Therapy control using motion prediction

1. A medical system, comprising:

a medical device, configured to be controlled according to a control protocol; and

a controller configured to:

receive a measured data set containing a current value at a current time and past values at past times of a physiologic signal measured on a recipient;

calculate a difference signal at the current time based on the measured data set, the difference signal including a first order finite difference of the physiologic signal at the current time and a second order finite difference of the physiologic signal at the current time, the first and second order finite differences respectively corresponding to first and second derivatives;

predict a difference value using the difference signal and a prediction algorithm, the difference value being a difference between a predicted future value of the physiologic signal at a future time and the current value;

calculate the predicted future value based on the difference value and the current value; and

update the control protocol according to the predicted future value.

2. The medical system of claim 1 , wherein the difference signal further includes one or more higher order finite differences of the physiologic signal at the current time, the one or more higher order finite differences having an order higher than the second order finite difference.

3. The medical system of claim 2 , wherein a highest order of the one or more higher order finite differences is determined based on at least one of the physiologic signal, or a time difference between the future time and the current time.

4. The medical system of claim 1 , wherein the controller is configured to calculate the first order finite difference using a first step size and the second order finite difference using a second step size.

5. The medical system of claim 4 , wherein the first step size equals to the second step size.

6. The medical system of claim 4 , wherein the first step size differs from the second step size.

7. The medical system of claim 4 , wherein the first step size and the second step size are determined based on at least one of characteristics of the physiologic signal, an application of the medical system, a measurement of the physiologic signal, or a time difference between the future time and the current time.

8. The medical system of claim 1 , wherein the prediction algorithm includes one of kernel density estimation, support vector regression with scaling, or random forest regression.

9. The medical system of claim 1 , wherein:

the prediction algorithm is trained using a training variable pair comprising a training difference signal and a training difference value;

the training difference signal represents a finite difference of the physiologic signal at a first time point that precedes the current time; and

the training difference value represents a difference between a second value of the physiologic signal at a second time point, that follows the first time point, and a first value of the physiologic signal at the first time point, the second time point preceding the current time.

10. The medical system of claim 1 , wherein the controller is further configured to:

receive historical measured values of the physiologic signal measured at a plurality of time points before predicting the difference value;

calculate a training difference signal at one or more time points based on the historical measured values, to obtain one or more training variable pairs each containing one training difference signal and a corresponding training difference value; and

train the prediction algorithm using the one or more training variable pairs.

11. A method for controlling a medical device, comprising:

receiving a measured data set containing a current value at a current time and past values at past times of a physiologic signal measured on a recipient;

calculating a difference signal at the current time based on the measured data set, the difference signal including a first order finite difference of the physiologic signal at the current time and a second order finite difference of the physiologic signal at the current time, the first and second order finite differences respectively corresponding to first and second derivatives;

predicting a difference value using the difference signal and a prediction algorithm, the difference value being a difference between a predicted future value of the physiologic signal at a future time and the current value;

calculating the predicted future value based on the difference value and the current value; and

updating a control protocol of the medical device according to the predicted future value.

12. The method of claim 11 , wherein the difference signal further includes one or more higher order finite differences of the physiologic signal at the current time, the one or more higher order finite differences having an order higher than the second order finite difference.

13. The method of claim 12 , further comprising:

determining a highest order of the one or more higher order finite differences based on at least one of the physiologic signal, an application of the medical device, or a time difference between the future time and the current time.

14. The method of claim 11 , wherein calculating the difference signal includes calculating the first order finite difference using a first step size and calculating the second order finite difference using a second step size.

15. The method of claim 14 , wherein the first step size equals to the second step size.

16. The method of claim 14 , wherein the first step size differs from the second step size.

17. The method of claim 14 , further comprising:

determining the first step size and the second step size based on at least one of characteristics of the physiologic signal, an application of the medical device, a measurement of the physiologic signal, or a time difference between the future time and the current time.

18. The method of claim 11 , wherein predicting the difference value uses one of kernel density estimation, support vector regression with scaling, or random forest regression.

19. The method of claim 11 , wherein:

the prediction algorithm is trained using a training variable pair comprising a training difference signal and a training difference value;

the training difference signal represents a finite difference of the physiologic signal at a first time point that precedes the current time; and

the training difference value represents a difference between a second value of the physiologic signal at a second time point, that follows the first time point, and a first value of the physiologic signal at the first time point, the second time point preceding the current time.

20. The method of claim 11 , further comprising training the prediction algorithm, wherein training the prediction algorithm includes:

receiving historical measured values of the physiologic signal measured at a plurality of time points before predicting the difference value;

calculating a training difference signal at one or more of the time points based on the historical measured values, to obtain one or more training variable pairs each containing one training difference signal and a corresponding training difference value; and

training the prediction algorithm using the one or more training variable pairs to obtain parameter values of the prediction algorithm.