Spinal cord stimulation system using neural dose determination optimized sub-aware therapy

Abstract

A spinal cord stimulation system using neural dose determination optimized sub-aware therapy is disclosed. In one example, an external device includes an algorithm configured to determine a stimulation procedure for a stimulator device. The algorithm includes a model that includes a predetermined energy value that causes a sub-perceived stimulus. The algorithm is configured to determine a stimulation parameter for the stimulation procedure that produces an energy value within the first model. The energy value in the model may be represented as a function of frequency. The model provides optimized sub-perceptual stimulation particularly at low frequencies such as 1 kHz and below, even 400 Hz and below.

Description

[0001] This application is a divisional application of the patent application with application number "202080048202.7", application date "July 1, 2020", and titled "Spinal Cord Stimulation System for Subsensory Therapy Using Neural Dose Determination Optimization". Technical Field

[0002] The present application relates to implantable medical devices (IMDs), generally to spinal cord stimulators, and more particularly to methods of controlling such devices. Background Art

[0003] Implantable neurostimulator devices are devices that generate and deliver electrical stimulation to body nerves and tissues for the treatment of various biological disorders, such as pacemakers for treating cardiac arrhythmias, defibrillators for treating cardiac fibrillation, cochlear stimulators for treating deafness, retinal stimulators for treating blindness, muscle stimulators for producing coordinated limb movements, spinal cord stimulators for treating chronic pain, cortical and deep brain stimulators for treating movement and psychological disorders, and other neurostimulators for treating urinary incontinence, sleep apnea, shoulder subluxation, etc. The following description will generally focus on the use of the present invention within a spinal cord stimulation (SCS) system, such as disclosed in U.S. Patent 6,516,227. However, the present invention may find applicability to any implantable neurostimulator device system.

[0004] SCS systems typically include Figure 1 1 and 1 . An implantable pulse generator (IPG) 10 is shown in FIG. The IPG 10 includes a biocompatible device housing 12 that houses circuitry and a battery 14 required for the operation of the IPG. The IPG 10 is coupled to electrodes 16 via one or more electrode leads 15 that form an electrode array 17. The electrodes 16 are configured to contact the patient's tissue and are carried on a flexible body 18, which also houses an independent lead wire 20 coupled to each electrode 16. The lead wires 20 are also coupled to proximal contacts 22, which can be inserted into a lead connector 24 within a header 23 secured to the IPG 10, which header may include, for example, epoxy. Once inserted, the proximal contacts 22 are connected to header contacts within the lead connector 24, which in turn are coupled to circuitry within the housing 12 by feedthrough pins through a housing feedthrough, although these details are not shown.

[0005] In the illustrated IPG 10, there are sixteen lead electrodes (E1 to E16) split between two leads 15, with the head 23 containing a 2x1 array of lead connectors 24. However, the number of leads and electrodes in an IPG is application specific and can therefore vary. The conductive housing 12 may also include electrodes (Ec). In SCS applications, the electrode leads 15 are typically implanted near the dura mater in the patient's spine on both sides of the midline of the spinal cord. The proximal electrodes 22 are tunneled through the patient's tissue to a distant location such as the buttocks where the IPG housing 12 is implanted, at which point they are coupled to the lead connector 24. In other IPG examples designed for direct implantation at a site requiring stimulation, the IPG may be leadless, with 16 electrodes optionally present on the body of the IPG for contacting the patient's tissue. The IPG leads 15 may be integrated with the housing 12 in other IPG solutions and permanently connected to the housing 12. The purpose of SCS therapy is to provide electrical stimulation from the electrodes 16 to relieve the patient's symptoms, most notably chronic back pain.

[0006] The IPG 10 may include an antenna 26a that allows it to communicate bidirectionally with a plurality of external devices, such as Figure 4 As shown in Figure 1 The antenna 26a depicted in FIG. 1 is shown as a conductive coil within the housing 12, although the coil antenna 26a may also be present in the head 23. When the antenna 26a is configured as a coil, communication with the external device preferably occurs using near-field magnetic induction. The IPG may also include a radio frequency (RF) antenna 26b. Figure 1 , the RF antenna 26b is shown in the head 23, but it can also be in the housing 12. The RF antenna 26b may include a patch, a slot, or a wire, and may operate as a single pole or a dipole. The RF antenna 26b preferably communicates using far-field electromagnetic waves. The RF antenna 26b may operate according to any number of known RF communication standards, such as Bluetooth, Zigbee, WiFi, and MICS, and the like.

[0007] Stimulation in the IPG 10 is typically provided by pulses, such as Figure 2 . The stimulation parameters typically include: the amplitude of the pulse (A; whether current or voltage); the frequency (F) and pulse width (PW) of the pulse; the electrodes 16 (E) that are activated to provide such stimulation; and the polarity (P) of such active electrodes, i.e., whether the active electrodes act as anodes (source current to tissue) or cathodes (sink current from tissue). These stimulation parameters, considered together, comprise a stimulation program that the IPG 10 can execute to provide therapeutic stimulation to a patient.

[0008] exist Figure 2 In the example of , electrode E5 has been selected as the anode, and thus provides a pulse that pulls a positive current of amplitude +A toward the tissue. Electrode E4 has been selected as the cathode, and thus provides a pulse that injects a corresponding negative current of amplitude -A from the tissue. This is an example of bipolar stimulation, in which only two lead-based electrodes (one anode, one cathode) are used to provide stimulation to the tissue. However, more than one electrode can serve as the anode at a given time, and more than one electrode can serve as the cathode at a given time (e.g., tripolar stimulation, quadrupolar stimulation, etc.).

[0009] like Figure 2 The pulses shown in are biphasic, including a first phase 30a, followed by a second phase 30b with opposite polarity. It is well known that the use of biphasic pulses is useful in effective charge recovery. For example, the current path to each electrode of the tissue may include a DC blocking capacitor connected in series, see, for example, U.S. Patent Application Publication 2016 / 0144183, which will be charged during the first phase 30a and discharged (restored) during the second phase 30b. In the example shown, the first phase 30a and the second phase 30b have the same duration and amplitude (although opposite polarity), which ensures the same amount of charge during the two phases. However, if the integral of the amplitude and duration of the two phases is equal in magnitude, the second phase 30b can also be charged balanced with the first phase 30a, as is well known. The width PW of each pulse is defined here as the duration of the first pulse phase 30a, although the pulse width can also refer to the total duration of the first pulse phase 30a and the second pulse phase 30b. Note that an interphase period (IP) can be provided between the two phases 30a and 30b, during which no stimulation is provided.

[0010] The IPG 10 includes a stimulation circuit 28 that can be programmed to generate stimulation pulses at the electrodes as defined by the stimulation program. The stimulation circuit 28 can, for example, include circuits described in U.S. Patent Application Publications 2018 / 0071513 and 2018 / 0071520 or described in USP8,606,362 and 8,620,436. These documents are incorporated herein by reference.

[0011] Figure 3An external test stimulation environment is shown that can be used before the IPG 10 is implanted in the patient. During the external test stimulation, stimulation can be attempted to the intended implant patient without implanting the IPG 10. In contrast, one or more test leads 15' are implanted in the patient's tissue 32 at the target location 34, such as in the spine as previously described. The proximal ends of the one or more test leads 15' leave the incision 36 and are connected to an external test stimulator (ETS) 40. The ETS 40 typically mimics the operation of the IPG 10, and can therefore provide stimulation pulses to the patient's tissue, as described above. See, for example, 9,259,574, which discloses a design for the ETS. The ETS 40 is typically worn externally by the patient for a short period of time (e.g., two weeks), which allows the patient and his or her clinician to experiment with different stimulation parameters to try and find a stimulation program that relieves the patient's symptoms (e.g., pain). If the external trial stimulation proves successful, the trial lead(s) 15' are removed and the complete IPG 10 and lead(s) 15 are implanted as described above; if unsuccessful, only the trial lead(s) 15' are removed.

[0012] Similar to IPG 10, ETS 40 may include one or more antennas to enable two-way communication with external devices. Figure 4 Further explanation. Such antennas may include near-field magnetic induction coil antenna 42a, and / or far-field RF antenna 42b, as previously described. ETS 40 may also include stimulation circuitry 44, which is capable of forming stimulation pulses according to a stimulation program, which circuitry may be similar to or include the same stimulation circuitry 28 present in IPG 10. ETS 40 may also include a battery (not shown) for operating power.

[0013] Figure 4 Various external devices are shown that can wirelessly communicate data with the IPG 10 and ETS 40, including a patient's handheld external controller 45 and a clinician programmer 50. Both devices 45 and 50 can be used to send stimulation programs to the IPG 10 or ETS 40—that is, to program its stimulation circuits 28 and 44 to produce pulses with desired shapes and timing as previously described. Both devices 45 and 50 can also be used to adjust one or more stimulation parameters of a stimulation program currently being executed by the IPG 10 or ETS 40. Devices 45 and 50 can also receive information from the IPG 10 or ETS 40, such as various status information, etc.

[0014] The external controller 45 may be as described, for example, in U.S. Patent Application Publication 2015 / 0080982, and may include any dedicated controller configured to work with the IPG 10. The external controller 45 may also include a general-purpose mobile electronic device such as a mobile phone that has been programmed with a medical device application (MDA) that allows it to operate as a wireless controller for the IPG 10 or ETS 40, as described in U.S. Patent Application Publication 2015 / 0231402. The external controller 45 includes a user interface that includes means for inputting commands (e.g., buttons or icons) and a display 46. The user interface of the external controller 45 enables the patient to adjust stimulation parameters, although it may have limited functionality when compared to a more powerful clinician programmer 50, described later.

[0015] The external controller 45 may have one or more antennas capable of communicating with the IPG 10 and the ETS 40. For example, the external controller 45 may have a near-field magnetic induction coil antenna 47a that is capable of wirelessly communicating with the coil antenna 26a or 42a in the IPG 10 or ETS 40. The external controller 45 may also have a far-field RF antenna 47b that is capable of wirelessly communicating with the RF antenna 26b or 42b in the IPG 10 or ETS 40.

[0016] The external controller 45 may also have control circuitry 48, such as a microprocessor, microcomputer, FPGA, other digital logic structure capable of executing instructions for an electronic device, etc. The control circuitry 48 may, for example, receive patient adjustments to stimulation parameters and create a stimulation program to be wirelessly transmitted to the IPG 10 or ETS 40.

[0017] The clinician programmer 50 is further described in U.S. Patent Application Publication 2015 / 0360038 and is only briefly described here. The clinician programmer 50 may include: a computing device 51, such as a desktop, a portable computer, a laptop, a tablet computer, a mobile smart phone, a personal digital assistant (PDA) type mobile computing device, etc. Figure 4 In FIG. 5 , computing device 51 is shown as a portable computer including typical computer user interface devices such as screen 52, mouse, keyboard, speaker, stylus, printer, etc., which are not all shown for convenience. Figure 4 Also shown are accessory devices for the clinician programmer 50 that are typically specific to its operation as a stimulation controller, such as a communications "wand" 54, and a joystick 58, which may be coupled to appropriate ports on the computing device 51, such as, for example, a USB port 59.

[0018] The antenna used in the clinician programmer 50 for communicating with the IPG 10 or ETS 40 can depend on the type of antennas included in those devices. If the patient's IPG 10 or ETS 40 includes a coil antenna 26a or 42a, the wand 54 can likewise include a coil antenna 56a for establishing near-field magnetic induction communications at close range. In this example, the wand 54 can be secured in close proximity to the patient, such as by placing it in a belt or sleeve that can be worn by the patient and near the patient's IPG 10 or ETS 40.

[0019] If the IPG 10 or ETS 40 includes an RF antenna 26 b or 42 b, the wand 54, computing device 41, or both may likewise include an RF antenna 56 b to establish communications with the IPG 10 or ETS 40 at greater distances. (The wand 54 may not be necessary in this case.) The clinician programmer 50 may also establish communications with other devices and networks (such as the Internet) wirelessly or via a wired link provided at an Ethernet or network port.

[0020] To program the stimulation program or parameters of the IPG 10 or ETS 40, the clinician interfaces with a clinician programmer graphical user interface (GUI) 64 provided on the display 52 of the computing device 51. As will be appreciated by those skilled in the art, the GUI 64 may be presented by executing clinician programmer software 66 on the computing device 51, which may be stored on the non-volatile memory 68 of the device. Those skilled in the art will additionally appreciate that the execution of the clinician programmer software 66 in the computing device 51 may be facilitated by a control circuit 70, such as a microprocessor, a microcomputer, an FPGA, other digital logic structures capable of executing programs in a computing device, etc. Such control circuit 70, in addition to executing the clinician programmer software 66 and presenting the GUI 64, may also enable communication via the antenna 56a or 56b to transmit the selected stimulation parameters to the patient's IPG 10 through the GUI 64.

[0021] exist Figure 5 A portion of GUI 64 is shown as an example in . Those skilled in the art will appreciate that the details of GUI 64 will depend on where in the clinician programmer software 66 is executed, which will depend on the GUI selections that the clinician has made. Figure 5A GUI 64 is shown at one point to allow stimulation parameters to be set for a patient and stored as a stimulation program. On the left is a program interface 72 that allows stimulation programs to be named, loaded, and saved for a patient as further described in the '038 disclosure. On the right is a stimulation parameter interface 82 in which specified stimulation parameters (A, D, F, E, P) can be defined for a stimulation program. Values ​​for stimulation parameters related to waveform shape (A; in this example, current), pulse width (PW), and frequency (F) are shown in the waveform parameter interface 84, which includes buttons that the clinician can use to increase or decrease these values.

[0022] The stimulation parameters associated with the electrode 16 (the activated electrode E and its polarity P) can be adjusted in the electrode parameter interface 86. The electrode stimulation parameters are also visible and can be manipulated in the following lead interface 92, which shows the leads 15 (or 15') approximately in their appropriate positions relative to each other, for example, on the left or right side of the spine. The cursor 94 (or other selection device such as a mouse pointer) can be used to select a specific electrode in the lead interface 92. Buttons in the electrode parameter interface 86 allow selected electrodes (including the shell electrode Ec) to be designated as anodes, cathodes, or off. The electrode parameter interface 86 also allows the relative strength of the anodic current or cathodic current of the selected electrode to be specified as a percentage X. This is particularly useful if, as described in the '038 disclosure, more than one electrode acts as an anode or cathode at a given time. According to the Figure 2 In the example waveforms shown in FIG. 9 , as shown in lead interface 92, electrode E5 has been selected as the only anode to source current, and this electrode receives X=100% of the specified anode current + A. Similarly, electrode E4 has been selected as the only cathode to sink current, and this electrode receives X=100% of the cathode current -A.

[0023] As shown, the GUI 64 specifies the pulse width PW of only the first pulse phase 30a. Nevertheless, the clinician programmer software 66 that runs the GUI 64 and receives input from it will ensure that the IPG 10 and ETS 40 are programmed to make the stimulation program present as a biphasic pulse when a biphasic pulse will be used. For example, the clinician programming software 66 can automatically determine the duration and amplitude of both pulse phases 30a and 30b (e.g., each has a duration of PW, and has opposite polarities +A and -A). The advanced menu 88 can also be used to (among other things) define the relative duration and amplitude of pulse phases 30a and 30b, and allow other more advanced modifications, such as setting the duty cycle (on / off time) of the stimulation pulse, and the rise time that the stimulation reaches its programmed amplitude (A) and the like. The mode menu 90 allows the clinician to select different modes for determining stimulation parameters. For example, as described in the '038 publication, mode menu 90 may be used to enable electronic trolling, which includes an automatic programming mode that performs current steering along an electrode array by moving the cathode in a bipolar manner.

