To Adapt Waveforms of Transcutaneous Electrical Stimulation for Effective Relief of Chronic Pain

Chronic pain is a serious disease that affects approximately one in five people worldwide. Transcutaneous electrical stimulation (TENS) can be used to stimulate C-nerve to promote endogenous opioid release, thus is expected to be a new method of chronic pain relief. It is a noninvasive, non-drug therapy that can be used at home. One problem is that TENS waveforms are inevitably deformed by the biological tissues between the skin surface and vicinity of the nerves. In addition, this deformation varies depend on the thickness of the individual's biological tissue layer. This deformation of waveforms might strongly affect the efficiency of chronic pain relief, though has not been taken into consideration by any related studies. In our previous study, a multi-layered tissue FEM (Finite Element Method) model was built to inspect the effect of the deformation of various TENS waveforms. However, the computation cost of FEM is too high to calculate a large amount of candidates to find better TENS waveforms. To solve this problem, this study proposed a machine learning based surrogate model for the FEM model and used it as a forward model to enable a gradient based adaption of TENS waveforms. Simulation results showed that the TENS waveforms taking into consideration the deformation could be found. This demonstrated the potential for adapting the TENS waveforms to individual variations.