Food intake monitor
AssigneeThe Board of Trustees of The University of Alabama
InventorsEdward Sazonov, Muhammad Farooq
Life of this patent
Estimating the term…
Abstract
Systems and methods for monitoring food intake include an air pressure sensor for detecting ear canal deformation, according to some implementations. For example, the air pressure sensor detects a change in air pressure in the ear canal resulting from mandible movement. Other implementations include systems and methods for monitoring food intake that include a temporalis muscle activity sensor for detecting temporalis muscle activity, wherein at least a portion of the temporalis muscle activity sensor is coupled adjacent a temple portion of eyeglasses and disposed between the temple tip and the frame end piece. The temporalis muscle activity sensor may include an accelerometer, for example, for detecting movement of the temple portion due to mandibular movement from chewing.
Claims20
A computer-implemented method for monitoring food intake, comprising:
- using a sensor to measure temporalis muscle activity;
- transmitting sensor data to a computer;
- using the computer to execute computer-implemented instructions that determine whether the temporalis muscle activity is associated with food intake by performing steps comprising:
- dividing the sensor data into epochs of a selected duration;
- computing a feature set for each epoch, wherein the respective feature sets comprise Fourier Transforms of the sensor data in each epoch;
- using a respective feature set, computing power spectral density coefficients for each epoch;
- compiling a predictor vector for each epoch with the power spectral density coefficients;
- using the respective predictor vectors as inputs to a multi-layer perception artificial neural network to classify the epochs as containing a swallow or not.
A system for monitoring food intake, comprising:
- a sensor to measure temporalis muscle activity of a subject;
- a computer receiving sensor data transmitted from the sensor;
- computer-implemented instructions stored on the computer that, when executed, determine whether the temporalis muscle activity is associated with food intake by performing steps comprising:
- dividing the sensor data into epochs of a selected duration;
- computing a feature set for each epoch, wherein the respective feature sets comprise Fourier Transforms of the sensor data in each epoch;
- using a respective feature set, computing power spectral density coefficients for each epoch;
- compiling a predictor vector for each epoch with the power spectral density coefficients;
- and
- using the respective predictor vectors as inputs to a multi-layer perception artificial neural network to classify the epochs as containing a swallow or not.
Dependencies are inferred from the claim text.