Granted Patent B2

US 12,564,526 · App. 16/233,239 · Granted Mar 3, 2026

Inventors: Ramazan Demirli (San Jose, CA); Omid Sayadi (San Jose, CA); Shawn Barr (San Jose, CA)

Assignee: Sleep Number Corporation

A61G7/018A47C21/003A61B5/0816A61B5/4806A61B5/4809A61B5/4812A61G7/015G06V40/10G16Y20/10G16Y40/10A47C27/083A47C31/008A61B5/0823A61B5/0826A61B5/11A61B5/4818A61B5/6891A61B5/6892A61B5/7267A61B2562/0204A61B2562/0247A61G7/05769G06V40/15

Abstract

Claims (83)

1 . A bed system comprising:

a mattress to support a user;

an acoustic sensor configured to sense acoustic energy in an environment of the user;

a pressure sensor configured to sense pressure applied to the mattress by the user laying on the mattress; and

a controller configured to:

receive an acoustic stream from the acoustic sensor, the acoustic stream representing acoustic energy sensed by the acoustic sensor;

receive a pressure stream from the pressure sensor, the pressure stream representing pressure sensed by the pressure sensor;

combine the acoustic stream and the pressure stream in order to generate a set of snore/breath parameters, wherein to combine the acoustic stream and the pressure stream, the controller is further configured to:

construct scaled and normalized values of acoustic and pressure data epochs based on the pressure stream and the acoustic stream, wherein the scaled and normalized values of acoustic and pressure data epochs are distributed across a common numerical range to provide balanced weighing of the acoustic stream and the pressure stream using a deep learning network; and

provide the scaled and normalized values of the acoustic and pressure data epochs as an input to the deep learning network and operate the deep learning network to output a prediction for the set of snore/breath parameters;

determine that a home-automation rule includes a condition that includes the set of snore/breath parameters; and

responsive to determining that a home-automation rule includes a condition that includes the set of snore/breath parameters, send an instruction to drive a controllable device to the controllable device.

2 . The bed system of claim 1 , further comprising the controllable device, the controllable device configured to:

receive the instruction to drive the controllable device; and

responsive to receiving the instruction to drive the controllable device, drive in order to alter the environment of the user.

3 . The bed system of claim 1 , wherein to combine the acoustic stream and the pressure stream, the controller is further configured to:

determine respiratory parameters based on instantaneous pressure signals;

determine respiratory parameters based on instantaneous acoustic signals; and

identify incidences of matching of both the respiratory parameters based on instantaneous pressure signals and of the respiratory parameters based on instantaneous acoustic signals.

4 . The bed system of claim 1 , wherein to combine the acoustic stream and the pressure stream, the controller is further configured to:

determine that the user is present in on the mattress;

determine that the user is asleep; and

responsive to determining that the user is present in on the mattress and that the user is asleep, generate a vote to represent a candidate set of snore/breath parameters.

5 . The bed system of claim 1 , wherein to combine the acoustic stream and the pressure stream, the controller is further configured to:

apply the acoustic stream and the pressure stream to one or more fusion algorithms that each are configured to, responsive to receiving both an acoustic stream and a pressure stream, generate a vote to represent a candidate set of snore/breath parameters that describe snoring and breathing action of the user on the mattress;

tally the votes in order to determine a winning set of snore/breath parameters; and

select the winning set of snore/breath parameters as the set of snore/breath parameters.

6 . The bed system of claim 1 , wherein at least one of the acoustic sensor and the pressure sensor comprise:

a signal conditioner; and

a digitizer;

wherein the acoustic stream is a digital stream of data for the acoustic sensor, and wherein the pressure stream is a digital stream of data for the pressure sensor.

7 . The bed system of claim 1 , wherein to combine the acoustic stream and the pressure stream in order to generate the set of snore/breath parameters, the controller is configured to:

receive both the acoustic stream and the pressure stream into a single buffer;

after receiving both the acoustic stream and the pressure stream into a single buffer, normalize the acoustic stream in a particular domain;

after receiving both the acoustic stream and the pressure stream into a single buffer, separately normalize the pressure stream in the particular domain; and

fuse the normalized pressure stream and the normalized acoustic stream to compute the set of snore/breath parameters.

8 . The bed system of claim 1 , wherein to combine the acoustic stream and the pressure stream in order to generate the set of snore/breath parameters, the controller is configured to:

receive the acoustic stream into an acoustic buffer;

receive the pressure stream into a pressure buffer separate from the acoustic buffer;

estimate acoustic parameters in a particular domain;

estimate pressure parameters in the particular domain; and

fuse the estimated acoustic parameters with the estimated pressure parameters.

