Systems and methods for classifying motion of a patient wearing an ambulatory medical device

A wearable cardioverter/defibrillator apparatus for monitoring arrhythmias during different types of patient activities, comprising:

• one or more sensing electrodes configured to sense electrical signals that are indicative of cardiac activity of a patient;
• one or more accelerometers configured to generate a plurality of motion signals representative of movement of the patient;
• at least one memory configured to store
• an arrhythmia detection process configured to execute in one of an activity-induced noise (AIN) sensitive mode and an AIN robust mode, and
• a motion classifier trained on motion signals annotated with a plurality of motion classifications;
• a user interface configured to receive patient input;
• one or more accelerometers configured to generate a plurality of motion signals representative of movement of a patient;
• and
• at least one processor coupled to the at least one memory and the one or more accelerometers and configured to:
• receive the electrical signals from the one or more sensing electrodes,
• determine electrocardiogram (ECG) data for the patient based on the electrical signals,
• cause the arrhythmia detection process to execute in the AIN sensitive mode, wherein the AIN sensitive mode comprises
• analyzing a first plurality of ECG metrics based on the ECG data to detect an arrhythmia,
• responsive to detecting the arrhythmia, initiating a process for determining whether treatment is warranted, and
• if it is determined that treatment is warranted, initiating the treatment,
• analyze motion signals of the one or more accelerometers to determine initiation of a high-noise activity, wherein analyzing comprises
• extracting, from the plurality of motion signals, a plurality of features relating to a current state or activity of the patient based on the plurality of motion signals,
• applying the motion classifier to a feature vector comprising the plurality of features to generate a respective classification of the plurality of motion classifications, and
• storing, to the at least one memory, the respective classification,
• cause, in response to determining the initiation of the high-noise activity, the arrhythmia detection process to execute in the AIN robust mode, wherein the AIN robust mode comprises
• analyzing a second plurality of ECG metrics based on the ECG data to detect the arrhythmia, wherein
• the second plurality of ECG metrics comprises a subset of the first plurality of ECG metrics, and
• responsive to detecting the arrhythmia,
• providing a notification to the patient to stop the high-noise activity, and
• causing the arrhythmia detection process to execute in the AIN sensitive mode,
• determine a termination of the high-noise activity based on one or more of a predetermined time out condition occurring after a predetermined time period, the patient input, or the plurality of motion signals, and
• cause, in response to determining the termination of the high-noise activity, the arrhythmia detection process to execute in the AIN sensitive mode.