Arrhythmia detection using hidden regularity to improve specificity

• US 8,897,863 B2
• Filed: 05/02/2013
• Issued: 11/25/2014
• Est. Priority Date: 05/03/2012
• Status: Expired due to Fees

• Current Assignee: University of Washington
• Original Assignee: University of Washington
• Inventors: Linker, David Thor
• Primary Examiner(s): Lavert, Nicole F
• Application Number:
  US13/886,240
• Publication Number:
  US 20130296680A1
• Time in Patent Office: 572 Days
• Field of Search: 600508-509, 600515-516
• US Class Current: 600/515
• CPC Class Codes: A61B 5/346 Analysis of electrocardiograms
• CPC Class Codes: A61B 5/361 Detecting fibrillation
• CPC Class Codes: A61B 5/7221 Determining signal validity, reliability or quality A61B5/7207 and A61B5/7217 take precedence

Abstract

A method for identifying a target arrhythmia, for example, atrial fibrillation, with very high specificity includes obtaining heart rhythm data such as EKG data, selecting and analyzing a segment of the data for arrhythmia, if an arrhythmia is found then reanalyzing the segment of data for a hidden regularity that would indicate the data is not the target arrhythmia. If no hidden regularity is found, then identifying the segment as the target arrhythmia.

Claims

1. A method for detecting a target arrhythmia in heart rhythm data with high specificity, comprising:

obtaining heart rhythm data generated by monitoring the activity of a heart with a medical monitoring device;

selecting a predetermined number of contiguous heart beat intervals from the heart rhythm data as an analysis segment;

analyzing the analysis segment to identify an irregularity in the heart beat intervals that satisfies a threshold test for arrhythmia;

if an irregularity that satisfies the threshold test is identified, then reanalyzing the analysis segment to identify any hidden regularity in the heart beat intervals that would indicate absence of the target arrhythmia; and

if the analysis segment was reanalyzed and no hidden regularity was identified, then designating the analysis segment as the target arrhythmia and storing the analysis segment and the designation on a non-transitory computer readable medium.

2. The method of claim 1, wherein the heart rhythm data comprises electrocardiograph data.

3. The method of claim 1, wherein the medical monitoring device comprises a wearable monitor.

4. The method of claim 3, wherein the wearable monitor comprises a plurality of electrodes configured to be adhesively attached to the subject, a power supply, a microprocessor configured to receive data from the plurality of electrodes, and a data storage device.

5. The method of claim 3, wherein the wearable monitor further comprises a QRS detector.

6. The method of claim 1, wherein the predetermined number of contiguous heart beat intervals is less than twenty-two heart beat intervals.

7. The method of claim 1, wherein the step of reanalyzing the analysis segment comprises (i) calculating a combinatorial property of the heart beat intervals in the analysis segment;

and (ii) comparing the calculated combinatorial property with a threshold value to determine if the hidden regularity is present.

8. The method of claim 7, wherein the threshold value is determined by (i) analyzing a database of heart rhythm data for which the arrhythmia properties of the heart rhythm data is known;

(ii) calculating an inaccuracy rate associated with comparing the calculated combinatorial property with a range of threshold values; and

(iii) selecting the threshold value that minimizes the calculated inaccuracy rate.

9. The method of claim 1, wherein the step of reanalyzing the analysis segment comprises (i) calculating the absolute value of the time difference between each sequential pair of intervals in the analysis segment;

and (ii) comparing the maximum time difference with a threshold value to determine if the hidden regularity is present.

10. The method of claim 1, wherein the step of reanalyzing the analysis segment comprises (i) calculating the time difference between each sequential pair of intervals in the analysis segment;

and (ii) comparing the maximum time difference with a threshold value to determine if the hidden regularity is present.

11. The method of claim 1, wherein the step of reanalyzing the analysis segment comprises (i) calculating the time difference between the longest interval in the analysis segment and the shortest interval in the analysis segment (“

MaxDiff”

);

(ii) calculating the average time difference between each sequential pair of intervals in the analysis segment for which the time difference is greater than a predetermined fraction of MaxDiff; and

(iii) comparing the calculated average time difference with a threshold value to determine if the hidden regularity is present.

12. The method of claim 1, wherein the step of reanalyzing the analysis segment comprises (i) calculating an average interval length for each two-interval sequence in the analysis segment (“

dvals”

) and an average interval length for each three-interval sequence in the analysis segment (“

tvals”

);

(ii) calculating an interval length for each two-interval sequence in the analysis segment (“

dvals2”

) and calculating one half of the interval length for each three-interval sequence in the analysis segment;

(iii) calculating the difference between all unique combinations of dvals, tvals, dvals2, and tvals2 (“

diffvals”

);

(iv) summing a predetermined number of the smallest diffvals values; and

(v) comparing the summed value with a predetermined threshold to determine if the hidden regularity is present.

13. The method of claim 1, wherein the step of reanalyzing the analysis segment comprises (i) calculating the sum of the absolute difference between all pairs of intervals in the analysis segment that are separated by one interval plus the sum of the absolute difference between all pairs of intervals in the analysis segment that are separated by two intervals;

(ii) summing a predetermined number of the smallest calculated sums; and

(iii) comparing the sum of the predetermined number of the smallest calculated sums with a predetermined threshold to determine if the hidden regularity is present.

14. The method of claim 1, wherein the step of reanalyzing the analysis comprises (i) calculating the average value of the absolute value of second order differences in the analysis segment;

and (ii) comparing the calculated second order difference with a predetermined threshold to determine if the hidden regularity is present.

15. The method of claim 1, wherein the step of reanalyzing the analysis segment comprises applying two or more of the steps described in claims 9-14 sequentially.

16. A method of detecting atrial fibrillation in heart rhythm data from a heart, comprising:

monitoring the electrical activity of a heart using a wearable detector comprising a plurality of electrodes, a microprocessor in signal communication with the electrodes, and a data storage component to obtain a heart rhythm data set;

selecting a segment comprising a predetermined number of contiguous heart beat intervals from the heart rhythm data set;

analyzing the segment of contiguous heart beat intervals to identify any irregularity in the segment of the heart that satisfies a threshold test for arrhythmia;

if an irregularity is identified in the segment, then reanalyzing the segment to identify any hidden regularity that would indicate the absence of atrial fibrillation; and

if the segment was reanalyzed and no hidden regularity was identified, then designating the segment as atrial fibrillation and storing the segment of the heart rhythm data on the data storage component.

17. The method of claim 16, wherein the heart rhythm data set comprises electrocardiograph data.

18. The method of claim 16, wherein the step of reanalyzing the segment of the heart rhythm data comprises (i) calculating a combinatorial property of the interval data in the segment;

and (ii) comparing the calculated combinatorial property with a threshold value to determine if the hidden regularity is present.

19. The method of claim 16, wherein the number of contiguous heart beat intervals in an analysis segment is not more than seven heart beat intervals.