[0024] Although GUI 64 is shown as operating within clinician programmer 50 , the user interface of external controller 45 may also provide similar functionality. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 An implantable pulse generator (IPG) that can be used for spinal cord stimulation (SCS) according to the prior art is shown.

[0026] Figure 2 An example of stimulation pulses that may be generated by an IPG according to the prior art is shown.

[0027] Figure 3 The use of an external trial stimulator (ETS) according to the prior art which can be used to provide stimulation prior to IPG implantation is shown.

[0028] Figure 4 Various external devices capable of communicating with and programming stimulation in the IPG and ETS according to the prior art are shown.

[0029] Figure 5 A graphical user interface (GUI) of a clinician programmer external device for setting or adjusting stimulation parameters according to the prior art is shown.

[0030] Figure 6 A sweet spot search for determining effective electrodes for a patient using a movable sub-sensing bipole is shown.

[0031] FIG. 7A to FIG. 7D A sweet spot search for determining effective electrodes for a patient using a movable super-sensing bipole is shown.

[0032] Figure 8 A stimulation circuit that may be used in IPG or ETS is shown that is capable of providing multiple independent current controls for independently setting the current at each of the electrodes.

[0033] Fig. 9 A flow chart showing a study conducted on various patients with back pain, designed to determine optimized subperceptual SCS stimulation parameters in the frequency range of 1 kHz to 10 kHz.

[0034] FIG. 10A to FIG. 10C Various results are shown as a function of stimulation frequency in the frequency range of 1 kHz to 10 kHz, including: average optimized pulse width ( Fig. 10A ), average charge per second, and optimized stimulation amplitude ( Fig. 10B ), and back pain scores ( Fig. 10C ).

[0035] FIG. 11A to FIG. 11C Further analysis of the relationship between average optimized pulse width and frequency over the frequency range of 1 KHz to 10 kHz is shown, and statistically-significant regions of optimization for these parameters are identified.

[0036] Fig. 12A Results are shown for patients tested with sub-perceptual therapy at frequencies at or below 1 kHz, and the optimized pulse width range determined at the tested frequencies, as well as the region of optimized pulse width v. frequency for sub-perceptual therapy are shown.

[0037] Fig. 12B Various modeled relationships between average optimized pulse width and frequencies at or below 1 kHz are shown.

[0038] Fig. 12C The duty cycle for the optimized pulse width as a function of frequency at or below 1 kHz is shown.

[0039] Fig.12D The average battery current and battery discharge time at optimized pulse widths as a function of frequency at or below 1 kHz are shown.

[0040] Fig.13A and Fig. 13B Results of additional testing are shown which validate the frequency versus pulse width relationship proposed previously.

[0041] Fig.14 A fitting module is shown that illustrates how the relationships and regions determined for optimized pulse width and frequency (≤ 10 kHz) can be used to set sub-sensory stimulation parameters for IPG or ETS.

[0042] Fig.15 The algorithm used for super-perceptual sweet spot search prior to sub-perceptual treatment is shown, as well as possible optimization of sub-perceptual treatment using a fitting module.

[0043] Fig.16 An alternative algorithm for optimizing sub-perceptual treatment using a fitting module is shown.

[0044] Fig.17A and Fig. 17B An analysis of optimized sub-perceptual stimulation parameters is shown, which includes a model of energy (average charges per second) versus frequency.

[0045] Figure 17C-17E Various algorithms are shown by which Fig.17A and Fig. 17B The model can be used to select optimized subperceptual stimulation parameters.

[0046] Fig.17F It is shown that the energy model can be viewed from the perspective of other stimulation parameters besides frequency. For example, energy can be modeled with pulse width.

[0047] Fig.18 A patient-derived model is shown showing a surface representing optimized sub-perceptual values ​​for frequencies and pulse widths, and also includes the patient's perception threshold pth as measured at these frequencies and pulse widths.

[0048] Fig.19A and Fig.19B Plotted perception threshold pth versus pulse width for a number of patients is shown, and it is shown how the results can be curve fit.

[0049] Fig. 20 A graph of parameter Z versus pulse width for a patient is shown, where Z comprises the patient's optimized amplitude A expressed as a percentage of the perception threshold pth (ie, Z=A / pth).

[0050] Figure 21A-Figure 21F An algorithm is shown for using Figure 18-Figure 20 The modeling information is obtained and perceptual threshold measurements made on the patient are used to derive the patient's optimized range of subperceptual stimulation parameters (e.g., F, PW, and A).

[0051] Fig. 22The use of optimized stimulation parameters in a patient external controller is shown, including a user interface that allows the patient to adjust stimulation within a range.

[0052] Figure 23A-23F The effect of statistical variance in modeling is shown, resulting in a volume that the optimal stimulation parameters determined for a patient may occupy. A user interface for a patient external controller is also shown to allow the patient to adjust stimulation within this volume.

[0053] Fig.24 A stimulation mode user interface is shown from which the patient can select different stimulation modes to provide stimulation or allow the patient to control stimulation using different subsets of stimulation parameters determined using optimized stimulation parameters.

[0054] Figures 25A-30B Examples of different subsets of stimulation parameters based on the patient's selection of different stimulation modes are shown. Fig.25A ) shows the frequency and pulse width of a subset, while the graph labeled B (e.g. Fig.25B ) show the amplitude and perception threshold of the subset. These figures show that the stimulation parameter subsets corresponding to different stimulation modes can include parameters that are completely constrained by (i.e., completely within) the determined optimized stimulation parameters, or can include parameters that are only partially constrained by the optimized stimulation parameters.

[0055] Fig.31 An automatic mode is shown, where the IPG and / or external controller is used to determine when a particular stimulation mode should be automatically entered based on sensed information.

[0056] Fig.32 Another example of a simulation mode user interface is shown, in which stimulation modes are presented for selection over a two-dimensional representation of stimulation parameters, although a three-dimensional representation indicating subset volumes may also be used.

[0057] Fig.33 A GUI aspect is shown that allows the patient to adjust the stimulation, wherein suggested stimulation areas for the patient are shown in conjunction with the adjustment aspect.

[0058] Fig.34A and Fig.34C Different ways in which an adjustment of one or more stimulation parameters can be automatically performed within a determined range or volume of optimized stimulation parameters are shown.

[0059] Fig.34B shows that it can be used to automatically generate Fig.34A The GUI for the adjustment.

[0060] Fig.35It is shown that in addition to one or more stimulation parameters within the determined optimized stimulation parameters, the position or focus of the pole configuration can also be changed.

[0061] Fig.36 Specific examples of adjustments within a determined range or volume of optimized stimulation parameters are shown, wherein pulse width and frequency are adjusted between different time periods.

[0062] Fig.37A and Fig.37B A specific example of regulation within a determined range or volume of optimized stimulation parameters is shown, wherein the stimulation is provided by a stimulation bolus.

[0063] Fig.38 The use of a fitting algorithm is shown that uses the patient fitting information to select the best stimulation parameters from the range or volume of optimized stimulation parameters determined for that patient.

[0064] Figure 39A-Figure 39C Receiving patient fit information at a GUI of an external device is shown, including pain information, mapping information, field information, and patient phenotype information.

[0065] Fig.40 It is shown that fitting information can be determined and used by a fitting algorithm as a function of the patient's posture.

[0066] Figure 41A-41C It is shown in flow chart form how the fitting algorithm may process the fitting information and the optimized stimulation parameters based on the training data to determine the best optimized stimulation parameters for the patient.

[0067] Fig.42 An alternative fitting algorithm is shown, in which the best optimized stimulation parameters are determined using patient fitting information and a non-patient specific model. DETAILED DESCRIPTION

[0068] Although spinal cord stimulation (SCS) therapy can be an effective means of relieving pain in patients, such stimulation can also cause paresthesias. Paresthesias (sometimes referred to as "supra-perception" therapy) are sensations that can accompany SCS therapy, such as numbness, tingling, heat, cold, etc. Typically, the effects of paresthesias are mild, or at least not overly concerning to the patient. In addition, for patients whose chronic pain is now under control by SCS therapy, paresthesias are usually moderately compromised. Some patients even find paresthesias to be comfortable and soothing.

[0069] Nevertheless, at least for some patients, SCS therapy would ideally provide complete pain relief without paresthesias - this is often referred to as "sub-perception" or subthreshold therapy that the patient cannot feel. Effective sub-perception therapy can provide pain relief without paresthesias by emitting stimulation pulses at a higher frequency. Unfortunately, such higher frequency stimulation may require more power, which tends to deplete the battery 14 of the IPG 10. See, for example, U.S. Patent Application Publication 2016 / 0367822. If the battery 14 of the IPG is a primary battery and is not rechargeable, high frequency stimulation means that the IPG 10 will need to be replaced more quickly. Alternatively, if the IPG battery 14 is rechargeable, the IPG 10 will need to be charged more frequently over a longer period of time. Either way, it will cause inconvenience to the patient.

[0070] In SCS applications, it is desirable to determine a stimulation program that will be effective for each patient. An important part of determining an effective stimulation program is determining the "sweet spot" for stimulation in each patient, i.e., selecting which electrodes should be active (E) and with what polarity (P) and relative amplitude (X%) to recruit and therefore treat the neural site where pain originates in the patient. Selecting electrodes proximal to this neural site of pain can be difficult to determine, and trials are typically performed to select the best combination of electrodes to provide treatment for the patient.

[0071] As described in U.S. Patent Application Serial No. 16 / 419,879, filed on May 22, 2019 (expressly incorporated herein by reference), when using subperceptual therapy, selecting electrodes for a given patient may be even more difficult because the patient does not feel the stimulation, and therefore it may be difficult for the patient to feel whether the stimulation is "covering" his pain and therefore whether the selected electrodes are effective. In addition, subperceptual stimulation therapy may require a "washin" period before it can become effective. The wash-in period may take a day or more, and therefore the subperceptual stimulation may not be effective immediately, which makes electrode selection even more difficult.

[0072] Figure 6 The techniques of the '879 application for sweet spot searching are briefly described, ie, how electrodes may be selected to be proximate to a painful neural site 298 in a patient when sub-sensory stimulation is used. Figure 6 The techniques in are particularly useful in a trial setting after a patient has first been implanted with an electrode array (ie, after receiving their IPG or ETS).

[0073] In the example shown, it is assumed that the pain site 298 is likely to be within a tissue region 299. Such a region 299 can then be derived by the clinician based on the patient's symptoms (e.g., by understanding which electrodes are adjacent to specific vertebrae (not shown), such as within the T9 to T10 gap). In the example shown, the region 299 is bounded by electrodes E2, E7, E15, and E10, meaning that electrodes outside of this region (e.g., E1, E8, E9, E16) are unlikely to have an effect on the patient's symptoms. Therefore, these electrodes may not be present in the patient's vertebrae. Figure 6 The dessert described in is selected during the search, as further described below.

[0074] exist Figure 6 In the embodiment of the present invention, a sub-sensing bipolar point 297a is selected, where one electrode (e.g., E2) is selected as an anode that will pull positive current (+A) toward the patient's tissue, and the other electrode (e.g., E3) is selected as a cathode that will sink negative current (-A) from the tissue. This is similar to the previous discussion of Figure 2 The content of the description, and biphasic stimulation pulses can be used with effective charge recovery. Because the bipole 297a provides subperceptual stimulation, the amplitude A used during the sweet spot search is titrated down until the patient no longer feels paresthesia. The subperceptual bipole 297a is provided to the patient over a duration (such as a few days), which allows the potential effectiveness of the subperceptual bipole to be "washed in" and allows the patient to provide feedback on how well the bipole 297a helps the patient's symptoms. Such patient feedback can include a pain scale ranking. For example, a patient can rank their pain using a numeric rating scale (NRS) or a visual analog scale (VAS) on a scale from 1 to 10, where 1 represents no pain or almost no pain, and 10 represents the worst pain imaginable. As discussed in the '879 application, such a pain scale ranking can be input into the patient's external controller 45.

[0075] After testing the bipole 297a in this first position, a different combination of electrodes (anodic electrode E3, cathodic electrode E4) is selected that will move the position of the bipole 297 within the patient's tissue. Again, the amplitude of the current A may need to be titrated to an appropriate sub-perceptual level. In the example shown, the bipole 297a is moved down one electrode lead and up the other electrode lead, as shown in the path 296 of finding a combination of electrodes that covers the pain site 298. Figure 6In the example of FIG, given that the pain site 298 is adjacent to electrodes E13 and E14, it can be expected that the bipole 297a at those electrodes will provide the best relief for the patient, as reflected by the patient's pain score ranking. The specific stimulation parameters selected when forming the bipole 297a can be selected at the GUI 64 of the clinician programmer 50 or other external device (such as the patient external controller 45) and wirelessly transmitted to the patient's IPG or ETS with a telemetry transmitter for execution.

[0076] Although Figure 6 The sweet spot search in can be effective, but it can also take a considerable time when using subperceptual stimulation. As already noted, subperceptual stimulation is provided for several days at each bipole 297 location, and because a large number of bipole locations are selected, the entire sweet spot search can take up to a month to complete.

[0077] The inventors have determined through testing of SCS patients that even if it is desired to eventually continue to use subperceptual treatment for the patient after the sweet spot search, it is beneficial to use superperceptual stimulation to select active electrodes for the patient during the sweet spot search. Compared to the use of subperceptual stimulation (which requires a wash-in period at each set of electrodes tested), the use of superperceptual stimulation during the sweet spot search greatly accelerates the determination of effective electrodes for the patient. After the electrodes for the patient are determined using superperceptual treatment, the treatment can be titrated to maintain the subperceptual level of the same electrodes determined for the patient during the sweet spot search. Since it is known that the selected electrodes are recruiting the patient's painful neural sites, it is more likely that the application of subperceptual treatment to those electrodes will have an immediate effect, thereby reducing or potentially eliminating the need for washing in subsequent subperceptual treatment. In summary, when a superperceptual sweet spot search is utilized, effective subperceptual treatment can be achieved more quickly for the patient. Preferably, the superperceptual sweet spot search occurs using symmetrical biphasic pulses occurring at low frequencies such as between 40 Hz and 200 Hz in one example.

[0078] According to one aspect of the disclosed technology, a patient will be provided with sub-aware therapy. A sweet spot search for determining electrodes that can be used during sub-aware therapy can precede such sub-aware therapy. In some aspects, when sub-aware therapy is used for a patient, the sweet spot search can use bipolar point 297a ( Figure 6 ), as just described. This can be relevant because a subperceptual sweet spot search can match the ultimate subperceptual treatment that the patient will receive.

[0079] However, the inventors have determined that even if sub-perceptual treatment is ultimately used for the patient, the use of super-perceptual stimulation (i.e., stimulation with accompanying paresthesias) during the sweet spot search may be beneficial. Fig. 7A, where the movable bipolar point 301a provides a super-sensory stimulation that can be felt by the patient. Figure 6 Providing bipole 301a as a suprasensory stimulus may simply involve increasing its amplitude (e.g., current A) when compared to sub-perceptual bipole 297a, although other stimulation parameters may also be adjustable—such as by providing a longer pulse width.

[0080] The inventors have determined that even if sub-sensory treatment is ultimately used for the patient, there are benefits to employing super-sensory stimulation during the sweet spot search.

