patterns via updating over time, and automatically notifies a predetermined third-party; the wearable claim further assigns and communicates a classification score for formal diagnosis and follow-up.

Stated Advantages

Documented Applications