Granted Patent B2

US 12,734,367 · App. 19/126,999 · Granted Sep 15, 2026

Inventors: Jason Coult (Seattle, WA); Thomas Rea (Seattle, WA); Peter J. Kudenchuk (Seattle, WA); Heemun Kwok (Seattle, WA); Jose Nathan Kutz (Seattle, WA)

Assignee: University of Washington

A61N1/3925A61N1/39044G16H50/70

Abstract

Claims (79)

1 . A medical device, comprising:

at least one sensor configured to detect an electrocardiogram (ECG) of a patient who is in a state of cardiac arrest;

an output device configured to output an instruction to administer a treatment; and

a processor configured to:

identify a pre-shock segment of the ECG, the pre-shock segment being detected by the sensor when a heart of the patient is in a state of ventricular fibrillation (VF) during a time interval preceding administration of an initial electrical shock to the heart of the patient;

identify at least one post-shock segment of the ECG, the at least one post-shock segment being detected by the sensor during a time interval following administration of the initial electrical shock to the heart of the patient;

predict, based on the pre-shock segment and the at least one post-shock segment, that the patient is exhibiting shock-refractory VF by:

converting the pre-shock segment into a first scalogram;

converting the at least one post-shock segment into at least one second scalogram;

generating comparisons between the first scalogram and the at least one second scalogram to eigenscalograms; and

determining whether the comparisons are indicative of shock-refractory VF; and

in response to predicting that the patient is exhibiting shock-refractory VF, cause the output device to output the instruction to administer a treatment.

2 . The medical device of claim 1 , wherein the processor is configured to convert the pre-shock segment into the first scalogram by performing a Morlet wavelet transform on the pre-shock segment, and

wherein the processor is configured to convert the at least one post-shock segment into the at least one second scalogram by performing the Morlet wavelet transform on the at least one post-shock segment.

3 . The medical device of claim 1 , wherein the patient is receiving first chest compressions during the time interval preceding administration of the initial electrical shock to the heart of the patient and the processor is further configured to remove artifacts associated with the first chest compressions from the pre-shock segment of the ECG, and/or

wherein the patient is receiving second chest compressions during the time interval following the administration of the initial electrical shock to the heart of the patient and the processor is further configured to reduce artifacts associated with the second chest compressions from the at least one post-shock segment of the ECG.

4 . The medical device of claim 3 , wherein the processor is further configured to:

reduce artifacts associated with the second chest compressions from the at least one post-shock segment of the ECG by normalizing wavelet coefficient magnitude values of the at least one second scalogram, and

normalize the first scalograms and the second scalograms by dividing wavelet coefficient magnitude values within each scalogram frequency by at least one of a maximum, median, or mean wavelet coefficient magnitude within each scalogram frequency.

5 . The medical device of claim 1 , wherein the treatment comprises administration of an antiarrhythmic medication to the patient, administration of a subsequent electrical shock at a different vector than the initial electrical shock, administration of a double-sequential defibrillation (DSD) treatment, administration of cardiac catheterization treatment, withholding of vasopressor treatment, or administration of extracorporeal membrane oxygenation treatment.

6 . The medical device of claim 1 , wherein generating comparisons between the first scalogram and the at least one second scalogram to eigenscalograms comprises:

identifying pre-shock dot products by comparing the pre-shock scalogram to the eigenscalograms;

identifying post-shock dot products by comparing the at least one post-shock scalogram to the eigenscalograms, and

wherein determining that the comparisons are indicative of shock-refractory VF comprises

inputting the pre-shock dot products and the post-shock dot products into a trained machine learning (ML) model; and

receiving a classification that the patient is susceptible to shock-refractory VF from the trained ML model.

7 . The medical device of claim 1 , wherein the output device is configured to output the instruction to administer the treatment during a CPR period following the administration of the initial electrical shock to the patient, and

wherein the CPR period ends about 2 minutes after administration of the initial electrical shock to the patient.

8 . A method, comprising:

identifying a first segment of an electrocardiogram (ECG) that ends prior to administration of an electrical shock;

identifying at least one second segment of the ECG that begins after the administration of the electrical shock;

predicting, based on the first segment and the at least one second segment, that the ECG is indicative of a defibrillation-resistant heart rhythm by:

converting the first segment into a first scalogram;

comparing the first scalogram to eigenscalograms;

converting the at least one second segment into at least one second scalogram; and

comparing the at least one second scalogram to eigenscalograms; and

outputting an indication of the predicted defibrillation-resistant heart rhythm.