[0081] First, as described above, the use of supersensory treatment by definition allows the patient to feel the stimulation, which enables the patient to provide essentially immediate feedback to the clinician whether the paresthesia appears to cover their pain site 298 well. In other words, it is not necessary to spend time washing in the bipole 301a at each location as it moves along the path 296. Therefore, the appropriate bipole 301a adjacent to the patient's pain site 298 can be established much more quickly, such as within a single clinician's visit, rather than over a period of days or weeks. In one example, when the supersensory sweet spot search is preceded by subsensory treatment, the time required to wash in the subsensory treatment can be 1 hour or less, 10 minutes or less, or even a few seconds or so. This allows the wash-in to occur during a single programming session of programming the patient's IPG or ETS, and does not require the patient to leave the clinician's office. In addition, it is noted that the subsensory treatment continues to be effective for a period of time after such treatment has stopped, that is, the treatment is effective during a "wash-out" period after the treatment has stopped. This wash-out period can be ten minutes or more, or even an hour or more. Note that this is beneficial because it means that therapy can be scaled back for a period of time (during a washout period), or more precisely, sub-aware therapy can be cycled on and off. In short, the IPG does not need to continuously provide sub-aware therapy and can effectively be turned off for a period of time, which saves power to the IPG.

[0082] Second, the use of supersensory stimulation during the sweet spot search ensures that electrodes are identified that well recruit the pain site 298. As a result, after the sweet spot search is complete and the final subsensory therapy is titrated to the patient, the wash-in of that subsensory therapy may not be performed for a long time because the electrodes required for good recruitment have been confidently identified.

[0083] FIG. 7B to FIG. 7DOther super-sensing bipoles 301b to 301d that can be used are shown, and in particular how a virtual bipole can be formed using a virtual pole by activating three or more of the electrodes 16. Virtual poles are further discussed in U.S. Patent Application Publication 2019 / 0175915, which is incorporated herein by reference in its entirety, and therefore only briefly described here. To assist in forming a virtual pole, if the stimulation circuit 28 or 44 used in IPG or ETS is capable of independently setting current at any of the electrodes - this is sometimes referred to as multiple independent current control (MICC), as described below with respect to Figure 8 This is further explained.

[0084] When using virtual bipolar, the clinician programmer 50 ( Figure 4 ) in the GUI 64( Figure 5 ) can be used to position 291 (which may not necessarily correspond to the location of the physical electrode 16 Figure 7B ) define the anode pole (+) and cathode pole (-) at these locations 291. The control circuit 70 in the clinician programmer 50 can calculate from these locations 291 and from other tissue modeling information which physical electrodes 16 will need to be selected and at what amplitude to form the virtual anode and virtual cathode at the specified locations 291. As previously described, the amplitude at the selected electrode can be expressed as a percentage X% of the total current amplitude A specified at the GUI 64 of the clinician programmer 50.

[0085] For example, in Figure 7B , the virtual anode pole is positioned at a position 291 between electrodes E2, E3, and E10. The clinician programmer 50 can then calculate, based on this position, the appropriate share (X%) of the total anode current +A that each of these electrodes will receive (during the first pulse phase 30a) to position the virtual anode at this position. Because the location of the virtual anode is closest to electrode E2, this electrode E2 can receive the largest share of the specified anode current +A (e.g., 75%*+A). Electrodes E3 and E10, which are adjacent to the location of the virtual anode pole but farther away, receive a smaller share of the anode current (e.g., 15%*+A and 10%*+A, respectively). Similarly, it can be seen that according to the specified position 291 of the virtual cathode pole adjacent to electrodes E4, E11, and E12, these electrodes will receive an appropriate share of the specified cathode current -A (e.g., again during the first pulse phase 30a, 20%*-A, 20%*-A, and 60%*-A, respectively). These polarities will then be reversed during the second phase 30b of the pulse, as Figure 7BIn any case, the use of virtual poles in the formation of the bipole 301b allows the field in the tissue to be shaped, and many different combinations of electrodes can be tried during the sweet spot search. In this regard, it is not strictly necessary to move the (virtual) bipole along the path 296 for each electrode in sequence, and the path can be random, perhaps guided by feedback from the patient.

[0086] Figure 7C A useful virtual dipole 301c configuration that can be used during a sweet spot search is shown. This virtual dipole 301c again defines the target anode and cathode, whose locations do not correspond to the locations of the physical electrodes. A virtual dipole 301c is formed along the lead (essentially spanning the length of the four electrodes from E1 to E5). This creates a larger field in the tissue that better recruits the patient's pain site 298. As it moves along the path 296, it is aligned with the Fig. 7A Compared to the smaller dipole configuration 301a in FIG. 3 , the dipole configuration 301c may need to be moved to fewer locations, thereby speeding up the detection of the pain site 298 . Fig.7D exist Figure 7C The bipole configuration of 301d is expanded to create a virtual bipole 301d using electrodes formed on two leads, for example, from electrode E1 to E5 and from electrode E9 to E13. This bipole 301d configuration needs to move only along a single path 296 parallel to the leads because its field is large enough to recruit neural tissue adjacent to both leads. This can further accelerate pain site detection.

[0087] In some aspects, the super-sensitive bi-poles 301a-301d used during the sweet spot search include symmetrical bi-phasic waveforms having pulse phases 30a and 30b of the same pulse width PW and the same amplitude (with flipped polarity during each phase) actively driven (e.g., by the stimulation circuit 28 or 44) (e.g., A 30a =A 30b , and PW 30a =PW 30b This is beneficial because the second pulse phase 30b provides for efficient charge recovery, where in this case the charge (Q 30a ) is equal to the charge (Q 30b) so that the pulses are charge balanced. The use of a biphasic waveform is also believed to be beneficial because, as is known, the cathode is heavily involved in the recruitment of neural tissue. When a biphasic pulse is used, the positions of the (virtual) anode and cathode will flip during the two phases of the pulse. This effectively doubles the neural tissue recruited for stimulation and therefore increases the likelihood that the pain site 298 will be covered by the bipolar point at the correct location.

[0088] The supersensitive bipolar points 301a-301b need not, however, include symmetrical biphasic pulses as just described. For example, the amplitudes and pulse widths of the two phases 30a and 30b may be different while keeping the charge (Q) of the two phases balanced (e.g., Q 30a =A 30a *PW 30a =A 30b *PW 30b =Q 30b Alternatively, the two phases 30a and 30b may be charge unbalanced (eg, Q 30a =A 30a *PW 30a >A 30b *PW 30b =Q 30b , or Q 30a =A 30a *PW 30a 30b *PW 30b =Q 30b ). In summary, the pulses in the bipolar points 301 to 301d can be bi-phase symmetric (and therefore inherently charge balanced), bi-phase asymmetric but still charge balanced, or bi-phase asymmetric and charge unbalanced.

[0089] ​In a preferred example, the frequency F of the supersensory pulses 301a to 301d used during the supersensory sweet spot search may be 10 kHz or less, 1 kHz or less, 500 Hz or less, 300 Hz or less, 200 Hz or less, 130 Hz or less, or 100 Hz or less, or a range bounded by two of these frequencies (e.g., 100 to 130 Hz, or 100 to 200 Hz). In a specific example, a frequency of 90 Hz, 40 Hz, or 10 Hz may be used, wherein the pulses comprise biphasic pulses that are preferably symmetrical. However, a single actively driven pulse phase followed by a passive recovery phase may also be used. The pulse width PW may also comprise values ​​in the range of hundreds of microseconds, such as 150 microseconds to 400 microseconds. Because the purpose of the supersensory sweet spot search is simply to determine electrodes that adequately cover the patient's pain, the frequency and pulse width may not be that important at this stage. Once the electrodes are selected for subsensory stimulation, the frequency and pulse width may be optimized, as discussed further below.

[0090] It should be understood that the super-sensitive bipoles 301a to 301d used during the sweet spot search are not necessarily the same electrodes selected when subsequently providing sub-sensitive treatment to the patient. Rather, the best position of the bipoles of interest during the search can be used as a basis for modifying the selected electrodes. Assuming, for example, that bipoles 301a ( Fig. 7A ), and it is determined that the bipolar point provides the best pain relief when located at electrodes E13 and E14. At that time, sub-perceptual treatment can be attempted for the patient using those electrodes E13 and E14. Alternatively, it may be wise to modify the selected electrodes to see if the patient's symptoms can be further improved before attempting sub-perceptual treatment. For example, using virtual poles as already described, the distance between the cathode and the anode (focus) can be varied. Alternatively, a triple pole (anode / cathode / anode) consisting of electrodes E12 / E13 / E14 or E13 / E14 / E15 can be attempted. See U.S. Patent Application Publication 2019 / 0175915 (discussing triple poles). Or electrodes on different leads can be attempted in conjunction with E13 and E14. For example, because electrodes E5 and E6 are typically adjacent to electrodes E13 and E14, it may be useful to add E5 or E6 as a source of anode current or cathode current (again creating a virtual pole). All of these types of adjustments should also be understood to include adjustments to the "steering" or "location" where the treatment is applied, even if the center point of stimulation does not change (as can occur, for example, when varying the distance or focus between the cathode and anode).

[0091] In one example about Figure 8 Explain Multiple Independent Current Control (MICC), in Figure 8 FIG. 2 shows a stimulation circuit 28 ( Figure 1 ) or 44( Figure 3 ). The stimulation circuit 28 or 44 can independently control the current or charge at each electrode and use the GUI 64 ( Figure 5 ) allows current or charge to be directed to different electrodes, which is useful, for example, when moving bipole 301i along path 296 during a sweet spot search ( FIG. 7A to FIG. 7D ). The stimulation circuit 28 or 44 includes one or more current sources 440 i and one or more current sinks 442 i Source 440 i Hesu442 i A digital-to-analog converter (DAC) may be included and may be referred to as a PDAC 440 based on the positive (sourced, anode) and negative (sunk, cathode) currents they source, respectively. i and NDAC 442 i In the example shown, the NDAC 440 i / PDAC 442 i The pairing is dedicated to (hardwired to) a specific electrode node ei 39. Each electrode node ei 39 is preferably connected to electrode Ei 16 via a DC blocking capacitor Ci38, which acts as a safety measure to prevent DC current injection into the patient if, for example, there is a circuit fault in the stimulation circuit 28 or 44. PDAC 440 i and DNAC 442 i A voltage source may also be included.

[0092] The PDAC 440 is controlled via the GUI 64 i and NDAC 442 i Appropriate control of allows either electrode 16 and housing electrode Ec 12 to act as an anode or cathode to generate current through the patient's tissue. Such control preferably takes the form of digital signals Iip and Iin that set the anode current and cathode current at each electrode Ei. If, for example, it is desired to set electrode E1 as an anode with a current of +3 mA, and electrodes E2 and E3 as cathodes each with a current of -1.5 mA, then control signal I1p would be set to have a 3 mA digital equivalent to cause PDAC 440 to 1 +3mA is generated, and the control signals I2n and I3n will be set to have a 1.5mA digital equivalent to enable the NDAC 442 2 and NDAC 442 3Each produces -1.5 mA. Note that the definition of these control signals may also occur using programmed amplitudes A and percentages X% set in the GUI 64. For example, A may be set to 3 mA, where E1 is designated as an anode with X=100%, and where E2 and E3 are designated as cathodes with X=50%. Alternatively, the control signals may not be set in percentages, and instead the GUI 64 may simply dictate that current will occur at each electrode at any point in time.

[0093] In summary, GUI 64 can be used to independently set the current at each electrode, or to direct the current between different electrodes. This is particularly useful in forming a virtual bipole, which was previously explained as including activating more than two electrodes. MICC also allows for the formation of a finer electric field in the patient's tissue.

[0094] Other stimulus circuits 28 may also be used to implement the MICC. In an example not shown, the switch matrix may be between one or more PDACs 440 i and electrode node ei 39, and one or more NDACs 442 i Between the electrode nodes. The switch matrix allows one or more of the PDACs and one or more of the NDACs to be connected to one or more electrode nodes at a given time. Various examples of stimulation circuits can be found in the following documents: USP6,181,969, 8,606,362, 8,620,436; and US Patent Application Publications 2018 / 0071513, 2018 / 0071520, and 2019 / 0083796.

[0095] Many of the stimulation circuits 28 or 44 (including the PDAC 440i and NDAC 442i, the switch matrix (if present), and the electrode nodes ei 39) may be integrated on one or more application specific integrated circuits (ASICs), as described in the following: U.S. Patent Application Publications 2012 / 0095529, 2012 / 0092031, and 2012 / 0095519. As described in these documents, one or more ASICs may also include: other circuits that may be used in the IPG 10, such as telemetry circuits (for interfacing the disconnect chip with the telemetry antenna of the IPG or ETS), circuits for generating a constant current output voltage (compliance voltage) VH to power the stimulation circuits, various measurement circuits, etc.

[0096] While it is preferred to use a sweet spot search (and in particular a super-perceptual sweet spot search) to determine the electrodes to be used during a subsequent sub-perceptual treatment, it should be noted that this is not strictly necessary. Sub-perceptual treatment may be led by a sub-perceptual sweet spot search, or may not be led by a sweet spot search at all. In short, sub-perceptual treatment as described next does not rely on the use of any sweet spot search.

[0097] In another aspect of the invention, the inventors have determined through testing of SCS patients that there is a statistically significant correlation between pulse width (PW) and frequency (F) at which an SCS patient will experience a reduction in back pain without paresthesia (subperceptual). Use of this information can help determine what pulse width might be optimized for a given SCS patient based on a particular frequency, and what frequency might be optimized for a given SCS patient based on a particular pulse width. Beneficially, this information suggests that subperceptual SCS stimulation without paresthesia can occur at frequencies of 10 kHz and below. Use of such low frequencies allows subperceptual therapy to be used with much lower power consumption in the patient's IPG or ETS.

[0098] Figures 9 to 11C Results derived from testing patients at frequencies ranging from 1 kHz to 10 kHz are shown. Fig. 9 It describes how data was collected from actual SCS patients, and the criteria for patient inclusion in the study. Patients with back pain who had not yet received SCS treatment were first identified. Key patient inclusion criteria included were: persistent low back pain for more than 90 days; NRS pain scale of 5 or higher (NRS described below); stable opioid treatment for 30 days; and baseline Oswestry Disability Index score greater than or equal to 20 and less than or equal to 80. Key patient exclusion criteria included were: back surgery within the previous 6 months; presence of other confounding medical / psychological conditions; and untreated major psychiatric illness or serious medication-related performance problems.

[0099] After such an initial screening, the patient periodically enters a qualitative indicator of their pain (i.e., a pain score) into a portable electronic diary device, which may include a patient external controller 45, and the external controller 45 may in turn transmit its data to a clinician programmer 50 ( Figure 4 Such pain scores may include a numeric rating scale (NRS) score from 1 to 10 and may be entered into an electronic diary three times daily. Fig. 10C As shown in , patients who were not ultimately excluded from the study and who had not yet received subperceptual stimulation treatment had a baseline NRS score of approximately 6.75 / 10, with a standard error SE (sigma / SQRT(n)) of 0.25.

[0100] Back to Fig. 9 , the patient then places the test lead 15'( Figure 3 ) are implanted on the left and right sides of the spine, and external trial stimulation is provided to the patient as previously described. A clinician programmer 50 is used to provide the stimulation program to each patient's ETS 40, as previously described. This is done to confirm that SCS therapy is helpful for a given patient in relieving their pain. If SCS therapy is not helpful for a given patient, the trial leads 15' are removed, and the patient is then excluded from the study.

[0101] Those patients for whom external trial stimulation was helpful eventually receive full implantation of a permanent IPG 10, as previously described. After the healing period, and again using the clinician programmer 50, the "sweet spot" for stimulation is located in each patient, i.e., which electrode should be active (E) and with what polarity (P) and relative amplitude (X%) to recruit and thus treat the neural sites in the patient 298. The sweet spot search can be performed with the previously described information about Figures 6 to 7D Any of the methods described may occur, but in preferred embodiments due to the previously described benefits will include extrasensory stimulation (e.g., such as, FIG. 7A to FIG. 7D ). However, this is not strictly necessary, and sub-perceptual stimulation can also be used during the sweet spot search. Fig. 9 In the example of , the sweet spot search occurs at 10kHz, but again the frequency used during the sweet spot search can be varied. Symmetrical biphasic pulses are used during the sweet spot search, but again this is not strictly necessary. The decision on which electrodes should be activated begins by selecting the electrode 16 that exists between thoracic vertebrae T9 and T10. However, electrodes as far away as T8 and T11 can also be activated if necessary. Fluoroscopy images of the leads 15 in each patient are used to determine which electrodes are adjacent to vertebrae T8, T9, T10, and T1.

