Non-invasive cardiac monitor and methods of using recorded cardiac data to infer a physiological characteristic of a patient
The present invention relates to a non-invasive cardiac monitoring device that records cardiac data to infer physiological characteristics of a human, such as cardiac arrhythmia. Some embodiments of the invention allow for long-term monitoring of physiological signals. Further embodiments allow for processing of the detected cardiac rhythm signals partially on the wearable cardiac monitor device, and partially on a remote computing system. Some embodiments include a wearable cardiac monitor device for long-term adhesion to a mammal for prolonged detection of cardiac rhythm signals.
1. A system comprising a wearable monitor device for monitoring signal data of a patient, the wearable monitor device comprising:
a housing comprising a surface configured to be engaged to the patient;
a sensor positioned to detect continuous physiological signals of the patient while the surface is engaged to the patient;
a hardware processor configured to:
determine an activity level of the patient;
select a machine learning model based at least in part on the activity level of the patient;
access an encoder of the machine learning model, wherein the encoder of the machine learning model is trained together with a decoder of the machine learning model; and
process the detected continuous physiological signals through the encoder; and
a transmitter configured to transmit an output of the encoder or a signal derived from the output of the encoder to a computing system, wherein the computing system is configured to process the output of the encoder or the signal derived from the output of the encoder via the decoder.
2. The system of claim 1 , wherein the encoder comprises a first subset of layers of the machine learning model.
3. The system of claim 1 , wherein the computing system is configured to infer a likelihood of an occurrence of cardiac arrhythmia by processing the output through the decoder.
4. The system of claim 3 , wherein the decoder comprises a second subset of layers.
5. The system of claim 3 , wherein the wearable monitor device comprises an accelerometer configured to gather motion data, wherein the computing system is configured to match motion data with the detected continuous physiological signals to infer the likelihood of the occurrence of cardiac arrhythmia.
6. The system of claim 1 , wherein the wearable monitor device is a wearable patch that is detachable by the patient.
7. The system of claim 1 , wherein the wearable monitor device is applied on the chest of the patient.
8. The system of claim 1 , wherein the wearable monitor device is an implantable device that is wearable by the patient by implanting the wearable monitor device within the patient.
9. The system of claim 1 , wherein the sensor includes one or more electrodes and the continuous physiological signals comprise electrocardiography (ECG) data.
10. The system of claim 1 , wherein the sensor includes an optical sensor and the continuous physiological signals comprise photoplethysmography (PPG) data.
11. The system of claim 1 , wherein the decoder comprises a second neural network.
12. A method comprising:
determining an activity level of a patient;
selecting, by a hardware processor of a wearable monitor device that detects continuous physiological signals of the patient, a neural network of a plurality of neural networks, wherein the neural network is selected based at least in part on the activity level of the patient, and wherein the selected neural network comprises a first subset and a second subset of layers;
processing the detected continuous physiological signals through an encoder without processing an output of the encoder through a decoder, wherein the encoder comprises the first subset of layers, wherein the encoder is trained on an external computing device with a decoder, wherein the decoder comprises the second subset of layers; and
transmitting the output of the encoder or a signal derived from the output of the encoder to a computing system.
13. The method of claim 12 , wherein the method further comprises quantizing the output of the encoder, wherein to transmit the output comprises transmitting the quantized output.
14. The method of claim 12 , wherein the first subset of layers and a second subset of layers are of the same neural network.
15. The method of claim 12 , wherein the wearable monitor device comprises a wearable cardiac monitor device, wherein the continuous physiological signals comprise cardiac rhythm signal data.
16. A system comprising a wearable monitor device for monitoring signal data of a patient, the wearable monitor device comprising:
a housing comprising a surface configured to be engaged to the patient;
a sensor positioned to detect continuous physiological signals of the patient while the surface is engaged to the patient;
a hardware processor configured to:
determine an activity level of the patient;
select a machine learning model of a plurality of machine learning models based at least in part on the activity level of the patient, wherein the selected machine learning model comprises an encoder and a decoder that are trained together on an external device; and
process the detected continuous physiological signals through a portion of the machine learning model stored on the wearable monitor device, wherein the portion includes the encoder; and
a transmitter configured to transmit an output of the encoder or a signal derived from the output of the encoder to a computing system, wherein the computing system is configured to process the output of the encoder or derived signal via the decoder.
17. The system of claim 16 , wherein a dimensionality of the encoder is smaller than the data of the detected continuous physiological signals from the sensor.
18. The system of claim 16 , wherein the output of the encoder is encrypted, wherein to transmit the output comprises transmitting the encrypted output.
19. The system of claim 16 , further comprising a receiver configured to receive an updated encoder and updating the encoder to the updated encoder, wherein the hardware processor is further configured to process signals through the updated encoder.
20. The system of claim 16 , wherein the hardware processor is further configured to infer a likelihood of an occurrence of cardiac arrhythmia by processing the output of the encoder through the decoder, wherein the encoder processes at least 24 hours of continuously detected, stored physiological signals.
21. The system of claim 16 , wherein the wearable monitor device further comprises a user trigger configured to depress and initiate recordation of an instance in time of a perceived cardiac event.
22. The system of claim 16 , wherein the hardware processor is further configured to select the machine learning model from the plurality of machine learning model based on a characteristic of the system.
23. The system of claim 22 , wherein the characteristic of the system comprises one or more of: a remaining amount of battery, a network characteristic between the system and an external computing system, or a wear duration.
24. The system of claim 22 , wherein the characteristic includes one or more of: a characteristic of the patient or a severity of cardiac arrhythmia.
25. The system of claim 16 , wherein the hardware processor is further configured to compress the output of encoder, wherein to transmit the output comprises transmitting the compressed output.
26. The system of claim 16 , wherein the hardware processor is further configured to quantize the output of the encoder.
27. The system of claim 26 , wherein quantizing comprises rounding, truncating, or reducing a number of bits for the output of the encoder.
28. The system of claim 26 , wherein the hardware processor is further configured to determine an amount of quantization based on one or more of: a characteristic of the system or a lossless compression performance.
29. The system of claim 26 , wherein the hardware processor is further configured to determine an amount of quantization based on at least one of: a processing power, a storage capacity, an amount of remaining storage capacity, a network characteristic, or an accuracy of the machine learning model.