Acoustic system and methodology for identifying the risk of obstructive sleep apnea during wakefulness

1. A method for screening a patient for obstructive sleep apnea during wakefulness comprising:

(a) positioning a sound input device adjacent a suprasternal notch of a trachea of the patient;

(b) recording tracheal breath sounds for at least five breathing cycles of the patient for each of nose breathing and mouth breathing with the patient in an upright position and at least five breathing cycles of the patient for each of nose breathing and mouth breathing with the patient in a supine position;

(c) estimating power spectrum density (PSD) and bispectrum of the at least 5 breathing cycles for each of nose breathing and mouth breathing in the upright and the supine positions;

(d) computing characteristic breathing sound features for the patient from the estimated PSD and the bispectrum of step (c), wherein the computed characteristic breathing sound features comprise a bispectral invariant P(a) defined by

P ( a )= a tan( I i ( a )/ I r ( a ))

with at least one characteristic breathing sound feature selected from the group consisting of the bispectral invariant at 15° for inspiration phase of mouth breathing in upright posture, the bispectral invariant at 24° for inspiration phase of mouth breathing in upright posture, and the bispectral invariant at 27° for inspiration phase of mouth breathing in upright posture;

(e) comparing the value of the computed characteristic breathing sound features for the patient to mean values for corresponding features from a non-Obstructive Sleep Apnea (non-OSA) training set and an Obstructive Sleep Apnea (OSA) training set;

(f) classifying the patient as non-OSA or OSA based on the comparison of step (e); and

(g) communicating to a user the classification of the patient as non-OSA or OSA.

2. The method of claim 1 , further comprising at least four of the computed characteristic breathing sound features selected from the group consisting of: spectral centroid of mouth breathing in supine position, spectral centroid of mouth breathing in upright position, spectral bandwidth of nose breathing in upright position, spectral bandwidth of nose breathing in supine position, spectral bandwidth of mouth breathing in supine position, normalized bispectral squared entropy of mouth breathing in supine position, bispectral invariant of mouth breathing in upright position, relative signal power in upright position, spectral centroid in supine position, spectral centroid in upright position, normalized bispectral entropy in supine position, normalized bispectral entropy in upright position, bispectral invariant in upright position, signal power of mouth breathing in supine position, signal power of nose breathing in upright position, relative signal power of nose breathing in upright position, spectral centroid of nose breathing in upright position, spectral centroid of mouth breathing in upright position, median bi-frequency of nose breathing in upright position, and bispectral invariant of nose breathing in upright position.

3. The method of claim 1 , wherein the computed characteristic breathing sound features further comprise:

I. Spectral centroid of PSD for the band of 450-600 Hz, for expiration phase of mouth breathing in upright posture

II. Spectral bandwidth for the band of 1200-1800 Hz, for expiration phase of mouth breathing in supine posture

III. Difference between nose and mouth breathing in spectral centroid for the band of 100-2500 Hz, for expiration phase in supine posture

IV. Difference between supine and upright postures in signal power for the band of 100-150 Hz, for inspiration phase of mouth breathing.

4. The method of claim 1 , further using age, gender, BMI, Mallampati score, neck circumference, and smoking history of the patient to aid in classifying the patient as OSA or non-OSA.

5. The method of claim 1 , further comprising confirming the classification of step (f) by subjecting the patient to polysomnography.

6. The method of claim 1 , wherein the estimated PSD is in the frequency range of 100 Hz to 2500 Hz.

7. The method of claim 1 , wherein the estimated bispectrum is in the frequency range of 100 Hz to 2500 Hz.

8. The method of claim 1 , wherein the computed characteristic breathing sound features are signal power, spectral centroid, spectral bandwidth, spectral flatness and crest factor, and are computed over a frequency range of 100 Hz to 2500 Hz.

9. The method of claim 1 , wherein the computed characteristic breathing sound features are signal power, spectral centroid and spectral bandwidth, and are computed in each of 100 to 150 Hz, 150 to 450 Hz, 450 to 600 Hz, 600 to 1200 Hz, 1200 to 1800 Hz and 1800 to 2500 Hz bandwidths.

10. The method of claim 1 , wherein the at least 5 breathing cycles for each of nose breathing and mouth breathing in the upright position and the at least 5 breathing cycles for each of nose breathing and mouth breathing in the supine position are separated into inspiration and expiration cycles.