[0102] During the sweet spot search, bipolar stimulation using only two electrodes was used for each patient, and only adjacent electrodes were used on a single lead 15, similar to Figure 6 and Fig. 7A Thus, a patient's sweet spot may include stimulation of adjacent electrodes E4 (as cathode) and E5 (as anode) on left lead 15, as previously described in Figure 2(the electrodes may be between T9 and T10), while another patient's sweet spot may include stimulation of adjacent electrodes E9 (as anode) and E10 (as cathode) on right lead 15 (the electrodes may be between T10 and T11). It is desirable to use only bipolar stimulation of adjacent electrodes and only between vertebrae T8 to T11 to minimize variability in treatment and symptoms between different patients in a study. However, more complex bipolar stimulation such as that on FIG. 7B to FIG. 7D Those described can also be used during the sweet spot search. Patients continued in the study if they had sweet spot electrodes at the desired chest location and if they experienced 30% or greater pain relief per NRS score; patients who did not meet these criteria were excluded from further study. While the study initially started with 39 patients, by Fig. 9 So far 19 patients have been excluded from the study, leaving a total of 20 remaining patients.

[0103] The remaining 20 patients then underwent a "washout" period, meaning that their IPG did not provide stimulation for a period of time. In particular, the patient's NRS pain score was monitored until their pain reached 80% of their initial baseline pain. This was to ensure that the benefit of the previous stimulation did not persist into the next analysis period.

[0104] The remaining patients then underwent sub-sensory SCS treatment at different frequencies ranging from 1 kHz to 10 kHz using the previously determined sweet spot active electrodes. However, this is not strictly necessary, as the current at each electrode is also independently controlled to help shape the electric field in the tissue as previously described. Fig. 9 As shown in , each patient was tested using stimulation pulses having frequencies of 10 kHz, 7 kHz, 4 kHz, and 1 kHz. For simplicity, Fig. 9 The frequencies are shown to be tested in this order for each patient, but in reality the frequencies are applied to each patient in a random order. Once testing at a given frequency is completed, there is a washout period before testing at another frequency begins.

[0105] At each measured frequency, the amplitude (A) and pulse width (PW) of the stimulation are adjusted and optimized for each patient (first pulse phase 30a; Figure 2) so that each patient experiences the best possible pain relief without paresthesia (subperception). In particular, using the clinician programmer 50, and leaving the same sweet spot electrode previously determined active (although again this is not strictly necessary), each patient is stimulated at a low amplitude (e.g., 0), which is increased to the maximum point at which paresthesia can be noticed by the patient (perception threshold). An initial stimulation is then selected for the patient at 50% of this maximum amplitude, i.e., so that the stimulation is subperception and therefore without paresthesia. However, other percentages of the maximum amplitude (80%, 90%, etc.) may also be selected, and may vary with patient activity or position, as further described below. In one example, the stimulation circuit 28 or 44 in the IPG or ETS may be configured to receive instructions from the GUI 64 via a selectable option (not shown) to reduce the amplitude of the stimulation pulse to a certain amount or percentage, or to reduce it by a certain amount or percentage for presentation, so that the pulse can be made subperception (if it has not already been made subperception). Other stimulation parameters (e.g., pulse width, charge) may also be reduced to the same effect.

[0106] The patient will then leave the clinician's office and thereafter and in communication with the clinician (or their technician or programmer) will use their external controller 45 ( Figure 4 ) make adjustments to their stimulation (amplitude and pulse width). At the same time, the patient will enter the NRS pain score in their electronic diary (e.g., an external controller), again three times a day. Patient adjustment of amplitude and pulse width is usually an iterative process, but essentially it is based on feedback from the patient to try to adjust the treatment to reduce their pain while still ensuring that the stimulation is sub-perceptual. Testing at each frequency lasted for about 3 weeks, and stimulation adjustments may be made every two days or so. At the end of the testing period at a given frequency, the optimized amplitude and pulse width have been determined for each patient and recorded for each patient, together with the patient's NRS pain score for those optimized parameters entered in their electronic diary.

[0107] In one example, the percentage of the maximum amplitude used to provide sub-perceptual stimulation can be selected depending on the patient's activity level or position. In this regard, the IPG or ETS can include a device for determining the patient's activity or position, such as an accelerometer. If the accelerometer indicates a high degree of patient activity or a position where the electrodes are farther from the spinal cord (e.g., lying down), the amplitude can be increased to a higher percentage to increase the current (e.g., 90% of the maximum amplitude). If the patient is experiencing a lower degree of activity or a position where the electrodes are closer to the spinal cord (e.g., standing up), the amplitude can be reduced (e.g., to 50% of the maximum amplitude). Although not shown, the GUI 64 ( Figure 5) may include setting a percentage of the maximum amplitude at which sensory abnormalities become apparent to the patient, thereby allowing the patient to adjust to sub-perceptual current amplitudes.

[0108] Preferably, multiple independent current controls (MICCs) are used to provide or regulate sub-perceptual therapy, as previously described with respect to Figure 8 as discussed. This allows the current at each electrode to be set independently to facilitate the guidance of current or charge between the electrodes, to help form a virtual bipole, and more generally to allow the electric field to be shaped in the patient's tissue. In particular, the MICC can be used to direct sub-perceptual treatment to different locations in the electrode array and therefore the spinal cord. For example, once a set of sub-perceptual stimulation parameters has been selected for a patient, one or more of the stimulation parameters can be changed. Such changes may be warranted or dictated by the treatment location. The patient's physiology may vary at different spinal locations, and tissue may be more or less conductive at different treatment locations. Therefore, if a sub-perceptual treatment location is directed to a new location along the spinal cord (a change in position of which may include changing the anode / cathode distance or focus), then at least one of the stimulation parameters, such as amplitude, may be warranted to be adjusted. As previously noted, performing sub-perceptual adjustments is facilitated and may occur during a programming session because a substantial wash-in period may be unnecessary.

[0109] Adjustments to sub-perceptual therapy may also include variations in other stimulation parameters, such as pulse width, frequency, and even interphase period (IP) duration ( Figure 2 ). The interphase duration can affect the neural dose, or rate of charge infusion, such that higher sub-perceptual amplitudes will be used with shorter interphase durations. In one example, the interphase duration can be varied between 0 and 3 ms. After the washout period, new frequencies can be tested using the same protocol as described.

[0110] The sub-sensory stimulation pulses used were symmetrical biphasic constant current amplitude pulses having a first pulse phase 30a and a second pulse phase 30b (of the same duration) (see Figure 2 ). However, constant voltage amplitude pulses may also be used. Pulses of different shapes (triangular, sine, etc.) may also be used. When providing sub-inductive therapy, pre-pulsing (i.e., providing a small current before providing one or more actively driven pulse phases) may also occur to affect the polarization or depolarization of neural tissue. See, for example, USP 9,008,790, which is incorporated herein by reference.

[0111] FIG. 10A to FIG. 10CResults are shown for patients tested at 10 kHz, 7 kHz, 4 Hz and 1 kHz. Data are shown in each graph as the mean for the 20 remaining patients at each frequency, with error bars reflecting the standard error (SE) between patients.

[0112] by Fig. 10B Initially, the optimized amplitudes A for the 20 remaining patients are shown at the tested frequencies. Interestingly, the optimized amplitude at each frequency is essentially constant, approximately 3 mA. Fig. 10B The amount of energy consumed at each frequency is also shown, more specifically the mean charge per second (MCS) attributable to the pulse in mC / s. The MCS is calculated by using an optimized pulse width ( Fig. 10A , discussed below) and multiplying it by the optimized amplitude (A) and frequency (F), the MCS value can include the nerve dose. The MCS is related to the current or power that the battery in the IPG 10 must consume to form the optimized pulse. It is worth noting that the MCS is significantly reduced at lower frequencies: for example, the MCS at F=1 kHz is approximately 1 / 3 of its value at higher frequencies (e.g., F=7 kHz or 10 kHz). This means that optimized SCS therapy for relief of back pain without paresthesia can be achieved at lower frequencies like F=1 kHz, with the added benefit of lower power draw being more considerate of the IPG 10's (or ETS 40's) battery.

[0113] Fig. 10A The optimized pulse width is shown as a function of frequency over the tested frequency range of 1 kHz to 10 kHz. As shown, the relationship follows a statistically significant trend: PW = -8.22F + 106 when modeled using linear regression 98a, where the pulse width is measured in microseconds and the frequency is measured in kilohertz, with a correlation coefficient R 2 When polynomial regression 98b is used for modeling, PW = 0.486F 2 –13.6F+116, again where the pulse width is measured in microseconds and the frequency is measured in kilohertz, with an even better correlation coefficient of R 2 = 0.998. Other fitting methods may be used to establish additional information related to the frequency and pulse width at which stimulation pulses are formed to provide pain relief without paresthesia in the frequency range of 1 kHz to 10 kHz.

[0114] Note that the relationship between optimized pulse width and frequency is not simply the expected relationship between frequency and duty cycle (DC) (i.e., the duration of a pulse being "on" divided by its period (1 / F)). In this regard, note that a given frequency has a natural effect on pulse width: it would be expected that a higher frequency pulse would have a smaller pulse width. Thus, for example, a 1 kHz waveform with a 100 microsecond pulse width would be expected to have the same clinical results as a 10 kHz waveform with a 10 microsecond frequency, since both waveforms have a 10% duty cycle. Fig.11A The duty cycle of the stimulation waveform generated using the optimized pulse width in the frequency range of 1 kHz to 10 kHz is shown. Here, the duty cycle is calculated by considering only the first pulse phase 30a ( Figure 2 ) is calculated from the total "on" time of the symmetrical second pulse phase; ignoring the duration of the symmetrical second pulse phase. The duty cycle is not constant over the frequency range of 1kHz to 10kHz: for example, the optimized pulse width at 1kHz (104 microseconds) is not just 10 times the optimized pulse width at 10kHz (28.5 microseconds). Therefore, the optimized pulse width exceeds the scaling significance only for frequency.

[0115] Fig. 10C The results show that for each frequency in the range of 1 kHz to 10 kHz, the optimal stimulation parameters (optimized amplitude ( Figure 7B ) and pulse width ( Fig. 7A ))). As previously noted, patients in this study, prior to receiving SCS treatment, initially reported pain scores with an average of 6.75. After SCS implantation and during the study, and with amplitude and pulse width optimized during temporary subperceptual treatment, their average pain scores dropped significantly to an average pain score of approximately 3 for all frequencies tested.

[0116] Fig.11A An in-depth analysis of the resulting relationship between optimized pulse width and frequency in the frequency range of 1kHz to 10kHz is provided. Fig.11A The graphs in show the average optimized pulse widths for the 20 patients in the study at each frequency, together with the standard errors resulting from the variation between them. These were normalized at each frequency by dividing the standard error by the optimized pulse width, with the variation ranging between 5.26% and 8.51% at each frequency. Thus, a variation of 5% (below all calculated values) can be assumed to be a statistically significant variation at all frequencies tested.

[0117] Based on this 5% change, the maximum average pulse width (PW+5%) and the minimum average pulse width (PW+5%) can be calculated for each frequency. For example, the optimized average pulse width PW at 1kHz is 104 microseconds, and 5% above this value (1.05*104μs) is 109 microseconds; 5% below this value (0.95*104μs) is 98.3 microseconds. Similarly, the optimized average pulse width AVG(PW) at 4kHz is 68.0 microseconds, and 5% above this value (1.05*68.0μs) is 71.4 microseconds; 5% below this value (0.95*68.0μs) is 64.6 microseconds. Thus, a statistically significant reduction in pain without paresthesia occurs within or above the linearly defined region 100a where points 102 are (1 kHz, 98.3 μs), (1 kHz, 109 μs), (4 kHz, 71.4 μs), and (4 kHz, 64.6 μs). A linearly defined region 100b around point 102 is also defined for frequencies greater than or equal to 4 kHz and less than or equal to 7 kHz: (4 kHz, 71.4 μs), (4 kHz, 64.6 μs), (7 kHz, 44.2 μs), (7 kHz, 48.8 μs). A linearly defined region 100c around point 102 is also defined for frequencies greater than or equal to 7 kHz and less than or equal to 10 kHz: (7 kHz, 44.2 μs), (7 kHz, 48.8 μs), (10 kHz, 29.9 μs), (10 kHz, 27.1 μs). Such a region 100 therefore includes information relating to the frequency and pulse width at which stimulation pulses are formed to provide pain relief without paresthesia within the frequency range of 1 kHz to 10 kHz.

[0118] Fig. 11B An alternative analysis of the resulting relationship between optimized pulse width and frequency is provided. In this example, regions 100a to 100c are defined based on the standard error (SE) calculated at each frequency. Therefore, the points 102 defining the corners of regions 100a to 100c are only within the range of the SE error bars at each frequency (PW+SE, and PW-SE), although these error bars have different sizes at each frequency. Therefore, a statistically significant reduction in pain without paresthesia occurs within or above the linearly defined region 100a at the points (1kHz, 96.3μs), (1kHz, 112μs), (4kHz, 73.8μs), and (4kHz, 62.2μs). Linearly defined regions 100b and 100c are similar, and because the points 102 defining them are within Fig. 11B This is explained in the diagram at the top of the figure, so it will not be repeated here.

[0119] Fig. 11CAnother analysis of the resulting relationship between the optimized pulse width and frequency is provided. In this example, regions 100a to 100c are defined based on a standard deviation (SD) calculated at each frequency that is greater than the standard error (SE) metric used for that point. Points 102 defining the corners of regions 100a to 100c are located within the SD error bars at each frequency (PW+SD, and PW-SD), although points 102 may also be set within the error bars, similar to the previous description of the Fig.11A In any case, a statistically significant reduction in pain without paresthesia occurred within or above the linearly defined region 100a at the points (1 kHz, 69.6 μs), (1 kHz, 138.4 μs), (4 kHz, 93.9 μs), and (4 kHz, 42.1 μs). The linearly defined regions 100b and 100c are similar and are defined by the points 102 at the Fig. 11C This is explained in the diagram at the top of the figure, so it will not be repeated here.

[0120] More generally, although not shown, regions within the frequency range of 1kHz to 10kHz that achieve sub-perceptual therapeutic effects include linearly defined regions 100a (1kHz, 50.0μs), (1kHz, 200.0μs), (4kHz, 110.0μs), and (4kHz, 30.0μs); and / or linearly defined regions 100b (4kHz, 110.0μs), (4kHz, 30.0μs), (7kHz, 30.0μs), and (7kHz, 60.0μs); and / or linearly defined regions 100c (7kHz, 30.0μs), (7kHz, 60.0μs), (10kHz, 40.0μs), and (10kHz, 20.0μs).

[0121] In summary, one or more statistically significant regions 100 can be defined for the following optimized pulse width and frequency data, taken for the patients in the study, to arrive at a combination of pulse width and frequency that reduces pain without paresthesia side effects in the frequency range of 1 kHz to 10 kHz, and different statistical measures of error can be used to so define one or more regions.

[0122] FIG. 12A to FIG. 12D Results of testing other patients with sub-perceptual stimulation therapy at frequencies at or below 1 kHz are shown. Testing of patients typically occurs during a suprasensory sweet spot search (see ) for selecting appropriate electrodes (E), polarity (P), and relative amplitude (X%) for each patient. FIG. 7A to FIG. 7D) occurs, although the sub-perceptual electrodes used again may vary from those used during the super-perceptual sweet spot search (e.g., using MICC). Although the form of the pulses used during sub-perceptual treatment may vary, symmetrical biphasic bipolar points are still used to test the patient with sub-perceptual stimulation.

