Detection of extreme hypoglycemia or hyperglycemia based on automatic analysis of speech patterns

• US 7,925,508 B1
• Filed: 08/22/2006
• Issued: 04/12/2011
• Est. Priority Date: 08/22/2006
• Status: Expired due to Fees

• Current Assignee: Avaya Incorporated
• Original Assignee: Avaya Incorporated
• Inventors: Michaelis, Paul Roller
• Primary Examiner(s): Wozniak, James S
• Assistant Examiner(s): KOVACEK, DAVID M
• Application Number:
  US11/508,477
• Time in Patent Office: 1,694 Days
• Field of Search: 704/231, 704234-240, 704243-255, 704/268, 704/270, 704/272, 704/274, 704E51001-E5105, 704E11001-E11007, 379/52, 37920101-2011, 37926501-2661, 381/54, 600/300, 600/481, 600/529, 600531-543
• US Class Current: 704/270
• CPC Class Codes: A61B 5/14532 for measuring glucose, e.g. by tissue impedance measurement
• CPC Class Codes: A61B 5/4803 Speech analysis specially adapted for diagnostic purposes
• CPC Class Codes: G10L 17/26 Recognition of special voice characteristics, e.g. for use in lie detectors; Recognition of animal voices

Abstract

In one embodiment, the present invention is directed to a communication device that analyzes received utterances and detects when the speaker has a target physical condition.

Claims

1. A method for determining a physical state of a user, comprising:

receiving via communication device a voice utterance from a first speaker in a communication between the first speaker and a second speaker;

analyzing the voice utterance to determine whether the first speaker likely has one of hypoglycemia or hyperglycemia, wherein if first speaker if speaking rapidly the first speaker is determined to be experiencing hypoglycemia and wherein if the first speaker is speaking lethargically, the first speaker is determined to be experiencing hyperglycemia;

when the first speaker likely has one of hypoglycemia or hyperglycemia, at least one of;

(i) sending to the user a notification associated with one of hypoglycemia or hyperglycemia, wherein the notification alerts the first speaker as to which of hypoglycemia or hyperglycemia the first speaker is suffering, and

(ii) sending to a third party a notification associated with one of hypoglycemia or hyperglycemia, wherein the notification provides which of hypoglycemia or hyperglycemia the first speaker is suffering.

2. The method of claim 1, wherein analyzing the voice utterance comprises performing acoustic-phonetic analysis of the voice utterance to determine a likelihood that the first speaker has the target physical condition or a likely severity of the target physical condition.

3. The method of claim 1, wherein the acoustic-phonetic analysis analyzes, with reference to a baseline utterance of the first speaker, at least one of a formant of the utterance, a pitch of the utterance, a speech or articulation rate of the utterance, a vocal intensity of the utterance, a number of speech errors in the utterance, a response time, a nonfluency level associated with the utterance, and a speech quality of the utterance.

4. The method of claim 1, wherein the likelihood that the first speaker has the target physical condition and likely severity of the target physical condition are determined, wherein step (ii) is performed when the likelihood and likely severity are at least a first set of thresholds and wherein step (i) but not step (ii) is performed when at least one of the likelihood and likely severity is less than the first set of thresholds but more than a second set of thresholds.

5. The method of claim 1, wherein analyzing the voice utterance comprises:

comparing speech units in the utterance with corresponding baseline speech units from the first speaker assuming that the first speaker does not have the target physical condition;

comparing speech units in the utterance with corresponding baseline speech units from the first speaker assuming that the first speaker has the target physical condition; and

based on the comparative results of substeps (b1) and (b2), determining a level of confidence that the first speaker has or does not have the target physical condition.

6. A non-transitory computer readable medium comprising thereon processor executable instructions to perform the steps of claim 1.

7. A communication device, comprising:

a user diagnostic functionality operable to;

(a) receive, as part of a voicemail, a voice utterance from a speaker;

(b) analyze the voice utterance to determine whether the speaker likely has one of hypoglycemia or hyperglycemia, wherein if the speaker is speaking rapidly, the first speaker is determined to be experiencing hypoglycemia, and wherein if the first speaker is speaking lethargically, the first speaker is determined to be experiencing hyperglycemia; and

(c) when the speaker likely has one of hypoglycemia or hyperglycemia, at least one of;

(i) send to the user a notification associated with one of hypoglycemia or hyperglycemia, and

(ii) send, to a third party, a notification associated with one of hypoglycemia or hyperglycemia.

