Approximate entropy

• US 5,191,524 A
• Filed: 09/08/1989
• Issued: 03/02/1993
• Est. Priority Date: 09/08/1989
• Status: Expired due to Term

• Current Assignee: Robert A. Neidorff, Steven M. Pincus
• Original Assignee: Robert A. Neidorff, Steven M. Pincus
• Inventors: Neidorff, Robert A., Pincus, Steven M.
• Primary Examiner(s): Envall, Jr., Roy N.
• Assistant Examiner(s): Huntley, David
• Application Number:
  US07/404,737
• Time in Patent Office: 1,271 Days
• Field of Search: 364/413.02, 364/413.05, 364/413.06, 364/413.03, 128/703, 128/706
• US Class Current: 600/508
• CPC Class Codes: A61B 5/35 by template matching
• CPC Class Codes: G16H 40/63 for local operation
• CPC Class Codes: G16Z 99/00 Subject matter not provided for in other main groups of this subclass
• CPC Class Codes: Y10S 128/923 by comparison of patient data to other data

Abstract

An approximation of entropy is determined on a set of data by comparing subsets of the data. The comparison reveals the regularity and stability of similar patterns amongst subsets of the data. The comparisons perform such that the contribution of noise to measurement of the regularity and stability is minimized. Quantitative values are assigned to measure the degree of regularity and stability. From these quantitative values a single output measure is generated indicative of the amount of patternness of the sequence of data. The calculations required to determine this approximate entropy are preferably performed within a data processing system. Numerous peripheral devices may be attached to such a data processing system. The types of data for which the approximate entropy may be calculated include any sets of data wherein the amount of patternness is sought.

Claims

1. A method for processing a set of medical data with a data processing system to determine a relative amount of patterness of the set of medical data comprising the steps of:

converting measurements of a medical parameter into a set of medical data having values representative of the measured medical parameter;

storing the set of medical data in the memory of a data processing system,

operating on the stored set of medical data with a processor of the data processing system, comprising the operating steps of;

a) defining a class of contiguous runs of prescribed length of the set of medical data;

b) comparing each element of this class to a plurality of elements of this class to determine regularity and stability of similar patterns among the elements of this class, wherein contribution of noise to determination of regularity and stability is minimized;

c) assigning and the stability of similar patterns among the elements of this class of medical data as determined by said comparing step;

d) providing an output signal indicative of a relative amount of patterness of the set of medical data based on an aggregate of the assigned quantitative values;

e) comparing the output signal to a predetermined value of patterness to produce a comparison result; and

performing a corrective medical procedure dependent on the comparison result.

2. A method as recited in claim 1 wherein the output measure of patterness of the medical data is an approximation of entropy of the medical data.

3. A method as recited in claim 1 wherein the contribution of noise to the output signal is minimized because the comparing step includes a noise filter.

4. A method as recited in claim 1 further comprising the step of converting the medical data into digital form before performing the comparing step.

5. A method as recited in claim 1 wherein the medical data is electrocardiograph (EKG) data.

6. A method as recited in claim 1 wherein the medical data is electroencephalograph (EEG) data.

7. A method as recited in claim 1 wherein the medical data is endocrine hormone secretion data.

8. A method as recited in claim 1 wherein the medical data is electro-oculogram (EOG) data.

9. A method as recited in claim 1 wherein the medical data is electromyogram (EMG) data.

10. A method as recited in claim 1 wherein the medical data is respiratory measurements.

11. A method for processing electrocardiograph (EKG) data with a data processing system to measure a relative amount of patterness in the set of electrocardiograph (EKG) beat-to-beat heart rate data, comprising the steps of:

converting an analog EKG signal into digital EKG data having values representative of the analog EKG signal;

storing the EKG data in the memory of a data processing system,

operating on the stored EKG data with a processor of the data processing system, comprising the operating steps of;

a) extracting R-R intervals from the EKG data;

b) averaging R-R intervals for a length of time to produce a set of R-R interval averages;

c) defining a class of contiguous runs of prescribed length of the set of R-R interval averages;

d) comparing each element of this class to a plurality of the elements of this class to determine regularity and stability of similar patterns among the elements of this class, wherein contribution of noise to determination of regularity and stability is minimized;

e) assigning quantitative values to measure the regularity and the stability of similar patterns among the elements of this class as determined by the comparing step;

f) computing a single measure of the relative amount of patterness in the EKG data based on an aggregate of the assigned quantitative values;

g) comparing the single measure of the relative amount of patterness in the EKG data to a predetermined value of patterness to produce a comparison result; and

performing a corrective medical procedure dependent on the comparison result.