[0123] Fig. 12A The relationship between frequency and pulse width at which patients reported effective subperceptual treatment for frequencies of 1 kHz and below 1 kHz is shown. Note that previously ( Fig. 9 ) can be used when evaluating frequencies at or below 1 kHz, with the frequency adjusted as appropriate.

[0124] As can be seen, at each measured frequency, the optimized pulse width again falls within a certain range. For example, at 800 Hz, patients reported good results when the pulse width fell within the range of 105 to 175 microseconds. The upper end of the pulse width range at each frequency is recorded as PW(high), while the lower end of the pulse width range at each frequency is recorded as PW(low). PW(medium) represents the middle (e.g., average) of PW(high) and PW(low) at each frequency. At each of the measured frequencies, the amplitude (A) of the provided current is titrated down to a sub-perceptual level so that the patient cannot feel the paresthesia. Typically, the current is titrated to 80% of the threshold at which paresthesia can be sensed. Because each patient's anatomy is unique, the sub-perceptual amplitude A can vary from patient to patient. The depicted pulse width data includes the pulse width of only the first phase of the stimulation pulse.

[0125] Table 1 below shows in tabular form the frequency range for frequencies at or below 1kHz. Fig. 12A Optimized width and frequency data from , where the pulse width is expressed in microseconds:

[0126]

[0127] Table 1

[0128] As previously described for frequencies in the 1kHz to 10kHz range ( FIG. 10A to FIG. 11C), the data can be broken down to define different regions 300i at which effective sub-perceptual therapy below 1 kHz is achieved. For example, the regions of effective sub-perceptual therapy can be linearly bounded between the various frequencies and high and low pulse widths that define effectiveness. For example, at 10 Hz, PW(low) = 265 microseconds and PW(high) = 435 microseconds. At 50 Hz, PW(low) = 230 microseconds and PW(high) = 370 microseconds. Thus, the region 300a that provides good sub-perceptual therapy is defined by the linearly bounded region of points (10 Hz, 265 μs), (10 Hz, 435 μs), (50 Hz, 370 μs), and (50 Hz, 230 μs).

[0129] Table 2 defines the linear constraints on Fig. 12A Points in each of the regions 300a to 300g are shown in:

[0130] area Defined by point (Hz, μs) 300a (10,265),(10,435),(50,370),(50,230) 300b (50,230),(50,370),(100,325),(100,195) 300c (100,195),(100,325),(200,260),(200,160) 300d (200,160),(200,260),(400,225),(400,140) 300e (400,140),(400,225),(600,200),(600,120) 300f (600,120),(600,200),(800,175),(800,105) 300g (800,105),(800,175),(1000,150),(1000,90)

[0131] Table 2

[0132] The region of subperceptual therapeutic effectiveness at frequencies at or below 1 kHz may be defined in other statistically significant ways, such as those previously described for frequencies in the 1 kHz to 10 kHz range ( Figures 11A-11C ). For example, region 300i can be defined by reference to a pulse width PW(middle) at the middle of each range at each frequency. PW(middle) can include, for example, an average optimized pulse width reported by patients at each frequency, rather than the strict middle of the effective range reported by those patients. PW(high) and PW(low) can then be determined as statistical variances from the average PW(middle) at each frequency, and can be used to set upper and lower boundaries of the effective sub-perceptual region. For example, PW(high) can include the average PW(middle) plus a standard deviation or standard error, or multiples of such statistical measures; PW(low) can also include the average PW(middle) minus the standard deviation or standard error, or multiples of such statistical measures. PW(high) and PW(low) can also be determined from the average PW(middle) in other ways. For example, PW(high) can include the average PW(middle) plus a certain percentage, and PW(low) can include PW(middle) minus a certain percentage. In summary, one or more statistically significant regions 300 can be defined for optimized pulse width and frequency data at frequencies at or below 1 kHz that reduce pain using sub-perceptual stimulation without paresthesia side effects.

[0133] In addition Fig. 12AShown in the are the mean patient pain scores (NRS scores) reported by patients when using optimized pulse widths for different frequencies at or below 1 kHz. Prior to receiving SCS treatment, the patients' initial reported pain scores averaged 7.92. After SCS implantation, and using subperceptual stimulation at optimized pulse widths (with ranges shown at each frequency), the patients' mean pain scores dropped significantly. At 1 kHz, 200 Hz, and 10 Hz, the mean patient reported pain scores were 2.38, 2.17, and 3.20, respectively. Thus clinical significance for pain relief was demonstrated when using optimized widths with subperceptual treatment at or below 1 kHz.

[0134] exist Fig. 12B The analysis is performed for frequencies of 1 kHz or less from the perspective of the median pulse width PW(middle) at each frequency (F). Fig. 12A As shown, the relationships 310a to 310d follow a statistically significant trend, as shown by Fig. 12B This is demonstrated by the various regression models shown in and summarized in Table 3 below:

[0135]

[0136]

[0137] Table 3

[0138] Other fitting methods may be used to establish additional information related to the frequency and pulse width at which stimulation pulses are formed to provide subperceptual pain relief without paresthesia.

[0139] Regression analysis can also be used to define statistically relevant regions, such as 300a to 300g, where sub-perceptual treatment is effective at or below 1 kHz. Fig. 12B , but regression may be performed for PW(low)vF to set the lower boundary of the correlation region 300i, and regression may be performed for PW(high)vF to set the upper boundary of the correlation region 300i.

[0140] Notice Fig. 12A The relationship between the optimized pulse width and frequency described in Fig. 12C The expected frequency vs. duty cycle (DC) relationship is shown in Figure 1. This is similar to the case when testing the 1kHz to 10kHz frequency range ( Fig.11A), the duty cycle of the optimized pulse width is not constant at 1kHz and below. Again, the optimized pulse width is only important for frequency scaling. Nevertheless, most pulse widths observed to be optimized at 1kHz and below are greater than 100 microseconds. Such pulse widths are not even possible at higher frequencies. For example, at 10kHz, the phases of the two pulses must fit within a 100 microsecond period, so PWs longer than 100 are not even possible.

[0141] Fig.12D The further benefit of using sub-perceptual therapy at frequencies of 1 kHz and below, namely reduced power consumption, is shown. Two sets of data are plotted. The first data set includes pulse widths ( ) optimized for the patient using a battery in the patient's IPG or ETS. Fig. 12A ) the average current drawn at each frequency (AVG Ibat), and the current amplitude A required to achieve subperceptual stimulation for that patient (again, this amplitude can vary for each of the patients). At 1 kHz, the average battery current is about 1700 microamps. However, as the frequency is reduced, the average battery current drops to about 200 microamps at 10 Hz. The second data set considers power consumption from a different vantage point, namely the number of days an IPG or ETS with a fully charged rechargeable battery can operate before needing to be recharged (the "discharge time"). Based on the average battery current data, it would be expected that when the average battery current is higher, the discharge time is lower at higher frequencies (e.g., about 3.9 days at 1 kHz, depending on various charging parameters and settings), and when the average battery current is lower, the discharge time is higher at lower frequencies (e.g., about 34 days at 10 Hz, depending on various charging parameters and settings). This is important: when using optimized pulse widths, not only can effective sub-apparent therapy be provided at frequencies of 1kHz and below; power consumption is greatly reduced, which can place less stress on the IPG or ETS and allow it to operate for longer periods of time. As noted above, excessive power consumption is a serious problem when sub-apparent therapy is conventionally used at higher frequencies. NOTE Fig.12D The data in can also be analyzed in terms of mean charges per second (MSC), as described previously for the 1 kHz to 10 kHz data ( Fig. 10B ).

[0142] Fig.13A and Fig. 13B Results of additional testing are shown which confirm the frequency vs. pulse width relationship just proposed. Here, data from 25 patients tested at frequencies of 10 kHz and below using sub-perceptual stimulation are shown. Fig.13ATwo different graphs are shown, showing the results for frequencies of 10k and below (lower graph) and for frequencies of 1kHz and below (upper graph). The mean shows the frequency and pulse width values ​​at which an optimized sub-perceptual treatment is produced. The upper and lower limits represent the variance of one standard deviation above and below the mean (+STD and –STD). Fig. 13B The curve fit results determined using the average value are shown. The data at 1 kHz and below were fitted using an exponential function and a power function, resulting in the relationship PW = 159e -0.01F +220e -0.00057F and PW = 761–317F 0.01 , both of which fit the data well. The data at 10kHz and below are fitted using a power function, giving PW = -1861 + 2356F -0.024 , again a good fit. The data can also be fit with other mathematical functions.

[0143] Once determined, information 350 relating to the frequency and pulse width of sub-perceptual therapy optimized for no paresthesia may be stored in an external device used to program the IPG 10 or ETS 40, such as the clinician programmer 50 or external controller 45 described previously. Fig.14 , wherein the control circuit 70 or 48 of the clinician programmer or external controller is associated with the following: regional information 100i or relationship information 98i for frequencies within the range of 1kHz to 10kHz, and regional information 300i or relationship information 310i for frequencies at or below 1kHz. Such information can be stored in a memory within the control circuit or in a memory associated with the control circuit. Storing this information using an external device is useful for helping clinicians perform sub-perceptual optimization, as further described below. Alternatively, and although not shown, information related to frequency and pulse width can be stored in the IPG 10 or ETS 40, thereby allowing the IPG or ETS to optimize itself without clinician or patient input.

[0144] The information 350 may be incorporated into the fitting module. For example, the fitting module 350 may be operated as a software module within the clinician programmer software 66, and may be implemented as a software module in the clinician programmer GUI 64 ( Figure 6 The fitting module 350 may also be operated in the control circuit of the IPG 10 or the ETS 40.

[0145] The fitting module 350 can be used to optimize the pulse width when the frequency is known, or vice versa. Fig.14As shown at the top of , a clinician or patient can input a frequency F into a clinician programmer 50 or external controller 45. The frequency F is passed to a fitting module 350 to determine a pulse width PW for the patient that is statistically likely to provide adequate pain relief without paresthesia. The frequency F can, for example, be input into a relationship 98i or 310i to determine the pulse width PW. Alternatively, the frequency can be compared to a relevant region 100i or 300i within which the frequency falls. Once the correct region 100i or 300i is determined, F can be compared to the data in the region to determine the pulse width PW, which can perhaps be taken as the pulse width between the PW+X and PW–X boundaries at a given frequency, as previously described. Other stimulation parameters (such as amplitude A, active electrodes E, their relative percentages X%, and electrode polarity P) can be determined in other ways, such as those described below, to arrive at a complete stimulation program (SP) for the patient. Based on the information from Fig. 10B Based on the data, an amplitude close to 3.0 mA may be a logical starting point, as this amplitude has been shown to be preferred by patients in the 1 kHz to 10 kHz range. However, other initial starting amplitudes may be selected, and the amplitude for sub-perceptual treatment may be frequency dependent. Fig.14 The bottom portion of FIG. 60 shows the use of the fitting module 350 in the opposite manner (i.e., picking frequency given pulse width). Note that in subsequent algorithms, or even in algorithms used outside of any algorithm, in one example, the system may allow the user to associate frequency and pulse width so that when either frequency or pulse width is changed, the other of the pulse width or frequency is automatically changed to correspond to the optimized setting. In one embodiment, associating frequency with pulse width in this manner may include an optional feature (e.g., in GUI 64) that may be used when sub-perceptual programming is desired, and associating frequency with pulse width may not be selected or selectable for use with other stimulation modes.

[0146] Fig.15 An algorithm 355 that can be used to provide subaural therapy to an SCS patient at frequencies of 10 kHz or less is shown, and summarizes some of the steps discussed above. Steps 320 to 328 describe a superaural sweet spot search. A user (e.g., a clinician) selects electrodes to create a bipole for the patient (320), such as by using a GUI of a clinician programmer. This bipole is preferably a symmetrical biphasic bipole and may include a virtual bipole, as previously described.

[0147] The bipole is transmitted to the IPG or ETS with other analog parameters using a telemetry transmitter for execution (321). Such other stimulation parameters can also be selected in the clinician programmer using a GUI. As a default, the frequency F can be equal to 90 Hz and the pulse width (PW) can be equal to 200 microseconds, although this is not strictly necessary and these values ​​can be modified. At this point, if the bipole provided by the IPG or ETS is not super-sensory, that is, if the patient does not feel paresthesia, the amplitude A or other stimulation parameters can be adjusted to make it so (322). The effectiveness of the bipole is then measured by the patient (324) to understand the extent to which the bipole is covering the patient's painful area. NRS or other scoring systems can be used to determine effectiveness.

[0148] If the bipole is invalid, or if it still needs to be searched, a new bipole can be tried (326). That is, the new electrode can preferably be selected in a manner that moves the bipole to a new position along path 296, as previously described with respect to FIG. 7A to FIG. 7D As described above. The new bipole can then again be transmitted to the IPG or ETS using a telemetry transmitter (321) and adjustments made if necessary to render the bipole super-aware (322). If the bipole is effective, or if the search has been completed and the most effective bipole has been located, the bipole can be optionally modified prior to sub-aware treatment (328). Such modifications as described above can involve selecting other electrodes adjacent to the electrode of the selected bipole to modify the field shape in the tissue to perhaps better cover the patient's pain. Thus, the modification of step 328 can change the bipole used during the search to a virtual bipole, or a triple pole, etc.

[0149] Modification of other stimulation parameters may also occur at this point. For example, the frequency and pulse width may be modified. In one example, an operating pulse width may be selected that provides good, comfortable paresthesia coverage (>80%). This may occur by using a frequency of, for example, 200 Hz, and starting with a pulse width of, for example, 120 microseconds. The pulse width may be increased at this frequency until good paresthesia coverage is felt. An amplitude, for example, in the range of 4 mA to 9 mA may be used.

[0150] At this point, the electrodes (E) selected for stimulation, their polarity (P), and the fraction (X%) of current they will receive (and possibly the operating pulse width) are known and will be used to provide sub-perceptual therapy. To ensure that sub-perceptual therapy is provided, the amplitude A of the stimulation is titrated down to a sub-perceptual, non-paresthesia level (330) and transmitted to the IPG or ETS using a telemetry transmitter. As described above, the amplitude A can be set below an amplitude threshold (e.g., 80% of a threshold) where the patient may just begin to experience paresthesias.

[0151] At this point, it may be useful to optimize the frequency and pulse width of the sub-perceptual therapy being provided to the patient (332). While the frequency (F) and pulse width (PW) used during the sweet spot search may be used for sub-perceptual therapy, it may also be beneficial to additionally adjust these parameters to optimized values ​​based on regions 100i or relationships 98i established at frequencies in the range of 1 kHz to 10 kHz, or regions 300i or relationships 310i established at frequencies at or below 1 kHz. Such optimization may be performed using Fig.14 The fitting module 350 in FIG. 1 and can occur in different ways, and some methods 332a to 332c of optimization are shown in FIG. Fig.15 Option 332a, for example, allows the software in the clinician programmer or IPG or ETS to automatically select the frequency (≤10kHz) and pulse width using the area or relationship data that relates the frequency to the pulse width. Option 332a can use the previously determined working pulse width (328) and select the frequency using the area or relationship. In contrast, option 332b allows the user (clinician) (using the GUI of the clinician program) to specify the frequency (≤10kHz) or pulse width. The software can then use the area or relationship again to select the appropriate value for other parameters (pulse width or frequency (≤10kHz)). Similarly, this option can use the previously determined working pulse width to select the appropriate frequency. Option 332c allows the user to enter the frequency (≤10kHz) and pulse width PW, but in a manner constrained by the area or relationship. Similarly, this option allows the user to enter the working pulse width and frequency suitable for the working frequency, which depends on the area or relationship. The clinician programmer's GUI 64 may not accept input of F and PW that do not fall within the region or along this relationship in this example because such values ​​would not provide optimized sub-perceptual therapy.

