An apparatus comprising: circuitry configured to classify a signal; and circuitry configured to control saving of the signal to a memory with a conditional resolution, wherein a signal that is classified as anomalous is saved at higher resolution as a higher resolution signal and a signal that is not classified as anomalous is saved at lower resolution as a lower resolution signal or is not saved.

Sensing circuitry of an implantable medical device (IMD) system may sense a cardiac signal that varies according to a cardiac cycle of a patient. Processing circuitry of the IMD system may determine a series of consecutive cardiac cycle length metric values based on the sensed cardiac signal, identify a plurality of pairs of the cardiac cycle length metrics, each of the pairs of cardiac cycle length metrics separated by an integer n of the cardiac cycle length metrics, and construct a distribution of the pairs of cardiac cycle length metrics based on values of the cardiac cycle length metrics for each of the pairs. The processing circuitry may detect a sleep apnea episode of the patient based on one or more characteristics of the constructed distribution, and control communication circuitry of the IMD system to transmit an indication of the detected sleep apnea episode to the external computing device.

A monitoring system and method for detecting an infection or for assessing a suitability of neuromodulation therapy for a patient. R-R intervals of a patient are detected and stored for a first time period. A heart rate variability (HRV) of the stored R-R intervals is determined using at least one of a time domain analysis, an entropy analysis, a frequency domain analysis, a wavelet analysis, or a detrended fluctuation analysis. The patient is identified as exhibiting symptoms of a systemic infection and/or identified as suitable for neuromodulation therapy if the HRV is higher than a first threshold.

An ECG sensor chip used in a wearable appliance includes; a switch controlled by a switching signal, an amplifier that amplifies a difference between first and second ECG signals, and a location indicator that generates the switching signal. The switch passes either a first ECG signal or second ECG signal in response to the switching signal.

Conventionally, Atrial Fibrillation (AF) has been detected using atrial analyses which is vulnerable to background noise. Again there is a dependency on statistical features which are extracted from R-R intervals of long ECG recordings. The present disclosure addresses AF detection from single lead short ECG recordings of less than one minute wherein automatic detection of P-R and P-Q intervals is difficult, which introduces error in feature computing from the segregated intervals and compromises the performance of the classifier. In the present disclosure, a Recurrent Neural Network (RNN) based architecture comprising two Long Short Term Memory (LSTM) networks is provided for temporal analysis of R-R intervals and P wave regions in an ECG signal respectively. Output sates of the two LSTM networks are merged at a dense layer along with a set of hand-crafted statistical features to create a composite feature set for classification of the AF.

Techniques are disclosed for explaining and visualizing an output of a machine learning system that detects cardiac arrhythmia in a patient. In one example, a computing device receives cardiac electrogram data sensed by a medical device. The computing device applies a machine learning model, trained using cardiac electrogram data for a plurality of patients, to the received cardiac electrogram data to determine, based on the machine learning model, that an episode of arrhythmia has occurred in the patient and a level of confidence in the determination that the episode of arrhythmia has occurred in the patient. In response to determining that the level of confidence is greater than a predetermined threshold, the computing device displays, to a user, a portion of the cardiac electrogram data, an indication that the episode of arrhythmia has occurred, and an indication of the level of confidence that the episode of arrhythmia has occurred.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for obtaining an electrocardiographic (ECG) signal of a user; obtaining a feature vector of the ECG signal of the user with neural network based feature extraction. Comparing the feature vector of the ECG signal with a stored feature vector of a registered user. Authenticating the user in response to determining that a similarity of the ECG feature vector of the ECG signal and the stored ECG feature vector of the registered user exceeds a pre-defined threshold value.

Methods, systems, and devices for measuring respiratory parameters from an ECG device are described. The method may include receiving an electrocardiogram (ECG) signal associated with a patient. The method may further include detecting a change in modulation of the ECG signal between a first portion of the ECG signal and a second portion of the ECG signal. The method may further include determining a change in respiratory effort of the patient based at least in part on the change in modulation.

Systems and methods for predicting and/or detecting cardiac events based on real-time biomedical signals are discussed herein. In various embodiments, a machine learning algorithm may be utilized to predict and/or detect one or more medical conditions based on obtained biomedical signals. For example, the systems and methods described herein may utilize ECG signals to predict and detect cardiac events. In various embodiments, patterns identified within a signal may be assigned letters (i.e., encoded as distributions of letters). Based on the known morphology of a signal, states within the signal may be identified based on the distribution of letters in the signal. When applied in the in-vehicle environment, drivers or passengers within the vehicle may be alerted when an individual within the vehicle is, or is about to, experience a cardiac event.

The present disclosure is directed towards systems and methods built for predictively timing the inflation and/or deflation of an intra-aortic balloon pump. A controller operates in three states: (1) initialization state, (2) learning state, and (3) peak detection state. The controller decomposes a patient's electrocardiogram signal to a power signal. It then learns characteristics of the patient's electrocardiogram signal during the learning state and computes adaptive threshold parameter values. During the peak detection state, the controller applies the learnt threshold parameter values on a current electrocardiogram signal to identify occurrence and timings of R peaks in the electrocardiogram signal. The R-to-R peak timings are then used to trigger inflation of an intra-aortic balloon pump.