Cough analysis

• US 9,526,458 B2
• Filed: 04/17/2014
• Issued: 12/27/2016
• Est. Priority Date: 09/18/2009
• Status: Active Grant

• Current Assignee: Speech Technology and Applied Research Corporation
• Original Assignee: Speech Technology and Applied Research Corporation
• Inventors: MacAuslan, Joel
• Primary Examiner(s): Kahelin, Michael
• Assistant Examiner(s): Alter, Mitchell E
• Application Number:
  US14/255,436
• Publication Number:
  US 20140343447A1
• Time in Patent Office: 985 Days
• US Class Current: 1/1
• CPC Class Codes: A61B 5/0823 Detecting or evaluating cough events
• CPC Class Codes: A61B 5/7282 Event detection, e.g. detecting unique waveforms indicative of a medical condition A61B5/4094, A61B5/4818, A61B5/349, A61B5/0823 take precedence
• CPC Class Codes: A61B 7/003 Detecting lung or respiration noise
• CPC Class Codes: A61B 7/04 Electric stethoscopes microphones, acoustic transducers therefor H04R

Abstract

A computer-implemented method comprises: (A) receiving first acoustic data representing a first cough train of a first human subject, wherein the first cough train comprises at least one first cough of the first human subject; (B) identifying at least one first value of at one first acoustic property of the first acoustic data; and (C) determining, based on the at least one first value of the at least one first acoustic property, whether the first acoustic data indicates that the first human subject has a severe respiratory illness.

Claims

1. A method performed by at least one computer processor, the method comprising:

(A) receiving first acoustic data representing at least one first cough of a first human subject using a standalone microphone not in contact with the first human subject'"'"'s body;

(B) identifying at least one first value of at least one first acoustic property of the first acoustic data, wherein the at least one first acoustic property comprises at least one of amplitude, temporal structure, and zero-crossing rate;

(C) determining, based on the at least one first value of the at least one first acoustic property, whether the first acoustic data indicates that the first human subject has a severe cough-transmissible respiratory illness, comprising;

(1) extracting, from the first acoustic data, a coughness contour that identifies at least one cough-like region in the first acoustic data;

(2) extracting, from the first acoustic data, contours of local and cumulative properties that are relevant for classifying coughs based on the first acoustic data;

(3) generating a single vector of acoustic data property values from the coughness contour and the contour other than the coughness contour; and

(4) determining, based on the single vector, whether the first acoustic data indicates that the first human subject has a severe cough-transmissible respiratory illness.

2. The method of claim 1, wherein the at least one first cough consists of exactly one cough of the first human subject.

3. The method of claim 1, wherein the at least one first cough comprises a plurality of coughs of the first human subject.

4. The method of claim 1, wherein (C) comprises determining, based on the at least one first value of the at least one first acoustic property, that the first acoustic data indicates that the first human subject has a severe cough-generating respiratory illness.

5. The method of claim 1, wherein (C) comprises determining that the first acoustic data indicates that the first human subject has a severe respiratory illness, and wherein the method further comprises:

(D) receiving second acoustic data representing at least one second cough of a second human subject;

(E) identifying at least one second value of at least one second acoustic property of the second acoustic data;

(F) determining, based on the at least one second value of the at least one second acoustic property, that the second acoustic data indicates that the second human subject does not have a severe respiratory illness.

6. The method of claim 1, further comprising:

(D) before (C), determining, based on at least one second value of at least one second acoustic property, that the first acoustic data represents a cough of the first human subject.

7. The method of claim 6, further comprising:

(E) receiving second acoustic data representing a sound generated by a second human subject;

(F) determining, based on at least one third value of at least one third acoustic property of the second acoustic data, that the second acoustic data does not represent a cough of the second human subject.

8. The method of claim 1, wherein (C) further comprises:

(C) (5) identifying, based on the at least one first value of the at least one first acoustic property, a type of the severe cough-transmissible respiratory illness.

9. The method of claim 8, wherein (C) (5) comprises determining whether the severe cough-transmissible respiratory illness is of a type that can propagate via epidemics.

10. The method of claim 1, wherein the at least one acoustic property comprises a landmark instance identifying a point in the first acoustic data that marks a discrete event.

11. The method of claim 10, wherein the landmark instance comprises an instance of a consonantal landmark.

12. The method of claim 10, wherein the landmark instance comprises an instance of a vocalic landmark.

13. The method of claim 10, wherein (C) comprises identifying, within the first acoustic data, an inspiration phase of a first cough in the at least one first cough, a compression phase of the first cough, and an expiration phase of the first cough.

20. The method of claim 1:

wherein the at least one first acoustic property includes temporal structure and zero-crossing,

wherein the at least one first value of the at least one first acoustic property includes a value of the temporal structure and a value of the zero-crossing; and

wherein determining comprises determining, based on the value of the temporal structure and the value of the zero-crossing, whether the first acoustic data indicates that the first human subject has a severe respiratory illness.

21. The method of claim 1:

wherein the at least one first acoustic property includes amplitude, temporal structure, and zero-crossing rate;

wherein the at least one first value of the at least one first acoustic property includes a value of the amplitude, a value of the temporal structure, and a value of the zero-crossing rate; and

wherein determining comprises determining, based on the value of the amplitude, the value of the temporal structure, and the value of the zero-crossing rate, whether the first acoustic data indicates that the first human subject has a severe respiratory illness.