[0152] Frequency or pulse width optimization can occur in other ways that more efficiently search the desired portion of parameter space. For example, gradient descent, binary search, simplex method, genetic algorithms, etc. can be used for the search. Machine learning algorithms that have been trained using data from patients can also be considered.

[0153] Preferably, when optimizing frequency (≤ 10 kHz) and pulse width at step 332, these parameters are selected in a manner that reduces power consumption. In this regard, it is preferred to select the lowest frequency, as this will reduce the average charge per second (MCS), reduce the average current drawn from the battery in the IPG or ETS, and therefore increase the discharge time, as previously discussed with respect to Fig. 10B and Fig.12DReducing the pulse width (if possible) will also reduce the battery drain and increase the discharge time.

[0154] At this point, all relevant stimulation parameters (E, P, X, I, PW, and F (≤10kHz)) are determined and can be sent from the clinician programmer to the IPG or ETS for execution (334) to provide sub-perceptual stimulation therapy to the patient. Adjustment (332) of the pulse width and frequency (≤10kHz) that may be optimized can result in these stimulation parameters providing paresthesia. Therefore, if necessary, the amplitude of current A can be titrated down to sub-perceptual levels again (336). If necessary, the prescribed sub-perceptual therapy can be allowed to wash in for a period of time (338), although as previously discussed this may not be necessary because the super-perceptual sweet spot search (320-328) has already selected electrodes that are well recruited to the patient's pain site.

[0155] If sub-perceptual therapy is not effective, or adjustments may be used, the algorithm may return to step 332 to select a new frequency (≤ 10 kHz) and / or pulse width based on the previously defined zone or relationship.

[0156] It should be pointed out that Fig.15 Not all parts of the steps of the algorithm in need to be performed in an actual implementation. For example, if the effective electrodes (ie, E, P, X) are known, the algorithm can start with sub-perceptual optimization using information related to frequency and pulse width.

[0157] Fig.16 Another approach is shown, in which the fitting module 350 ( Fig.14 ) can be used to determine the optimal sub-perceptual stimulation for a patient at frequencies of 10 kHz or less. Fig.16 In the embodiment of the present invention, the fitting module 350 is again incorporated into or used by the algorithm 150, which again can be executed on the control circuit of the external device as part of its software, or executed in the IPG 10. In the algorithm 105, the fitting module 350 is used to select the initial pulse width given a specific frequency. However, the algorithm 105 is more comprehensive because it will test and optimize the amplitude and also optimize the pulse width at different frequencies. As further described below, the algorithm 105 also optionally helps to select the optimized stimulation parameters that will result in the lowest power requirements of the battery 14 of the IPG. For the algorithm 105 in Fig.16Some of the steps shown in are optional, and other steps may be added. It is assumed that the sweet spot search for the test patient by the algorithm 105 has occurred, and the electrodes (E, P, X) have been selected and preferably will remain constant throughout the operation of the algorithm. However, this is not strictly required, as these electrode parameters may also be modified, as described above.

[0158] The algorithm 105 begins by picking an initial frequency (e.g., F1) within a range of interest (e.g., ≤ 10 kHz). The algorithm 105 then passes that frequency to the fitting module 350, which picks an initial pulse width PW1 using the previously determined relationships and / or regions. For simplicity, Fig.16 The fitting module 350 is shown as a simple lookup table of pulse width and frequency, which may include another form of information related to the frequency and pulse width at which stimulation pulses are formed to provide pain relief without paresthesia. The selection of pulse width using the fitting module 350 may be more sophisticated, as previously described.

[0159] After selecting the pulse width for a given frequency, the stimulation amplitude A is optimized (120). Here, multiple amplitudes are selected and applied to the patient. In this example, the selected amplitudes preferably use the optimized amplitude A determined at each frequency (see, e.g., Fig. 10B ). Therefore, the patient tries amplitudes of A=A2, lower than (A1) and higher than (A3) over a period of time (e.g., every two days). The best of these is picked by the patient. At this point, further adjustments to the amplitude can be tried to try and hone in on an amplitude that is optimized for the patient. For example, if A2 is preferred, an amplitude slightly higher than this (A2+Δ) and an amplitude slightly lower than this (A2-Δ) can be tried over a period of time. If a lower value of A1 is preferred, an even lower amplitude (A1-Δ) can be tried. If a higher value of A3 is preferred, an even higher amplitude (A3+Δ) can be tried. Finally, iterative testing of such amplitudes arrives at an effective amplitude for the patient that does not cause paresthesia.

[0160] Next, the pulse width can be optimized for the patient (130). As with the amplitude, this can occur by slightly reducing or increasing the previously selected pulse width (350). For example, at a frequency of F1 and an initial pulse width of PW1, the pulse width can be reduced (PW1-Δ) and increased (PW1+Δ) to see if such settings made by the patient are preferred. Further iterative adjustments to the amplitude and pulse width can occur at this point, although this is not shown.

[0161] In summary, at a given frequency, an initial pulse width (350) (and preferably also an initial amplitude (120)) is selected for the patient because it is expected that these values ​​will likely provide effective and non-paresthesia pain relief. However, because each patient is different, the amplitude (120) and pulse width (130) are also adjusted according to the initial values ​​for each patient.

[0162] Thereafter, the optimized stimulation parameters determined for the patient at the tested frequency are stored in the software (135). Optionally, the average charge per second (MCS) indicating the neural delivery received by the patient, or other information indicating power draw (e.g., average Ibat, discharge time) is also calculated and stored. If yet other frequencies in the range of interest (e.g., F2) have not been tested, they are tested as described above.

[0163] Once one or more frequencies have been tested, stimulation parameters for the patient can be selected (140) using the optimized stimulation parameters (135) previously stored for the patient at each frequency. Because the stimulation parameters are applicable to the patient at each frequency, the selected stimulation parameters can include stimulation parameters that result in the lowest power draw (e.g., lowest) MSC. This is desirable because these stimulation parameters will be easiest on the IPG's battery. It can be expected that the stimulation parameters with the lowest MCS determined by algorithm 105 will include those stimulation parameters taken at the lowest frequency. However, each patient is different, and therefore this may not be the case. Once the stimulation parameters have been selected, further amplitude optimization (150) can be undertaken, where the goal of selecting the minimum amplitude is to provide subperceptual pain relief without paresthesia.

[0164] As previously mentioned, an interesting aspect of the disclosed trends and modeling of optimized sub-perceptual stimulation parameters is the realization that lower frequencies of stimulation can provide good results. As previously mentioned, using lower frequencies results in stimulation parameters that require lower power draw from the IPG. This was evaluated earlier in several different ways. For example, in Fig. 10B In , it is explained for frequencies ranging from 1 kHz to 10 kHz that optimized subperceptual stimulation parameters (including pulse width and amplitude) result in a decrease in mean charges per second (MCS) values ​​with frequency. Fig.12D We similarly see for frequencies of 1 kHz and below that the total energy expended to produce the optimal subperceptual stimulus decreases with frequency. Fig.12D In , energy draw is expressed by an estimate of the IPG's battery current (Ibat) and in terms of discharge time (ie, the amount of time an IPG with a rechargeable battery can run before needing to be recharged). Fig.12DIt is shown that lower frequencies result in lower battery currents and longer discharge times, reflecting that lower MCS values ​​and lower energy are used to form the optimized stimulation parameters.

[0165] Fig.17A and Fig. 17B The analysis of neural dose continued by again considering the mean charge per second (MCS) data as a function of frequency for the optimized stimulation parameters acquired from the patient population. Fig.17A The relationship between MSC and the logarithm of frequency for frequencies from 10 Hz to 10 kHz is shown; Fig. 17B More precisely focus on the MSC data in the lower frequency range of 10 Hz to 1 kHz. As mentioned before, the MCS can be calculated by multiplying the optimal frequency, pulse width, and amplitude for each patient tested. Since the optimized subperceptual stimulation parameters are different for each patient, the error bars are similar to Fig.17A and Fig. 17B The MCS data in is associated with + / - one standard deviation (STD). Fig.17A and Fig. 17B It can be seen that the MCS data produced by the optimized sub-perceptual parameters follows a predictable trend and increases non-linearly with frequency, as discussed further below.

[0166] exist Fig. 17B In the MSC data are analyzed with particularity for frequencies of 1kHz and below, which, as mentioned previously, are of particular interest because such frequencies provide good sub-perceptual treatment results, but at significantly lower energies. That said, although not shown, data up to 10kHz can also be analyzed and used in the various algorithms shown below, but not shown. Fig. 17B The data in are shown in Table 4 below, which also shows the average optimized stimulation parameters for the test patients:

[0167]

[0168] Table 4

[0169] Although the MCS vs. frequency data can be curve-fitted to other functions, when modeled as a polynomial (relation 380), the data fit well, with a result of MSC = -0.0002F 2 +0.55F+9.64(R 2 =0.9986), where MSC is calculated in microcoulombs per second. A statistically significant region 381i around the relationship 380 can also be established, which can be defined by various error metrics as described above. Fig. 17B, these regions 381i are bounded by standard deviation error bars. For example, at 10 Hz, MCS-STD = 6 μC / s and MCS+STD = 12 μC / s. At 50 Hz, MCS-STD = 27 μC / s and MCS+STD = 55 μC / s. Thus, the region 381a that provides good sub-perceptual treatment is defined by the linearly bounded region of points (10 Hz, 6 μC / s), (10 Hz, 12 μC / s), (50 Hz, 55 μC / s), and (50 Hz, 27 μC / s). Table 4 defines the linearly bounded region 381a. Fig. 17B Points in each of the regions 381a to 381f are shown in:

[0170] area Defined by point (Hz, μC / s) 381a (10,6),(10,12),(50,55),(50,27) 381b (50,27),(50,55),(100,88),(100,40) 381c (100,40),(100,88),(200,151),(200,67) 381d (200,67),(200,151),(400,274),(400,118) 381e (400,118),(400,274),(600,357),(600,141) 381f (600,141),(600,357),(1000,471),(1000,181)

[0171] Table 5

[0172] Of particular interest are the results for frequencies at or below 400 Hz (e.g., regions 381a to 381d). Such frequencies produce particularly low energy values ​​(i.e., low MSC values), and these frequencies are not known to the inventors who have studied in the field of sub-perceptual therapy. As mentioned elsewhere, conventional sub-perceptual therapy is believed to focus on higher stimulation frequencies.

[0173] Fig.17A and 17B The deviation in the MCS is shown to increase as frequency increases. It is hypothesized that this occurs because the MCS becomes more sensitive as frequency increases to the step size that was used to adjust the amplitude (0.1 mA) and pulse width (10 μs). This suggests that it is wise to use smaller step sizes when higher frequencies are used. In this regard, and although not shown in the figure, an algorithm can be employed to adjust the step size of the amplitude and pulse width as a function of frequency, with larger step sizes used for lower frequencies and smaller step sizes used for higher frequencies. Note that the ability to adjust the step size may depend on the clinician programming software used to adjust the stimulation parameters, as well as on the DAC circuit used for a given patient.

[0174] The data in Table 4 represent the expansion of amplitude and pulse width data across patient populations, further indicating specific optimized subperceptual parameters. Such optimized stimulation parameters can be expressed in terms of volume of frequency, amplitude, and pulse width. For example, at 10 Hz, A-STD = 1.73 mA, A+STD = 3.39 mA, PW-STD = 326 μs, and PW+STD = 374 μs. At 50 Hz, A-STD = 1.65 mA, A+STD = 3.43 mA, PW-STD = 283 μs, and PW+STD = 333 μs. Therefore, the volume that provides good sub-perceptual treatment is defined by the linear bounded volume of the following eight points: (10 Hz, 1.73 mA, 326 μs), (10 Hz, 1.73 mA, 374 μs), (10 Hz, 3.39 mA, 326 μs), (10 Hz, 3.39 mA, 374 μs), (50 Hz, 1.65 mA, 283 μs), (50 Hz, 1.65 mA, 333 μs), (50 Hz, 3.43 mA, 283 μs), (50 Hz, 3.43 mA, 333 μs). Table 6 below lists these optimized stimulation parameter volumes:

[0175]

[0176]

[0177] Table 6

[0178] The fact that the MCS of optimized stimulation parameters follows a predictable trend across patients suggests that such trend data can be used to select optimized stimulation parameters for a given patient, and Figure 17C-17E Different ways or algorithms in which this may occur are shown in . As mentioned elsewhere, the algorithms shown can be implemented as software in an external device used to control the patient's IPG or ETS. These algorithms can also be implemented in the IPG or ETS itself, which will allow at least semi-automatic selection of optimized sub-perceptual stimulation parameters for the patient.

[0179] Algorithms for selecting optimized subthreshold stimulation parameters can use MSC data alone, or can also take into account other previously established modeling data, such as the relationship between frequency and pulse width (see, e.g. Fig. 10A and Fig. 12A ). Fig. 17C The algorithm 379 in step 382 uses this frequency and pulse width data to select frequencies and pulse widths that are consistent with the various relationships (98i, 310i) or statistically significant regions (100i, 300i) described previously. Fig. 17C An example of frequency versus pulse width relationship 310 and region 300c is reproduced in the left figure of FIG. Fig. 17CIn the example shown, a frequency of 300 Hz and 200 microseconds are selected at step 382, ​​and the point falls within region 300c (see Fig. 12A Alternatively, using relationship 310 (see Fig. 12B ) can also be used to select the frequency and pulse width.

[0180] Next, in step 383, an MCS value or range is determined using the selected frequency, and one of the relationships 380 or regions 381i. For example, using relationship 380, when F = 300 Hz, an MSC value of 157 μC / s can be calculated. Alternatively, an error bar defining region 381d (where F = 300 Hz falls) allows the MSC range to be determined at that frequency, such as between approximately 96 and 273 μC / s, as shown in FIG. Fig. 17C as shown in the figure on the right.

[0181] Based on the determined MCS value (or range), an amplitude value (or range) can be determined, as shown in step 384. The amplitude will include the MSC value (or range) divided by the product of the selected frequency (300 Hz) and the pulse width (200 μs), which in the example shown produces a single amplitude value of 2.61 mA, or an amplitude range of 1.6 mA to 4.55 mA. In short, algorithm 379 produces optimized sub-perceptual stimulation parameters, including frequency (e.g., 300 Hz), pulse width (e.g., 200 μs), and amplitude (e.g., 2.61 mA). Because the underlying data can reflect some extension (e.g., regions 300, 381), algorithm 379 can determine multiple potential candidate stimulation parameter sets to try for a given patient. Multiple different stimulation sets can also be applied to the patient in a time-multiplexed manner. This will be referred to later. FIG. 34A to FIG. 35 Describe in more detail.

