Acoustical patient monitoring using a sound classifier and a microphone

• US 10,269,228 B2
• Filed: 05/29/2009
• Issued: 04/23/2019
• Est. Priority Date: 06/17/2008
• Status: Active Grant

• Current Assignee: Koninklijke Philips Electronics N.V. (Koninklijke Philips N.V.)
• Original Assignee: Koninklijke Philips N.V.
• Inventors: Aarts, Ronald M.
• Primary Examiner(s): Dougherty, Sean P
• Application Number:
  US12/996,034
• Publication Number:
  US 20110087079A1
• Time in Patent Office: 3,616 Days
• Field of Search: 600529, 600586
• CPC Class Codes: A61B 5/4806 Sleep evaluation A61B5/4821 takes precedence; devices for inducing sleep A61M21/02
• CPC Class Codes: A61B 7/003 Detecting lung or respiration noise
• CPC Class Codes: G08B 21/043 detecting an emergency event, e.g. a fall
• CPC Class Codes: G08B 21/0461 integrated or attached to an item closely associated with the person but not worn by the person, e.g. chair, walking stick, bed sensor
• CPC Class Codes: G16H 40/63 for local operation
• CPC Class Codes: G16H 40/67 for remote operation

Abstract

When monitoring a patient, acoustic events (e.g., coughs, snores, impact sounds, verbalizations, etc.) relevant to the patient'"'"'s status are detected by a microphone array (12) and timestamped. Detected event signals generated by the microphone array (12) are filtered to identify signatures such as zero crossings, corner frequencies, amplitude, pitch, etc., for classification purposes. The filtered signals are digitized and classified into one of a plurality of acoustic event classes (e.g., snore, cough, wheeze, breath, etc.) and/or subclasses. The classified events are displayed to a user (e.g., graphically, textually, etc.) with their timestamps to indicate chronology. A user can review the acoustic events, select one or more events, and listen to a recording of the selected event (s). Additionally, specified acoustic events can trigger an alarm to alert a nurse or the like that the patient requires immediate attention.

Claims

1. A patient monitoring system, including:

one or more microphones configured to detect acoustic events generated by a patient, and generate signals comprising information describing the acoustic events;

a processor configured to timestamp the acoustic event signals with a timestamp;

a classifier configured to classify each acoustic event signal into one of a plurality of acoustic event classes;

a memory configured to store computer-executable instructions, which when executed by the processor, cause the processor to;

filter the acoustic event signal and identifying signatures associated with the acoustic event signal prior to digitizing the acoustic event signal;

digitize the acoustic event signal;

as a function of the identified signatures, classify the acoustic event signal into;

one of a plurality of acoustic event classes; and

one of a plurality of subclasses of the one of the plurality of acoustic event classes; and

trigger an alarm that the patient may have fallen out of a bed by;

(i) determining an impact sound; and

(ii) determining that there has been an absence of an acoustic event for a predetermined duration following the impact sound; and

a user interface configured to display;

(i) the timestamp and a textual description of one or more classified acoustic events on a display; and

(ii) a playback icon for selecting and playing a recorded audio file of the acoustic event.

2. The system according to claim 1, wherein:

the plurality of acoustic event classes includes a cough class; and

the cough class includes subclasses of at least one of;

a wheezing cough subclass;

a productive cough subclass;

a gargled cough subclass; and

a dry cough subclass.

3. The system according to claim 1, wherein:

the plurality of acoustic event classes includes a cough class; and

the cough class includes subclasses of all of;

a wheezing cough subclass;

a productive cough subclass;

a gargled cough subclass; and

a dry cough subclass.

4. The system according to claim 1, further including:

an active noise cancellation component configured to suppress non-patient-generated acoustic event sounds.

5. The system according to claim 1, wherein the acoustic event classes include classes for patient-generated sounds including at least one of a yell, groan, impact sound, belch, or verbal speech.