22. The method of claim 21, wherein the at least one first cough includes a first cough and a second cough;

and

wherein (C) comprises determining whether the first cough and the second cough are separated from each other by at least 50 ms.

23. The method of claim 1, wherein (C) comprises determining whether the onset of the at least one first cough includes a rise of at least 15 dB in at most 12 ms.

24. The method of claim 1, wherein (C) comprises determining whether the onset of the at least one first cough to its maximum is at least 15 ms.

25. The method of claim 1, wherein all broad frequency ranges within 1-7 kHz in the first acoustic data rise and fall together.

26. The method of claim 1, wherein the difference between the maximum over frequency and the median over frequency in the first acoustic data lies in the range 30-45 dB.

27. The method of claim 1, wherein, at each time during the first acoustic data, the standard deviation over frequency is at least 3.5 dB.

28. The method of claim 1, wherein (C) comprises determining whether one-half of the zero-crossing rate is at least 2 kHz for at least 20 ms.

14. A non-transitory computer-readable medium tangibly storing computer program instructions, wherein the computer program instructions are executable by at least one computer processor to perform a method, the method comprising:

(A) receiving first acoustic data representing at least one first cough of a first human subject using a standalone microphone not in contact with the first human subject'"'"'s body;

(B) identifying at least one first value of at least one first acoustic property of the first acoustic data, wherein the at least one first acoustic property comprises at least one of amplitude, temporal structure, and zero-crossing rate;

(C) determining, based on the at least one first value of the at least one first acoustic property, whether the first acoustic data indicates that the first human subject has a severe cough-transmissible respiratory illness, comprising;

(1) extracting, from the first acoustic data, a coughness contour that identifies at least one cough-like region in the first acoustic data;

(2) extracting, from the first acoustic data, contours of local and cumulative properties that are relevant for classifying coughs based on the first acoustic data;

(3) generating a single vector of acoustic data property values from the coughness contour and the contour other than the coughness contour; and

(4) determining, based on the single vector, whether the first acoustic data indicates that the first human subject has a severe cough-transmissible respiratory illness.

15. A method performed by at least one computer processor, the method comprising:

(A) receiving first acoustic data representing at least one first cough of a first human subject using a standalone microphone not in contact with the first human subject'"'"'s body;

(B) identifying at least one first value of at least one first acoustic property of the first acoustic data, wherein the at least one first acoustic property comprises at least one of amplitude, temporal structure, and zero-crossing rate;

(C) determining, based on the at least one first value of the at least one first acoustic property, whether the first acoustic data indicates that the first human subject has an abnormal pulmonary system, comprising;

(1) extracting, from the first acoustic data, a coughness contour that identifies at least one cough-like region in the first acoustic data;

(2) extracting, from the first acoustic data, contours of local and cumulative properties that are relevant for classifying coughs based on the first acoustic data;

(3) generating a single vector of acoustic data property values from the coughness contour and the contour other than the coughness contour; and

(4) determining, based on the single vector, whether the first acoustic data indicates that the first human subject has an abnormal pulmonary system; and

(D) if the first acoustic data is determined to indicate that the first human subject has an abnormal pulmonary system, then determining, based on the at least one first value of the at least one first acoustic property, whether the first acoustic data indicates that the first human subject has a severe cough-transmissible respiratory illness.

16. The method of claim 15, wherein (C) comprises determining, based on the at least one first value of the at least one first acoustic property, that the first acoustic data indicates that the first human subject has a normal pulmonary system.

17. The method of claim 15, wherein the at least one first cough consists of exactly one cough of the first human subject.

18. The method of claim 15, wherein the at least one first cough comprises a plurality of coughs of the first human subject.

19. A non-transitory computer-readable medium tangibly storing computer program instructions, wherein the computer program instructions are executable by at least one computer processor to perform a method, the method comprising:

(A) receiving first acoustic data representing at least one first cough of a first human subject using a standalone microphone not in contact with the first human subject'"'"'s body;

(B) identifying at least one first value of at least one first acoustic property of the first acoustic data, wherein the at least one first acoustic property comprises at least one of amplitude, temporal structure, and zero-crossing rate;

(C) determining, based on the at least one first value of the at least one first acoustic property, whether the first acoustic data indicates that the first human subject has an abnormal pulmonary system, comprising;

(1) extracting, from the first acoustic data, a coughness contour that identifies at least one cough-like region in the first acoustic data;

(2) extracting, from the first acoustic data, contours of local and cumulative properties that are relevant for classifying coughs based on the first acoustic data;

(3) generating a single vector of acoustic data property values from the coughness contour and the contour other than the coughness contour; and

(4) determining, based on the single vector, whether the first acoustic data indicates that the first human subject has an abnormal pulmonary system; and

(D) if the first acoustic data is determined to indicate that the first human subject has an abnormal pulmonary system, then determining, based on the at least one first value of the at least one first acoustic property, whether the first acoustic data indicates that the first human subject has a severe cough-transmissible respiratory illness.