[0182] The trend of MSC vs. frequency suggests that it may not be strictly necessary to select pulse width according to frequency (or vice versa) when selecting optimized subperceptual parameters, and Fig.17D Another example of an algorithm 385 is shown that can be used to select optimized sub-perceptual stimulation parameters. In this example, the frequency versus pulse width trends discussed previously (e.g., regions 100i, 300i or relationships 98i and 310i) are not used. Instead, only the MSC versus frequency trends are used (using regions 381i or relationship 380), with the goal of selecting a frequency, pulse width, and amplitude that is consistent with such trend data. The operation of algorithm 385 can select a single set of stimulation parameters (i.e., a single frequency, pulse width, and amplitude), although Fig.17D A selection of three such parameter sets is shown. Parameter set "A" includes the set F = 150 Hz, PW = 100 μs and A = 4 mA, which results in an MCS value of 60 μC / s. This stimulus parameter set "A" is shown in FIG. Fig.17D , and note that it is within the statistically significant region 381c, although it is not a point on relationship 380. Parameter set "B" includes the set F = 500 Hz, PW = 100 μs, and A = 4.7 mA, which results in an MCS value of 234 μC / s. This point is within region 381e, although it is also assumed that this includes a point on relationship 380 - that is, when F = 500 Hz, relationship 380 calculates the MSC as 234. Parameter set "C" includes the set F = 800 Hz, PW = 87.5 μs, and A = 5 mA, which results in an MCS value of 350 μC / s. This stimulation parameter set "C" is within region 381f, although it is not a point on relationship 380.

[0183] Fig.17D The resulting waveforms for these different stimulation parameter sets are also shown in FIG. 1 . In this example, it is assumed that active charge recovery is used and that the active charge recovery phase (the second phase of each pulse) is symmetrical to the first phase (but of opposite polarity). However, this is not strictly necessary and the charge recovery phase may include an asymmetrical active or passive phase, as described previously and subsequently as shown in FIG. Fig.17E Note that these stimulation parameter sets may or may not include frequencies and pulse widths consistent with the previously explained pulse width to frequency trends (100i, 300i, 98i, 310i). Similar to algorithm 379, algorithm 385 may select a single stimulation parameter set for a given patient, or multiple such parameter sets may be attempted or applied in a time multiplexed manner.

[0184] Because the MSC and frequency data ultimately reflect the energy of the stimulation, other algorithms can determine waveforms for use with a patient that are not strictly defined by pulses, or at least not strictly defined by constant frequency, pulse width, or amplitude. Fig.17E The algorithm 387 in selects a waveform that conforms to the MSC and frequency trend data (region 381 or relationship 380), wherein the waveform has an irregular shape. For example, the pulses in the waveform selected by the algorithm 387 may not have a constant amplitude, or may be subdivided into multiple pulses.

[0185] For example, the pulses in stimulation parameter set "D" have the same pulse width (100 μs) and frequency (150 Hz) as the pulses in set "A" ( Fig. 17C ). However, the amplitude of the pulse is not constant, but ramps up from 0 to a maximum amplitude (Amax) of 8 mA over the pulse width. Therefore, the area under the pulse in set "D" (Area 1) is equal to the area under the pulse in set "A". Since the MSC can also be defined as the pulse area multiplied by the frequency, the MSC for the pulses in set "A" and set "D" is the same (60 μC / s), which is again consistent with the MSC vs. frequency trend mentioned (380, 381).

[0186] In another example, the pulses in stimulation parameter set "E" have the same maximum amplitude (4.7 mA) and frequency (500 Hz) as the pulses in set "B" ( Fig. 17C ). However, the pulse width is doubled (to 200 μs), and the amplitude ramps up and down during the pulse width. Therefore, the area under the pulse in set “E” is 2 ) is equal to the area under the pulse in set "B", and both sets again have the same MSC value (234 μC / s), which is again consistent with the MSC trend data.

[0187] Stimulation parameter set "F" shows that each pulse can be formed by algorithm 387 into a group of pulses, sometimes called a "burst" of pulses. In this example, the pulses are combined with the pulses in set "C" ( Fig. 17C ) and have the same pulse width (87.5 μs) and frequency (800 Hz). However, set "F" creates multiple pulses (two, but there could be more) for each pulse shown in set "C" and their amplitude is half (2.5 mA). The sum of the areas under the two pulses in set "F" (Area 3 ) is equal to the area under the single pulse in set "C", and thus the pulses in set "C" and set "F" again have the same MCS value (350 μC / s), which is again consistent with the MSC trend data. In short, algorithm 387 allows for a more general definition of waveforms that are consistent with the MSC trend data.

[0188] Thus far, MSC trend data (e.g., 380, 381) have been shown to vary with frequency. However, MSC trend data useful in the disclosed algorithm may also be defined with respect to other stimulation parameters. For example, Fig.17F As shown in , and as shown in Table 4 above, MSC values ​​predicting good subperceptual treatment also vary with pulse width in a predictable manner, as shown in relationship 380'. This relationship 380' can also be modeled (e.g., curve fit) and associated with error bars to determine area 381', which can then be used in the algorithm just described to help select optimized subperceptual stimulation parameters. Fig.17FThe variation of MSC values ​​with amplitude is also shown in Table 4 (and again taken from Table 4). Note that the amplitude is fairly constant (e.g., approximately 2.6 mA) as MSC varies. While this trend may not be very useful in selecting stimulation parameters, it is noted that it is consistent with previously reported optimized subperceptual stimulation amplitudes (see, e.g., Fig. 10B ).

[0189] Further investigation revealed Figure 18-23D , the purpose is to provide optimized sub-perceptual modeling that takes into account the perception threshold (pth) as well as the frequency (F) and pulse width (PW). The perception threshold may be an important factor to consider when modeling sub-perceptual stimulation and using such modeling information to determine the optimized sub-threshold stimulation parameters for each patient. The perception threshold pth includes the lowest amplitude (e.g., in mA) at which the patient can feel the effects of paresthesia, and amplitudes below this will cause sub-perceptual stimulation. In fact, different patients will have different perception thresholds. The different perception thresholds are caused in large part because the electrode arrays in some patients may be closer to the spinal nerve fibers than other patients. Therefore, such patients will experience perception at a lower amplitude, that is, for these patients, pth will be lower. If the electrode arrays in other patients are farther away from the spinal nerve fibers, the perception threshold pth will be higher. The improved modeling takes into account the understanding of pth, because in addition to the optimized frequency and pulse width, the inclusion of this parameter can also be used to suggest an optimized amplitude A for the patient's sub-perceptual stimulation.

[0190] With this in mind, data were obtained from patients to determine not only the frequencies and pulse widths that they found to be optimal as previously described, but also the perception thresholds at those frequencies and pulse widths. The resulting model 390 was Fig.18 This model 390 was determined based on testing of a sample of patients (N=25), where Fig.18 The average values ​​determined by the three-dimensional regression fit are shown, which produces the model 390 as a surface in the frequency-pulse width-perception threshold space. Fig.18 The data presented were acquired at frequencies of 1 kHz and below. The data at these frequencies are of particular interest because, as already mentioned, lower frequencies are more of a concern for energy usage in IPG or ETS, and therefore, there is a particular need to demonstrate the utility of sub-perceptual stimulation in this frequency range. Fig.18As can be seen from the equations in Figure 390, by assuming that the frequency varies according to the power function with the pulse width (a(PW)b) and the perception threshold pth (c(pth)d), the data obtained from the patient is modeled with a good fit. Although these functions provide a suitable fit, other types of mathematical equations can also be used for the fit. Model 390 gives the following result as a surface fit: F(PW, pth) = 4.94x10 8 (PW)-2.749+1.358(pth) 2 Note that frequency, pulse width, and perception threshold are not simply proportionally or inversely related as in model 390, but are instead related by a nonlinear function.

[0191] Fig.19A Further observations noted from the tested patients are shown and provide another modeling aspect that, together with model 390, can be used to determine optimized subthreshold stimulation parameters for the patient. Fig.19A shows how the perception threshold pth of the tested patients varies according to the pulse width, where each patient Fig.19A The analysis of each line shows that the relationship between pth and PW can be well modeled using a power function, ie, pth(PW) = i(PW)j + k, although other mathematical functions can also be used for fitting. Fig.19A The data were acquired for each patient at nominal frequencies (such as 200 to 500 Hz), and further analysis confirmed that the results did not vary significantly with frequency (at least at frequencies of 150 Hz and higher, using biphasic pulses with active charging). Fig.19A The pulse widths in the were limited to the range of approximately 100 to 400 microseconds. It was reasonable to restrict the analysis to these pulse widths because previous tests (e.g. Fig. 12A ) shows that pulse widths within this range have only subperceptible therapeutic effects at frequencies of 1 kHz and below. Fig.19B Another example of pth versus pulse width for different patients is shown, and another equation that can be used to model the data is shown. Specifically, the Weiss-Lapicque or intensity-duration equation is used in this example, which relates the amplitude and pulse width required to reach the threshold. The equation takes the form of pth=(1 / a)(1+b / PW), and when the data for different patients are averaged, constants a=0.60 and b=317 have good fit results, where these values ​​represent the average constant parameters extracted from the population data.

[0192] Fig. 20 A further observation noted from the patients tested is shown and provides yet another aspect of modeling. Fig. 20It shows how the patient's optimized subperceptual amplitude A varies depending on the patient's perception threshold pth and pulse width. Fig. 20 In the graph of , the vertical axis plots the parameter Z, which is related to the patient's perception threshold pth and its optimized amplitude A (which will be lower than pth in subperception treatment). Specifically, Z is the optimized amplitude expressed as a percentage of pth, i.e., Z = A / pth. Fig. 20 As shown, Z varies with pulse width. At smaller pulse widths (e.g., 150 microseconds), Z is relatively low, meaning that the patient's optimized amplitude A is noted to be well below their perception threshold (e.g., A=40% of pth). At longer pulse widths (e.g., 350 microseconds), Z is higher, meaning that the patient's optimized amplitude A is noted to be closer to their perception threshold (e.g., A=70% of pth). As noted from testing various patients, Z and PW generally have a linear relationship over the pulse widths tested, and therefore linear regression is used to determine the relationship between them, yielding Z=0.0017(PW)+0.1524(395). Similarly, Fig. 20 The testing in was limited to the approximate range of 100 to 400 microseconds, which was noted to be useful for subperceptual treatments below 1 kHz. It is expected that testing over a wider range of pulse widths (e.g., less than 100 microseconds or greater than 400 microseconds) will show some variation from the linear relationship noted. For example, for pulse widths greater than 400 microseconds, Z may flatten out to some value less than 1, and for pulse widths less than 100 microseconds, Z may flatten out to a value greater than 0. Because Z varies with pulse width as the curve fit does, and because Z also varies with the optimized amplitude A and perceptual threshold pth (Z = A / pth), Fig. 20 Modeling allows the optimized amplitude A to be modeled as a function of the perception threshold pth and the pulse width PW, i.e., A = pth [0.0017 (PW) + 0.1524] (396). The inventors observed that the optimized amplitude A is generally invariant to changes in frequency and pulse width. However, the perception threshold varies with pulse width. Therefore, Z varies with pulse width, while the optimized amplitude A may not.

[0193] Recognizing and modeling these observations, the inventors have developed an algorithm 400 that can be used to provide personalized sub-perceptual therapy for a particular patient. The algorithm 400 can be largely implemented on a clinician programmer 50 and results in the determination of a range of optimized sub-perceptual parameters (e.g., F, PW, and A) for the patient. Preferably, as a final step of the algorithm 400, the range or volume of optimized sub-perceptual parameters is sent to the patient's external controller 45 to allow the patient to adjust their sub-perceptual therapy within this range or volume.

[0194] from Fig.21A The algorithm 400 shown at the beginning begins in step 402 by determining the sweet spot in the electrode array on which therapy should be applied for a given patient, i.e., by identifying which electrodes should be activated and with what polarity and percentage (X%). For a given patient, the results of the sweet spot search may be known, and therefore step 402 should be understood to be optional. Step 402 and subsequent steps can be accomplished using the clinician programmer 50.

[0195] At step 404, a new patient is tested by providing a situation pulse, and in algorithm 400, such testing involves measuring the patient's perception threshold pth at various pulse widths using the sweet spot electrode that has been identified at step 402 during the test procedure. Fig.19A and Fig.19B As discussed, testing of different pulse widths may occur at a nominal frequency, such as in the range of 200 to 500 Hz. Determining the PTH at each given pulse width involves applying the pulse width and gradually increasing the amplitude A to the point at which the patient reports feeling stimulation (paresthesia), thereby producing a PTH expressed in amplitude (e.g., milliamperes). Alternatively, determining the PTH at each given pulse width may involve decreasing the amplitude A to the point at which the patient reports no longer feeling stimulation (subthreshold). Fig.21A 404. Here, it is assumed that the patient in question has a paresthesia threshold pth of 10.2 mA at a pulse width of 120 microseconds. At a pulse width of 350 microseconds, the pth is 5.9 mA, and other values ​​are between these.

[0196] Next, in step 406, the algorithm 400 in the clinician programmer 50 models the pthv.PW data points measured in step 404 and curve fits them to a mathematical function. This mathematical function may be noted earlier to model pth and PW in other patients well, such as a power function pth(PW)=i(PW) j + k or Weiss-Lapicque equation, as previously discussed Fig.19A and Fig.19B However, any other mathematical function can be used to curve fit the data measurement results of the current patient, such as polynomial functions and exponential functions. In the data shown, the power function models the data well, resulting in pth(PW) = 116.5xPW -0.509 (For simplicity, the constant "k" has been ignored.) The measured data in table 404 and the determined curve fit relationship pth(PW) 406 determined for the patient may be stored in the memory of clinician programmer 50 for use in subsequent steps.

[0197] Next, and refer to Fig. 21B , the algorithm 400 continues by comparing the pth(PW) relationship determined in step 406 with the model 390. This will refer to Fig. 21B The table shown in FIG. 1 is used to illustrate this. In this table, the Fig.21A As can be seen, discrete pulse width values ​​of interest (100 microseconds, 150 microseconds, etc.) can be used (which may be different from the exact pulse width used during the patient test in step 404). Although Fig. 21B Only six rows of PWv.pth values ​​are shown in the table of , but this could be a longer vector of values ​​where pth is determined in discrete PW steps (such as 10 microsecond steps).

[0198] In step 408, the pthv.PW values ​​(from function 406) are compared to the three-dimensional model 390 to determine the frequency F that is optimal at these various pthv.PW pairs. In other words, the pth and PW values ​​are provided as variables to Fig.18 The surface fitting equation (F(PW, pth)) 390 in FIG. 1 is used to determine the optimized frequencies, which are also shown as being filled into Fig. 21B At this point, Fig. 21B The table in represents a vector 410 that relates to pulse width and frequency that are optimized for the patient and also includes the perception threshold for the patient at these pulse width and frequency values. In other words, vector 410 represents the values ​​within model 390 that are optimized for the patient. Note that vector 410 for the patient can be represented as a curve along three-dimensional model 390, such as Fig. 21B As shown in .

[0199] Next, and if Fig. 21C As shown in step 412 of , vector 410 can optionally be used to form another vector 413 that contains values ​​of interest or, more realistically, values ​​that can be supported by the IPG or ETS. For example, note that vector 410 for the patient includes frequencies at higher values, such as 1719 Hz, or frequencies at odd values, such as 627 and 197 Hz. It may not be desirable to use frequencies at higher values ​​because such frequencies, even if effective for the patient, may involve excessive power draw. See, e.g. Fig.12D. In addition, the IPG or ETS in question may only be able to provide pulses with frequencies at discrete intervals (such as in increments of 10 Hz). Therefore, in vector 413, a frequency of interest or supported frequency (e.g., 1000 Hz, 400 Hz, 200 Hz, 100 Hz, etc.) is selected, and then the corresponding values ​​of PW and pth are interpolated using vector 410. Although not shown, it may be useful to formulate vector 410 as an equation F(PW, pth)) to make vector 413 easier to fill in. Nevertheless, vector 413 includes essentially the same information as vector 410, albeit at the desired frequency. It is recognized that the IPG or ETS may only support certain pulse widths (e.g., in increments of 10 microseconds). Therefore, although this is not shown in the accompanying drawings, the pulse widths in vector 413 can be adjusted (e.g., rounded) to the nearest supported value.