6. The system according to claim 1, further including one or more filters that detects acoustic event signatures associated with the acoustic event signals, the one or more filters includes a bandpass filter;

wherein the acoustic event signatures include all of;

average zero crossing time, amplitude, frequency, one or more corner frequencies, and pitch of the acoustic event signals.

7. The system according to claim 1, wherein the acoustic event classes include classes for patient-generated sounds including all of a yell, groan, impact sound, and belch.

8. A method of monitoring a patient, including:

detecting, with one or more microphones, an acoustic event generated by the patient;

timestamping, with a timestamp, an acoustic event signal that includes information related to the acoustic event;

filtering the acoustic event signal to identify signatures associated with the acoustic event signal prior to digitizing the acoustic event signal;

digitizing the acoustic event signal;

as a function of the identified signatures, classifying the acoustic event signal into;

one of a plurality of acoustic event classes; and

one of a plurality of subclasses of the one of the plurality of acoustic event classes;

storing each classified acoustic event with the timestamp; and

triggering an alarm that the patient may have fallen out of a bed by;

(i) determining an impact sound; and

(ii) determining that there has been an absence of an acoustic event for a predetermined duration following the impact sound.

9. The method according to claim 8,

the plurality of acoustic event classes includes a cough class;

and

the cough class includes subclasses of at least one of;

a wheezing cough subclass;

a productive cough subclass;

a gargled cough subclass; and

a dry cough subclass.

10. The method according to claim 8, further including:

using active noise cancellation to suppress ambient sounds when detecting the acoustic event.

11. A patient monitoring system comprising a non-transitory computer-readable medium including computer readable instructions, which when executed by a processor, cause the processor to:

detect an acoustic event generated by a patient;

using information from an accelerometer, verify that the acoustic event was generated by the patient;

timestamp an acoustic event signal that includes information related to the acoustic event;

filter the acoustic event signal to identify signatures associated with the acoustic event signal prior to digitizing the acoustic event signal;

digitize the acoustic event signal;

classify the acoustic event into one of a plurality of acoustic event classes as a function of the identified signatures;

store a classified acoustic event with the timestamp;

group acoustic events together into a group, wherein each event within the group;

has a same class as other events of the group; and

is contiguous with another event of the group; and

control a display to display the group to a user.

12. The patient monitoring system of claim 11, wherein the computer readable instructions, which when executed by a processor, further cause the processor to:

control a display to display an icon corresponding to the group, the icon allowing the user to listen to an entire period of the group.

13. The patient monitoring system of claim 11, wherein the computer readable instructions, which when executed by a processor, further cause the processor to:

control a display to allow the user to select a textual description of the group to view individual events of the group.

14. The patient monitoring system of claim 11, wherein the computer readable instructions, which when executed by a processor, further cause the processor to:

train the processor to classify acoustic events for the patient by, upon an occurrence of a specific acoustic event, receiving an input from a user that identifies the specific acoustic event as belonging to a specific event class.

15. The patient monitoring system of claim 11, wherein the computer readable instructions, which when executed by a processor, further cause the processor to:

control the display the allow the user to scroll vertically and horizontally to view a list of acoustic events.

16. A method of monitoring a patient, including:

detecting, with one or more microphones, an acoustic event generated by a patient;

using information from an accelerometer, verifying that the acoustic event was generated by the patient;

timestamping an acoustic event signal that includes information related to the acoustic event;

filtering the acoustic event signal to identify signatures associated with the acoustic event signal prior to digitizing the acoustic event signal;

digitizing the acoustic event signal;

classifying the acoustic event into one of a plurality of acoustic event classes as a function of the identified signatures;

storing a classified acoustic event with the timestamp;

grouping acoustic events together into a group, wherein each event within the group;

has a same class as other events of the group; and

is contiguous with another event of the group; and

controlling a display to display the group to a user by displaying an icon corresponding to the group, the icon allowing the user to listen to an entire period of the group;

wherein the acoustic event classes include classes for patient-generated sounds including all of a yell, groan, impact sound, and belch.

17. The patient monitoring method of claim 16, further comprising:

controlling a display to allow the user to select a textual description of the group to view individual events of the group.