Multi-parametric analysis of snore sounds for the community screening of sleep apnea with non-Gaussianity index

1. A method for diagnosing a sleeping disorder of a subject including:

processing a digitized audio signal from the subject with at least one electronic processor, said processing including:

estimating values for one or more parameters of said audio signal;

applying the values to a predetermined diagnostic model; and

indicating the presence of a sleeping disorder based on the output of said model;

wherein the one or more parameters includes a non-Gaussianity index quantifying deviation from Gaussianity distribution of said audio signal, and wherein the method further includes:

dividing the digitized audio signal into a number of segments;

calculating non-Gaussianity scores for each of the segments; and

computing the non-Gaussianity index based on said scores.

2. A method according to claim 1 , wherein the calculating of the non-Gaussianity scores includes calculating a normal probability of each said segment of the digitized audio signal.

3. A method according to claim 1 , further comprising comparing the non-Gaussianity scores to a predetermined threshold value.

4. A method according to claim 3 , wherein the computing of the non-Gaussianity index based on said scores includes taking into account the comparison of the scores to the predetermined threshold value.

5. A method according to claim 1 , wherein the one or more parameters further include one or more of: pitch parameters, higher order spectrum parameters, cepstral coefficients and slice parameters.

6. A method according to claim 5 , wherein the cepstral coefficients are mel-frequency cepstral coefficients.

7. A method according to claim 5 , wherein the slice parameters are diagonal slice parameters.

8. A method according to claim 5 , wherein the higher order spectrum parameters are bispectrum parameters.

9. A method according to claim 1 , wherein the processing of the digitized audio signal from the subject includes dividing the digitized signal into a plurality of segments.

10. A method according to claim 9 , including identifying said segments indicating voiced snoring prior to the step of estimating values for the one or more parameters.

11. An apparatus for diagnosing a sleeping disorder of a subject, including:

at least one digital signal input interface configured to receive a digitized audio signal of the subject;

at least one electronic processor in communication with the digital signal input interface;

a user display viewable by an operator of the apparatus; and

at least one electronic memory in communication with the electronic processor, said electronic memory containing tangible instructions executable by the said electronic processor, said instructions configured at least:

to detect snore segments of the digitized audio signal;

to estimate values for a plurality of parameters of the snore segments, said parameters including a parameter quantifying deviation from Gaussianity distribution of said signal;

to apply the values to a predetermined diagnostic model; and

to indicate the presence of a sleeping disorder based on the model upon the user display.

12. An apparatus for diagnosing a sleeping disorder of a subject, including:

at least one audio signal input interface configured to receive an audio signal of a sleeping subject;

an analog to digital converter coupled to the audio signal input interface configured to create a digitized audio signal;

one or more sleep-related parameter calculation modules responsive to the digitized audio signal and configured to produce signals indicating values for said sleep-related parameters;

a classification module responsive to the signals indicating values for said sleep-related parameters and configured to classify the audio signal as either sleep dysfunctional or non-sleep dysfunctional; and

a display assembly responsive to the classification module and configured to display said classification;

wherein the one or more sleep-related parameter calculation modules include a module configured to calculate a parameter indicating deviation from Gaussianity distribution of the digitized audio signal.

13. An apparatus according to claim 12 , wherein the assembly includes a data logger to record the digitized audio signal and to record the signals indicating values for said sleep-related parameters.

14. An apparatus according to claim 12 , wherein the assembly includes a segmentation module configured to segment the digitized audio signal, said segmentation module being disposed in a signal path between the analog to digital converter and the one or more sleep-related parameter calculation modules.

15. An apparatus according to claim 12 , wherein the sleep-related parameter calculation modules further include one or more of: a total airways response calculator; a pitch calculator; and a cepstral coefficient calculator.

16. An apparatus according to claim 12 , wherein the module configured to calculate a parameter indicating deviation from Gaussiantity distribution of the digitized audio signal comprises a non-Gaussianity Index (NGI) calculator.

17. An apparatus according to claim 16 , further including a pre-emphasis module coupled to an input side of the NGI calculator configured to filter the digitized audio signal prior to processing by the NGI calculator.

18. A computer software product comprising a program storage device, readable by machine, tangibly embodying a program of non-transitory instructions executable by the machine to cause the machine to perform a method according to claim 1 .

19. A method according to claim 1 , wherein the calculating of the non-Gaussianity index comprises measuring a deviation of a plot of said audio signal from a straight line, said straight line being a segment of a reference Gaussian probability plot.

20. An apparatus according to claim 16 , wherein the module is configured to calculate the non-Gaussianity index by measuring a deviation of a plot of said audio signal from a straight line, said straight line being a segment of a reference Gaussian probability plot.