9 . The method of claim 8 , wherein converting the first segment into a first scalogram comprises performing a wavelet transform on the first segment, and

wherein converting the at least one second segment into at least one second scalogram comprises performing a wavelet transform on the at least one second segment.

10 . The method of claim 8 , wherein the at least one second segment comprises chest compression artifacts, and

wherein the method further comprises:

reducing the chest compression artifacts from the at least one segment by normalizing wavelet coefficient magnitude values of the second scalograms.

11 . The method of claim 10 , wherein normalizing the wavelet coefficient magnitude values of the second scalograms comprises dividing the wavelet coefficient magnitude values within each scalogram frequency by at least one of a maximum, a median, or a mean wavelet coefficient magnitude within each scalogram frequency.

12 . The method of claim 8 , further comprising:

trimming the first scalogram to a first frequency range,

trimming the second scalogram to a second frequency range, and

wherein the first frequency range is broader than the second frequency range.

13 . The method of claim 8 , wherein comparing the first scalogram to eigenscalograms comprises determining first dot products of the first segment,

wherein comparing the at least one second scalogram to the eigenscalograms comprises determining second dot products of the at least one second segment, and

wherein determining that the ECG is indicative of the defibrillation-resistant heart rhythm comprises inputting the first dot products and the second dot products into a classifier, the classifier being at least one of a trained random forest model, SVM, or k-nearest neighbor model.

14 . The method of claim 8 , the electrical shock being a first electrical shock, the method further comprising:

outputting an instruction to administer an additional treatment,

wherein the additional treatment comprises at least one of administration of an antiarrhythmic medication, administration of a second electrical shock at a different vector than the first electric shock, administration of the second electrical shock at a higher energy level than the first electrical shock, administration of extracorporeal membrane oxygenation, administration of a cardiac catheterization treatment, or administration of DSD.

15 . The method of claim 8 , wherein predicting, based on the first segment and the second segment, that the ECG is indicative of the defibrillation-resistant heart rhythm further comprises:

determining that the first segment exhibits VF; and

determining that at least one of a frequency, energy, or magnitude of the VF is below a threshold.

16 . The method of claim 8 , wherein outputting the indication of the predicted defibrillation-resistant heart rhythm occurs during a CPR period following the electrical shock, and/or

wherein outputting the indication of the predicted defibrillation-resistant heart rhythm occurs within two minutes of administration of the electrical shock.

17 . The method of claim 8 , wherein identifying the at least one second segment of the ECG that begins after the administration of the electrical shock comprises:

identifying a third segment of the ECG that begins after the administration of the electrical shock;

determining that the defibrillation-resistant heart rhythm cannot be predicted based on the first segment and the third segment; and

in response to determining that the defibrillation-resistant heart rhythm cannot be predicted based on the first segment and the third segment, identifying the at least one second segment of the ECG that comprises the third segment and a fourth segment of the ECG that begins after the third segment.

18 . The method of claim 8 , wherein the at least one second segment of the ECG comprises multiple segments of the ECG that begin after the administration of the electrical shock, and

wherein the multiple segments occur within about 1 minute after the administration of the electrical shock.

19 . A computing device, comprising:

at least one processor; and

memory storing instructions that, when executed by the at least one processor, cause the at least one processor to perform operations comprising:

identifying a first segment of an electrocardiogram (ECG) that ends prior to administration of an electrical shock;

identifying at least one second segment of the ECG that begins after the administration of the electrical shock;

predicting, based on the first segment and the at least one second segment, that the ECG is indicative of a defibrillation-resistant heart rhythm by:

converting the first segment into a first scalogram;

comparing the first scalogram to eigenscalograms:

converting the at least one second segment into at least one second scalogram; and

comparing the at least one second scalogram to eigenscalograms; and

outputting an indication of the predicted defibrillation-resistant heart rhythm.

20 . The computing device of claim 19 , wherein the operations further comprise:

trimming the first scalogram to a first frequency range,

trimming the second scalogram to a second frequency range, and

wherein the first frequency range is broader than the second frequency range.

Assignments (2)

Continuity (2)

References Cited (46)