8. The communication device of claim 7, wherein operation (b) comprises the sub-operation:

(b1) performing acoustic-phonetic analysis of the voice utterance to determine a likelihood that the speaker has the target physical condition or a likely severity of the target physical condition.

9. The communication device of claim 7, wherein the acoustic-phonetic analysis analyzes, with reference to a baseline utterance of the speaker, at least one of a formant of the utterance, a pitch of the utterance, a speech or articulation rate of the utterance, a vocal intensity of the utterance, a number of speech errors in the utterance, a response time, a nonfluency level associated with the utterance, and a speech quality of the utterance.

10. The communication device of claim 7, wherein the likelihood that the speaker has the target physical condition and likely severity of the target physical condition are determined, wherein step (ii) is performed when the likelihood and likely severity are at least a first set of thresholds and wherein step (i) but not step (ii) is performed when at least one of the likelihood and likely severity is less than the first set of thresholds but more than a second set of thresholds.

11. The communication device of claim 7, wherein operation (b) comprises the sub-operations:

(b1) comparing speech units in the utterance with corresponding baseline speech units from the speaker assuming that the speaker does not have the target physical condition;

(b2) comparing speech units in the utterance with corresponding baseline speech units from the speaker assuming that the speaker has the target physical condition; and

(b3) based on the comparative results of sub-operations (b1) and (b2), determining a level of confidence that the speaker has or does not have the target physical condition.

12. A method, comprising:

receiving, by a communication device, a voice utterance from a speaker;

sending the voice utterance to a communication node; and

analyzing, by a communication node, the voice utterance to determine whether the speaker likely has one of hypoglycemia or hyperglycemia, wherein if the speaker is speaking rapidly it is determined that the speaker is experiencing hypoglycemia, and wherein if the speaker is speaking lethargically it is determined that the speaker is experiencing hyperglycemia.

13. The method of claim 12, wherein the voice utterance is received during a live conversation with another party and wherein, when the speaker likely has hypoglycemia and hyperglycemia, at least one of (i) sending to the user a notification associated with the hyperglycemia or hypoglycemia and (ii) sending to a third party a notification associated with the hyperglycemia or hypoglycemia.

14. The method of claim 13, wherein analyzing the voice utterance comprises performing acoustic-phonetic analysis of the voice utterance to determine a likelihood that the speaker has hyperglycemia or hypoglycemia or a likely severity of the hyperglycemia or hypoglycemia.

15. The method claim 14, wherein the acoustic-phonetic analysis analyzes, with reference to a baseline utterance of the speaker, at least one of a formant of the utterance, a pitch of the utterance, a speech or articulation rate of the utterance, a vocal intensity of the utterance, a number of speech errors in the utterance, a response time, a nonfluency level associated with the utterance, and a speech quality of the utterance.

16. The method of claim 13, wherein analyzing the voice utterance comprises:

comparing speech units in the utterance with corresponding baseline speech units from the speaker assuming that the speaker does not have hyperglycemia or hypoglycemia;

comparing speech units in the utterance with corresponding baseline speech units from the speaker assuming that the speaker has hyperglycemia or hypoglycemia; and

based on the comparative results of the above comparing sub-steps, determining a level of confidence that the speaker has or does not have hyperglycemia or hypoglycemia.

17. The method of claim 13, wherein the likelihood that the speaker has hyperglycemia or hypoglycemia and likely severity of the hyperglycemia or hypoglycemia are determined, wherein step (ii) is performed when the likelihood and likely severity are at least a first set of thresholds and wherein step (i) but not step (ii) is performed when at least one of the likelihood and likely severity is less than the first set of thresholds but more than a second set of thresholds.

18. A non-transitory computer readable medium comprising thereon processor-executable instructions to perform the steps of claim 12.