12. A method as recited in claim 11 wherein the single measure of patterness is an approximation of entropy.

13. A method as recited in claim 11 wherein the contribution of noise to the single measure is minimized because the comparing step includes a noise filter.

14. A method for processing a set of hormone secretion level data with a data processing system to measure the relative amount of patterness in the set of hormone secretion level data, comprising the steps of:

converting measurements of the level of hormone secretion over a period of time into a set of hormone secretion level data having values representative of the measurements of hormone secretion;

storing the set of hormone secretion level data in the memory of a data processing system, and

operating on the stored set of hormone secretion level data with a processor of the data processing system, comprising the operating steps of;

a) defining a class of continuous runs of prescribed length of the hormone secretion level data;

b) comparing each element of this class to a plurality of the elements of this class to determine regularity and stability of similar patterns among the elements of this class, wherein contribution of noise to determination of regularity and stability is minimized;

c) assigning quantitative values to measure the regularity and the stability of similar patterns among the elements of this class as determined by the comparing step;

d) computing a single measure of the relative amount of patterness of the hormone secretion data based on an aggregate of the assigned quantitative values;

e) comparing the single measure of the relative amount of patterness in the hormone secretion data to a predetermined value of patterness to produce a comparison result; and

performing a corrective medical procedure dependent on the comparison result.

15. A method as recited in claim 14 wherein the single measure of patterness is an approximation of entropy.

16. A method as recited in claim 14 wherein the contribution of noise to the measurement of regularity and stability is minimized because the comparing step includes a noise filter.

17. A medical monitor for monitoring heart activity of a patient, comprising:

a) an electrocardiograph for monitoring the heart activity of the patient and producing a corresponding set of signals of said heart activity;

b) a processor for receiving the set of signals of heart activity and processing the signals wherein said processor comprises;

1) means for defining a class of contiguous runs of prescribed length of the set of signals of heart activity;

2) comparision means for comparing each element of this class to a plurality of the elements of this class to determine regularity and stability of similar patterns among the elements of this class, wherein contributions of noise to determination of regularity and stability is minimized;

3) assignment means for assigning quantitative values to measure the regularity and the stability of similar patterns among the elements of this class of the set of signals of heart activity as determined by the comparing of the elements of this class; and

4) output means for providing an output signal indicative of a relative amount of patterness of the set of signals of heart activity based on an aggregate of the assigned quantitative values; and

c) an output device for receiving and responding to the output signal and for providing a corrective medical procedure dependent on the output signal.

18. A medical monitor as recited in claim 17 wherein the output device includes an alarm which sounds when the output signal indicates a low degree of patterness in the set of signals of heart activity.

19. A medical monitor as recited in claim 17 wherein the output device includes a meter that displays an amount of patterness that is indicated by the output signal in the set of signals of heart activity.

20. A medical monitor as recited in claim 17 wherein the output device includes a recording means for recording an amount of patterness that is indicated by the output signal in the set of signals of heart activity.

21. A medical monitor as recited in claim 17 wherein the output device includes an automated corrective medical procedure means that executes a corrective medical procedure if the output signal indicates a degree of patterness in the set of signals of heart activity.

22. A data processing system for operating on a sequence of data comprising:

storage means for storing the sequence of data in the memory of a data processing system,

processing means for operating on the stored sequence of data, the processing means comprising;

a) an assigning means for defining a class of contiguous runs of prescribed length of the sequence of data;

b) a comparing means for comparing each element of this class to a plurality of the elements of this class to determine regularity and stability of similar patterns among the element of this class wherein said comparing means assigns quantitative values to an extent of regularity and stability of similar patterns among elements of this class;

c) a filtering means for minimizing effects of noise from the sequence of data from the values that are compared; and

d) a computational means for using said quantitative values produced by said comparing means to compute a single output approximation of entropy; and

automated dispensing means for automatically dispensing medication in response to the approximation of entropy.