9 . The bed system of claim 1 , wherein combine the acoustic stream and the pressure stream in order to generate the set of snore/breath parameters, the controller is configured to:

compute acoustic features in a particular domain;

compute pressure features in another particular domain; and

apply machine learning classifiers to both the acoustic features and the pressure features, to compute the snore/breathing parameters.

10 . A controller configured to:

receive an acoustic stream from an acoustic sensor configured to sense acoustic energy in an environment of a user, the acoustic stream representing acoustic energy sensed by the acoustic sensor;

receive a pressure stream from a pressure sensor, the pressure stream representing pressure sensed by the pressure sensor configured to sense pressure applied to a mattress by the user laying on the mattress;

combine the acoustic stream and the pressure stream in order to generate a set of snore/breath parameters, wherein to combine the acoustic stream and the pressure stream, the controller is further configured to:

construct scaled and normalized values of acoustic and pressure data epochs based on the pressure stream and the acoustic stream, wherein the scaled and normalized values of acoustic and pressure data epochs are distributed across a common numerical range to provide balanced weighing of the acoustic stream and the pressure stream using a deep learning network; and

provide the scaled and normalized values of the acoustic and pressure data epochs as an input to the deep learning network and operate the deep learning network to output a prediction for the set of snore/breath parameters;

determine that a home-automation rule includes a condition that includes the set of snore/breath parameters; and

responsive to determining that a home-automation rule includes a condition that includes the set of snore/breath parameters, send an instruction to drive a controllable device to the controllable device.

11 . The controller of claim 10 , wherein to combine the acoustic stream and the pressure stream, the controller is further configured to:

determine respiratory parameters based on instantaneous pressure signals;

determine respiratory parameters based on instantaneous acoustic signals; and

identify incidences of matching of both the respiratory parameters based on instantaneous pressure signals and of the respiratory parameters based on instantaneous acoustic signals.

12 . The controller of claim 10 , wherein to combine the acoustic stream and the pressure stream, the controller is further configured to:

identify features of potential snores and breaths from the acoustic stream;

identify features of potential snores and breaths from the pressure stream; and

identify agreements of the potential snores and breaths from the acoustic stream and of the potential snores and breaths from the pressure stream.

13 . The controller of claim 10 , wherein to combine the acoustic stream and the pressure stream, the controller is further configured to:

determine that the user is present in on the mattress;

determine that the user is asleep; and

responsive to determining that the user is present in on the mattress and that the user is asleep, generate a vote to represent a candidate set of snore/breath parameters.

14 . The controller of claim 10 , wherein to combine the acoustic stream and the pressure stream, the controller is further configured to:

apply the acoustic stream and the pressure stream to one or more fusion algorithms that each are configured to, responsive to receiving both an acoustic stream and a pressure stream, generate a vote to represent a candidate set of snore/breath parameters that describe snoring and breathing action of the user on the mattress;

tally the votes in order to determine a winning set of snore/breath parameters; and

select the winning set of snore/breath parameters as the set of snore/breath parameters.

15 . The controller of claim 10 , wherein to combine the acoustic stream and the pressure stream in order to generate the set of snore/breath parameters, the controller is configured to:

receive both the acoustic stream and the pressure stream into a single buffer;

after receiving both the acoustic stream and the pressure stream into a single buffer, normalize the acoustic stream in a particular domain;

after receiving both the acoustic stream and the pressure stream into a single buffer, separately normalize the pressure stream in the particular domain; and

fuse the normalized pressure stream and the normalized acoustic stream to optimally determine the set of snore/breath parameters.

16 . The controller of claim 10 , wherein combine the acoustic stream and the pressure stream in order to generate the set of snore/breath parameters, the controller is configured to:

compute acoustic features in a particular domain;

compute pressure features in the same or another particular domain; and

apply machine learning classifiers to both the acoustic features and the pressure features to optimally determine the set of snore/breathing parameters.

17 . The controller of claim 10 , wherein the deep learning network is trained based on labeled outcomes for training acoustic and pressure data, and wherein the training includes iteratively adjusting parameters of the deep learning network to minimize loss between the labeled outcomes for the training acoustic and pressure data and corresponding predictions output from the deep learning network.

18 . The controller of claim 10 , wherein the scaled and normalized values of acoustic and pressure data epochs are reweighed to adjust an influence of the acoustic and pressure data epochs before the scaled and normalized values of acoustic and pressure data epochs are provided as the input to the deep learning network.

19 . The bed system of claim 1 , wherein responsive to combining the acoustic stream and the pressure stream, the controller is configured to detect which of a first user laying on the mattress or a second user laying on the mattress is snoring.

20 . The bed system of claim 1 , wherein the controller is configured to provide a balanced weighing of the acoustic stream and the pressure stream using the deep learning network when the controller operates the deep learning network to output the prediction for the set of snore/breath parameters.

Assignments (3)

Continuity (2)

References Cited (124)