Food intake monitoring system using apnea detection in breathing signals

Patent metadata

Field	Value
Publication number	US-10327698-B2
Application number	US-201314383983-A
Country	US
Kind code	B2
Filing date	Feb 28, 2013
Priority date	Mar 15, 2012
Publication date	Jun 25, 2019
Grant date	Jun 25, 2019

Abstract

A wearable breathing sensor, such as a piezoelectric chest belt system, generates a breathing signal that is analyzed by a classifier to identify apnea patterns indicating that the subject has swallowed during breathing. These breathing signals are computer-analyzed to extract inferences regarding the subject's eating and drinking patterns and thereby provide useful data for monitoring food or beverage intake for remote health monitoring.

First claim

What is claimed is:

1. A system for monitoring food or beverage intake of a living subject, comprising:
a wearable breathing sensor adapted to be worn around the torso of the subject and being responsive of inhale-exhale movement of the subject's torso to produce a breathing signal expressed as electrical data, wherein the wearable breathing sensor includes (i) a chest-worn belt and (ii) at least one piezoelectric sensor placed between two elastic strips;
a classifier receptive of the electrical data and operative to classify the electrical data according to a predefined set of breathing patterns that include at least one apnea pattern indicating that the subject has swallowed during a breathing cycle;
a swallow pattern analyzer coupled to the classifier and operable to recognize food and beverage intake patterns in the electrical data, wherein the food and beverage intake patterns are associated with eating and drinking different types of food and beverage;
a processor programmed to store the classified electrical data in an associated data storage device;
a food intake analyzer that correlates a record of classified apnea patterns with food and beverage consumption logs entered by or on behalf of the subject; and
a server configured to receive the correlation to enable remote monitoring of the living subject,
wherein:
the classifier is configured to classify the electrical data according to a predefined set of breathing patterns that include:
a normal breathing pattern,
a breathing cycle with exhale swallow pattern, and
a breathing cycle with inhale swallow pattern,

the classifier employs a plurality of matched filters that includes (i) a first matched filter for the normal breathing pattern, (ii) a second matched filter for the breathing cycle with exhale swallow pattern, and (iii) a third matched filter for the breathing cycle with inhale swallow pattern,
the classifier is configured to use the plurality of matched filters to determine a plurality of similarity scores for the electrical data,
each score of the plurality of similarity scores is associated with a pattern of the predefined set of breathing patterns, and
the classifier is configured to classify the electrical data based on the plurality of similarity scores.

2. The system of claim 1 further comprising a signal shaping circuit that processes the breathing signal prior to submission to said classifier.

3. The system of claim 2 wherein said signal shaping circuit employs plural stages selected from the group consisting of impedance matching, drift control, DC damping, low pass filtering and amplification.

4. The system of claim 1 wherein said classifier is implemented by said processor programmed to analyze and classify the electrical data according to a predefined set of trained models stored in the associated data storage device.

5. The system of claim 4 wherein said processor is disposed within a portable electronic device.

6. The system of claim 4 wherein the wearable breathing sensor communicates wirelessly with said processor.

7. The system of claim 1 wherein the classifier is a support vector machine.

8. The system of claim 1 wherein the classifier is a support vector machine based on a polykernel function where the decision boundary is determined by maximizing the geometrical margin of training set data representing solid swallowing and liquid swallowing cases.

9. A method of monitoring food or beverage intake of a living subject, comprising:
placing a breathing sensor around the subject's torso to measure the subject's inhale-exhale movement and generating a breathing signal expressed as electrical data, wherein the breathing sensor includes (i) a chest-worn belt and (ii) at least one piezoelectric sensor placed between two elastic strips;
using electrical circuitry to automatically classify said electrical data according to a predefined set of breathing patterns that include at least one apnea pattern indicating that the subject has swallowed during a breathing cycle;
using the classified electrical data as a measure of the subject's food or beverage intake;
using a processor receptive of the classified electrical data to (i) store the classified electrical data in a data storage device associated with said processor and (ii) analyze the classified data to recognize food and beverage intake patterns associated with eating and drinking different types of food and beverage;
using the processor to correlate a record of classified apnea patterns with food and beverage consumption logs entered by or on behalf of the subject; and
transmitting the correlation to a server to enable remote monitoring of the living subject,
wherein:
the step of classifying said electrical data is performed by classifying the electrical data according to a predefined set of breathing patterns that include:
a normal breathing pattern,
a breathing cycle with exhale swallow pattern, and
a breathing cycle with inhale swallow pattern,

the electrical circuitry employs a plurality of matched filters that includes (i) a first matched filter for the normal breathing pattern, (ii) a second matched filter for the breathing cycle with exhale swallow pattern, and (iii) a third matched filter for the breathing cycle with inhale swallow pattern,
classifying said electrical data includes:
using the plurality of matched filters to determine a plurality of similarity scores for the electrical data, wherein each score of the plurality of similarity scores is associated with a pattern of the predefined set of breathing patterns, and

the classifier is configured to classify the electrical data based on the plurality of similarity scores.

10. The method of claim 9 wherein the electrical circuitry used to automatically classify comprises a processor-implemented classifier utilizing at least one trained model.

11. The method of claim 9 further comprising performing signal shaping on said breathing signal prior to said classifying step.

12. The method of claim 9 wherein analyzing the classified electrical data is performed by comparing the classified electrical data to at least one trained model stored in the associated data storage device.

13. The method of claim 9 further comprising disposing the electrical circuitry to automatically classify said electrical data within the breathing sensor.

14. The method of claim 9 further comprising disposing the electrical circuitry to automatically classify said electrical data within a portable electronic device.

15. The method of claim 9 further comprising wirelessly communicating the electrical data from the breathing sensor to the electrical circuitry to automatically classify said electrical data.

16. The method of claim 9 wherein the electrical circuitry to automatically classify implements a support vector machine.

17. The method of claim 9 wherein the electrical circuitry to automatically classify implements a support vector machine based on a polykernel function where the decision boundary is determined by maximizing the geometrical margin of training set data representing solid swallowing and liquid swallowing cases.

Assignees

• Univ Michigan State

Inventors

• Biswas Subir
• Dong Bo

Classifications

• A61B5/1135

  by monitoring thoracic expansion · CPC title

• A61B5/7246

  using correlation, e.g. template matching or determination of similarity · CPC title

• A61B5/7282

  Event detection, e.g. detecting unique waveforms indicative of a medical condition (cough events A61B5/0823; seizures A61B5/4094; sleep apnoea A61B5/4818) · CPC title

• A61B5/083

  Measuring rate of metabolism by using breath test, e.g. measuring rate of oxygen consumption · CPC title

• A61B5/4866 Primary

  Evaluating metabolism (using breath test A61B5/083) · CPC title