Machine learning to optimize spinal cord stimulation
View Patent ↗An example of a system may include a processor and a memory device comprising instructions, which when executed by the processor, cause the processor to: access a patient metric of a subject; use the patient metric as an input to a machine learning algorithm, the machine learning algorithm to search a plurality of neuromodulation parameter sets and to identify a candidate neuromodulation parameter set of the plurality of neuromodulation parameter sets, the candidate neuromodulation parameter set designed to produce a non-regular waveform that varies over a time domain and a space domain; and program a neuromodulator using the candidate neuromodulation parameter set to stimulate the subject.
1. A system comprising:
a processor; and
a memory device comprising instructions, which when executed by the processor, cause the processor to:
access a patient pain metric of a subject;
use the patient pain metric as an input to an algorithm that is configured for use to identify a candidate neuromodulation parameter set for a use by a neuromodulator to deliver and a non-regular electrical waveform for neuromodulation using electrodes, wherein the non-regular electrical waveform varies over a time domain including variations in at least one of timing, size or shape of the non-regular electrical waveform and the non-regular electrical waveform varies over a space domain including variations in distribution of modulation energy fractionalized across the electrodes; and
enable the neuromodulator to treat patient pain using the candidate neuromodulation parameter set to electrically stimulate the subject using the non-regular electrical waveform that varies over the time domain and that varies over the space domain including the variations in the distribution.
2. The system of claim 1 , wherein the subject is a patient.
3. The system of claim 1 , wherein the subject is an animal from a preclinical trial.
4. The system of claim 1 , wherein the algorithm comprises a genetic algorithm.
5. The system of claim 1 , wherein the algorithm comprises a neural network.
6. The system of claim 1 , wherein the system is a cloud-based system, and wherein the instructions to provide the candidate neuromodulation parameter set comprise instructions to transmit the candidate neuromodulation parameter set to a client device of the cloud-based system.
7. The system of claim 1 , wherein the patient pain metric is an objective pain metric.
8. The system of claim 7 , wherein the objective pain metric is a physiological indication sensed by a sensor worn by the subject.
9. The system of claim 1 , wherein the patient pain metric is a subjective pain metric.
10. The system of claim 9 , wherein the subjective pain metric is obtained from querying the subject.
11. The system of claim 1 , wherein the space domain includes at least one parameter related to fractionalization or polarity.
12. The system of claim 1 , wherein the candidate neuromodulation parameter set is designed to produce a pulse burst that varies pulse-by-pulse.
13. The system of claim 1 , wherein the algorithm is a machine learning algorithm, and the memory device further comprises instructions, which when executed by the processor, cause the processor to:
receive from a user, a selected parameter; and
focus the machine learning algorithm using the selected parameter.
14. A method comprising:
accessing, at a computerized system, a patient pain metric of a subject;
using the patient pain metric as an input to an algorithm executing on the computerized system that is configured for use to identify a candidate neuromodulation parameter set for a use by a neuromodulator to deliver a non-regular electrical waveform for neuromodulation using electrodes, wherein the non-regular waveform varies over a time domain including variations in at least one of timing, size or shape of the waveform and the non-regular waveform varies over a space domain including variations in distribution of modulation energy fractionalized across the electrodes;
use the candidate neuromodulation parameter set to enable the neuromodulator to electrically stimulate the subject using the non-regular electrical waveform that varies over the time domain and that varies over the space domain including the variations in the distribution; and
using the neuromodulator to produce the non-regular waveform that varies over the time domain and the space domain.
15. The method of claim 14 , wherein the subject is a patient.
16. The method of claim 14 , wherein the subject is an animal from a preclinical trial.
17. The method of claim 14 , wherein the algorithm comprises a genetic algorithm.
18. The method of claim 14 , wherein the candidate neuromodulation parameter set is designed to produce a pulse burst that varies pulse-by-pulse.
19. The method of claim 14 , further comprising:
receiving from a user, a selected parameter; and
focusing the algorithm using the selected parameter.
20. A non-transitory machine-readable medium including instructions, which when executed by a machine, cause the machine to:
access a patient pain metric of a subject;
use the patient pain metric as an input to an algorithm that is configured for use to identify a candidate neuromodulation parameter set for a use by a neuromodulator to deliver a non-regular electrical waveform for neuromodulation using electrodes, wherein the a non-regular electrical waveform varies over a time domain including variations in at least one of timing, size or shape of the waveform and the non-regular waveform varies over a space domain including variations in distribution of modulation energy fractionalized across the electrodes; and
enable the neuromodulator to treat pain using the candidate neuromodulation parameter set by electrically stimulating the subject using the non-regular electrical waveform that varies over the time domain and that varies over the space domain including the variations in the distribution.