Evaluating pain of a user via time series of parameters from portable monitoring devices

Inventors

Rezai, Ali • Finomore, Victor

Abstract

Systems and methods are provided for evaluating pain for a user. A first pain-relevant parameter representing the user is monitored at an in-vivo sensing device over a defined period to produce a time series for the first pain-relevant parameter. A value for a second pain-relevant parameter for the user is obtained at first and second times in the defined period from the user via a portable computing device to provide respective first and second values for the second pain-relevant parameter. A value is assigned to the user via a predictive model according to the time series for the first pain-relevant parameter, the first value for the second pain-relevant parameter, and the second value for the second pain-relevant parameter.

Core Innovation

The invention relates to an objective pain-evaluation and pain-treatment system that assigns a value representing one of a predicted level of pain for a user at a future time and a current level of pain for the user. The system monitors a first pain-relevant parameter representing the user at an in-vivo sensing device or wearable device over a defined period to produce a time series, and obtains a value for a second pain-relevant parameter at first and second times in the defined period via a portable computing device to provide respective first and second values. The assigned value is produced by a recurrent neural network according to the time series and the first and second values.

The invention further includes receiving a self-reported pain level from the user, comparing the self-reported pain level to the value assigned via the recurrent neural network, and changing a parameter associated with the recurrent neural network. The parameter comprises one of an individualized threshold used to produce categorical outputs for the user and a baseline value for at least one biological rhythm of the user, and in some implementations the parameter is changed via a reinforcement learning process according to a comparison of the measured outcome to the value assigned.

The invention also actuates a worn or implanted therapeutic device associated with the user when the value representing the predicted level of pain for the user or the current level of pain for the user exceeds the individualized threshold. The document describes optional wavelet decomposition and wavelet coefficients for the time series, and an optional facial-expression classification and feature extraction that considers departures from established patterns of biological rhythms.

Claims Coverage

The partial content includes three independent claims: one method claim and two system claims. Each independent claim centers on a recurrent neural network that assigns a pain value from a time series of a first pain-relevant parameter and user-provided values for a second pain-relevant parameter, with feedback updating network-associated parameters and actuation of a therapeutic device when an individualized threshold is exceeded.

Across the independent claims, the invention uses a recurrent neural network that assigns a pain value from a time series derived from an in-vivo sensing or wearable first pain-relevant parameter and first and second values of a second pain-relevant parameter obtained at user input times. Feedback compares the assigned value to a self-reported pain level and changes network-associated parameters that include an individualized threshold and/or a baseline value for at least one biological rhythm. A worn or implanted therapeutic device is actuated when the assigned pain value exceeds the individualized threshold.

Stated Advantages

Documented Applications

No documented applications found