[0200] Next, and refer to Fig.21D , algorithm 400 determines in step 414 the optimized magnitude of the pulse width and pth value in vector 413 (or vector 410 if vector 413 is not used). This is done by using Fig. 20 This occurs using the amplitude function 396 determined earlier in the table, namely A(pth, PW). Using this function, an optimized amplitude A can be determined for each pth, PW pairing in the table.

[0201] At this point, in step 416, the optimized subthreshold stimulation parameters F, PW, A420 are determined as a patient-specific model. The optimized stimulation parameters 420 may not need to include the perception threshold pth: although pth is useful for determining the patient's optimized subthreshold amplitude A (step 414), it may no longer be a parameter of interest because it is not a parameter generated by the IPG or ETS. However, in other examples discussed later, it may be useful to include pth in the optimized parameters 420 because this may allow the patient to adjust their stimulation to a supra-perceptual level if desired. At this point, the optimized stimulation parameters 420 may then be sent to the IPG or ETS for execution, or, as shown in step 422, they may be sent to the patient's external controller 45, as described below.

[0202] Fig.21E and Fig.21F The optimized parameters 420 are depicted in graphical form. Although the optimized parameters 420 in this example include three-dimensional ranges or coordinate lines (F, PW, and A), for ease of illustration, they are depicted in two two-dimensional graphs: Fig.21E shows the relationship between frequency and pulse width, and Fig.21F shows the relationship between frequency and amplitude. It should also be noted that Fig.21FThe paresthesia threshold pth is shown, and the X-axis also shows the Fig.21E The pulse widths corresponding to various frequencies are shown in Table 1. Note that the shape of the data on these plots can vary from patient to patient (e.g., based on Fig.21A pth measurement results), and may also vary depending on the underlying modeling used (e.g., Figure 18-Figure 20 ). Therefore, the various shapes of the trends shown should not be interpreted as limiting.

[0203] The optimized stimulation parameters 420 determined by the algorithm 400 include a range or vector of values, including those based on modeling ( Figure 18-Figure 20 ) and patient testing ( Fig.21A , the frequency / pulse width / amplitude coordinates of step 404) will result in a subthreshold stimulation that is optimized for this patient. Although for simplicity, in Fig. 22 Optimized parameters 420 are shown in tabular form in FIG. 4 , but it should be understood that these optimized parameters (O) can be curve-fitted using an equation involving frequency, pulse, and amplitude (i.e., O=f(F, PW, A)). Because each of these coordinates is optimized, it may be reasonable to allow the patient to use it with their IPG or ETS, and as a result, the optimized parameters 420 can be sent from the clinician programmer 50 to the patient external controller 45 ( Figure 4 ) to allow the patient to choose between them. In this regard, the optimized parameters 420, whether in table form or equation form, can be loaded into the control circuit 48 of the external controller 45.

[0204] Once loaded, the patient can access a menu in the external controller 45 to adjust the therapy provided by the IPG or ETS in accordance with these optimized parameters 420. For example, Fig. 22A graphical user interface (GUI) of an external controller 45 is shown as displayed on its screen 46. The GUI includes a device that allows the patient to adjust the stimulation simultaneously within the range of the determined optimized stimulation parameters 420. In one example, a slider with a cursor 430 is included in the GUI. The patient can select the cursor 430 and, in this example, move the cursor left or right to adjust the frequency of the stimulation pulses in their IPG or ETS. Moving it to the left reduces the frequency to the minimum value (e.g., 50 Hz) included in the optimized parameters 420. Moving the cursor 430 to the right increases the frequency to the maximum value (e.g., 1000 Hz) included in the optimized parameters. As the cursor 430 moves and the stimulation frequency changes, the pulse width and amplitude will be adjusted simultaneously, as reflected by the optimized parameters 420. For example, at F=50 Hz, the amplitude is automatically set to A=4.2 mA, and the pulse width is set to 413 microseconds. At F=1000 Hz, the amplitude is set to A=3.7 mA, and the pulse width is set to 132 microseconds. In effect, cursor 430 allows the patient to browse through optimized parameters 420 to find their preferred F / PW / A settings, or simply select stimulation parameters that are still effective but require drawing less power from the IPG or ETS (e.g., at a lower frequency). Note that frequency, pulse width, and amplitude may not be adjusted proportionally or inversely relative to each other, but will follow a nonlinear relationship based on the underlying modeling.

[0205] In another example, it may be useful to allow the patient to adjust the stimulation without knowing the stimulation parameters, i.e., without displaying the parameters, which may be too technical for the patient to understand. In this regard, the sliders may be labeled with more general parameters, such as The patient can adjust this parameter, such as between 0% and 100%. The three-dimensional simulation parameters A, PW and F can be mapped to the one-dimensional parameter (For example, as shown, 4.2 mA, 413 μs, and 50 Hz may equal 0%). In general, the patient may set the parameters It is understood as a kind of "intensity" or "neural dose" with increasing percentages. In practice, it depends on the optimized stimulation parameters 420 being mapped to In another modification not shown, the mean charge per second (MSC) model (e.g., 380, 381) can be used to select the optimized stimulation parameters, as previously described with respect to Figures 17A-17F The GUI may then also allow the user to navigate within the MSC model to select optimized stimulation parameters.

[0206] It should be understood that while the GUI of the external controller 45 does allow the patient some flexibility to modify the stimulation parameters of their IPG or ETS, it is also simple and advantageously allows the patient to adjust all three stimulation parameters simultaneously using a single user interface element while ensuring that the resulting stimulation parameters will provide optimized subthreshold stimulation.

[0207] Other stimulation regulation controls may also be provided by the external controller 45. For example, Fig. 22 As shown in , another slider may allow the patient to adjust the duty cycle to control the extent to which the pulse will run continuously (100%) or be completely off (0%). An intermediate duty cycle (e.g., 50%) would mean that the pulse would run for a period of time (from seconds to minutes) and would then be off for the same duration. Since "duty cycle" may be a technical concept that patients may not intuitively understand, note that the duty cycle may be labeled in a more intuitive manner. Therefore, and as shown, the duty cycle adjustment may be labeled differently. For example, since a lower duty cycle affects a lower power draw, the duty cycle slider may be labeled as a "power save" function, a "total energy" function, a "total neural charge dose" function, or the like, which may be easier for the patient to understand. The duty cycle may also include a function that is locked into the patient's external controller 45 and is only accessible to the clinician after, for example, entering an appropriate password or other credentials. Note that the duty cycle can be adjusted smoothly or in preset logical increments (such as 0%, 10%, 20%, etc.). For simplicity, duty cycle adjustment is not shown in subsequent user interface examples, but may be used in such examples as well.

[0208] Figure 23A-23D The practicality of modeling that results in determining the optimized parameters 420 may be imperfect is addressed. For example, the model 390 (frequency as a function of PW and pth (F(PW, pth); Fig.18 ) is averaged across patients and can have some statistical variance. Fig.23A This is briefly illustrated by showing surfaces 390+ and 390- that are higher and lower than the average value reflected in the surface model 390. The surfaces 390+ and 390- may represent some degree of statistical variance or error measure, such as plus or minus one sigma, and in fact may generally include error bars beyond which the model 390 is no longer reliable. These error bars 390+ and 390- (which may not be constant across the surface 390) may also be determined from an understanding of the statistical variance of the various constants assumed during modeling. For example, different confidence measures may be utilized to determine the values ​​a, b, c, and d in the model 390. As shown in FIG. Fig.18 As shown in Figure 1, the constants a, b, and c vary within a 95% confidence interval. For example, the constant "a" can range from 5.53x107 to 9.32x10 8 , as shown in the figure. (Here, it is assumed that the constant d is only 2 and does not change.) Similarly, the value used to model the relationship between pth and pulse width ( Fig.19A and Fig.19B ) may have different confidence measures, and the values ​​m and n used to model the relationship between the optional amplitude, pth, and PW may also be different ( Fig. 20 ). Over time, as data from more patients is collected, it is expected that the confidence in these models will increase. In this regard, it is noted that the algorithm 400 can be easily updated from time to time with new modeling information by loading the new modeling information into the clinician programmer 50.

[0209] Statistical variance means that the optimized stimulation parameters may not include discrete values, but instead may fall within a volume. Fig.23A 410 (see Fig. 21B ). Given the statistical variance, vector 410 may include a rigid line within volume 410'. In other words, there may not be a one-to-one correspondence between PW, pth, and F, such as Fig. 21B Alternatively, for any given variable (such as pulse width), the pth determined for the patient (using the pth(PW) model in step 406) may vary within a range between a statistically significant maximum and minimum value, such as Fig. 23B Model 390 ( Fig.18 ) may also mean that the maximum and minimum frequencies may be determined for each maximum and minimum pth in step 408. Since this is trickled through the algorithm 400, the optimized stimulation parameters 420 may also not have a one-to-one relationship between frequency, pulse width, and amplitude. Instead, and as Fig. 23B As shown in , for any frequency, there may be a range of maximum and minimum optimized pulse widths that are statistically significant, and likewise a range of optimized amplitudes A. Effectively, then, the optimized stimulation parameters 420' may define a coordinate volume that is statistically significant in the frequency-pulse width-amplitude space rather than in the coordinate lines. The paresthesia threshold pth may also vary within a range, and as mentioned previously, it may be useful to include it in the optimized stimulation parameters 420', because pth may help allow the patient to change the stimulation from sub-perception to suprasensory, as discussed in some of the later examples.

[0210] Fig.23C and Fig.23DThe optimized parameters 420' are depicted in graphical form, showing the statistically relevant range of pulse widths and the statistically relevant range of amplitudes suitable for the patient at each frequency. Although the optimized parameters 420' in this example include three-dimensional coordinate volumes (F, PW, and A), for ease of illustration, they are depicted in two two-dimensional graphs, similar to those previously described in Fig.21E and Fig.21F What happens in: Fig.23C shows the relationship between frequency and pulse width, and Fig.23D shows the relationship between frequency and amplitude. Also note that Fig.23D The paresthesia threshold pth is shown, which, like the pulse width and amplitude, can vary statistically within a certain range. Also shown are the optimized stimulation parameters 420 for each of the parameters (determined without statistical variance, see Fig.21E and Fig.21F ), and as expected falls within the wider volume of the parameters specified by 420'.

[0211] In case a volume of optimized parameters 420' is defined, it may then be useful to allow the patient to navigate different settings within this volume of optimized parameters 420' using his external controller 45. Fig.23E . Here, the GUI of the external controller 45 does not display a single linear slider, but rather displays a three-dimensional volume representing the volume 420' of the optimized parameters, where the different axes represent the changes that the patient can make in frequency, pulse width and amplitude. As previously described, the GUI of the external controller 45 allows the patient some flexibility to modify the stimulation parameters of their IPG or ETS, and allows the patient to adjust all three stimulation parameters simultaneously with one adjustment action and using a single user interface element.

[0212] Different GUIs are possible that allow the patient to browse the determined optimized parameter volume 420', and Fig.23F Another example is shown. Fig.23F, two sliders are shown. The first linear slider, controlled by cursor 430a, allows the patient to adjust the frequency based on the frequency reflected in the optimized volume 420'. The second two-dimensional slider, controlled by cursor 430b, allows the patient to adjust the pulse width and amplitude at that frequency. Preferably, the range of pulse width and amplitude is constrained by the optimized parameters 420' and the frequency that has been selected using cursor 430a. For example, if the user chooses to use a frequency F=400 Hz, the external controller 45 can consult the optimized parameters 420' to automatically determine the optimized range of pulse width (e.g., 175 to 210 microseconds) and amplitude (3.7 to 4.1 mA) for the patient to use at that frequency. When the patient changes the frequency using cursor 430a, the range of allowed pulse width and amplitude that can be selected using cursor 430b will automatically change to ensure that subthreshold stimulation remains within the volume 420' determined to be statistically useful for the patient. In another modification not shown, the mean charge per second (MSC) model (e.g., 380, 381) can be used to select optimized stimulation parameters, as previously described with respect to Figures 17A-17F The GUI may then also allow the user to navigate within the MSC model to select optimized stimulation parameters.

[0213] Fig.24 Another example is shown in which a user can use the derived optimized stimulation parameters to program the settings of their IPG 10 (or ETS). The subsequent examples for completeness use the determined volume 420' of the optimized stimulation parameters, but a vector or range 420 of optimized stimulation parameters may also be used. The average charge per second (MSC) model (e.g., 380, 381) may also be used to select the optimized stimulation parameters, as previously described with respect to Figures 17A-17F described.

[0214] Fig.24A user interface is shown on the screen 46 of the patient's external controller 45 that allows the patient to select from a variety of stimulation modes. Such stimulation modes may include various ways in which the IPG may be programmed to be consistent with the optimized stimulation parameters 420' determined for the patient, such as: an economy mode 500 that provides stimulation parameters with low power draw; a sleep mode 502 that optimizes the stimulation parameters for the patient while sleeping; a sensory mode 504 that allows the patient to feel the stimulation (supersensory); a comfort mode 506 for general daily use; an exercise mode 508 that provides stimulation parameters suitable for when the patient is exercising; and an intense mode 510 that may be used, for example, if the patient is experiencing pain and would benefit from more intense stimulation. Such stimulation modes may be indicative of the patient's posture or activity. For example, the sleep mode 502 provides stimulation optimized for sleep (e.g., when the patient is lying down and not moving significantly), while the exercise mode 508 provides stimulation optimized for exercise (e.g., when the patient is standing up and moving significantly). Although not shown, stimulation modes may also be included that provide stimulation optimized for different patient postures (such as supine, prone, standing, sitting, etc.) or for different conditions (such as cold or inclement weather). Although described in the context of an external controller for the patient, it is also recognized that in other examples, another external device that can be used to program the patient's IPG can also be used to select a stimulation mode, such as a clinician programmer 50.

[0215] The patient can select from these stimulation modes, and such selection can program the IPG 10 to provide a subset of stimulation parameters useful for that mode controlled by the optimized stimulation parameters 420'. For certain stimulation modes, the subset of stimulation parameters may be completely constrained by (completely within) the volume of the optimized stimulation parameters 420' determined for the patient, and will therefore provide the patient with optimized sub-perceptual stimulation therapy. As further explained below, subsets of other modes may be only partially constrained by the optimized stimulation parameters. However, in all cases, the subset is determined using the optimized stimulation parameters (420 or 420'). Preferably, the subset is determined for the patient at the clinician programmer 50 and sent to the patient's external controller 45 and stored therein. Alternatively, the determined optimized stimulation parameters may be sent to the external controller 45, leaving it to the external controller 45 to determine the subset from the optimized stimulation parameters.

[0216] The number of stimulation modes available for the patient to select on the external controller 45 can be limited or programmed by the clinician. This may be warranted because certain stimulation modes may not be relevant to certain patients. In this regard, the clinician can program the patient's external controller 45 to specify the available stimulation modes, such as by entering an appropriate clinician's password. Alternatively, the clinician can program the external controller 45 using a clinician programmer 50 to program the external controller 45.

[0217] An example of a subset of stimulation parameters 425x is shown in Figures 25A-30B Shown in. Fig.25A and Fig.25B A subset 425a of stimulation parameter coordinates used when the economy mode 500 is selected is shown, which includes a subset of the optimized stimulation parameters 420' with low power draw. Like the optimized stimulation parameters 420', the subset 425a may include three-dimensional volumes of F, PW, and A parameters, and also (compare Fig.23C and Fig.23D ) uses two two-dimensional graphs to represent subset 425a, where Fig.25A shows the relationship between frequency and pulse width, and Fig.25B The relationship between frequency and amplitude is shown.

[0218] To affect low power draw, the frequencies within subset 425a are low, such as limited to a frequency range of 10 to 100 Hz, even though the optimized stimulation parameters 420' may have been determined over a wider range, such as 10 to 1000 Hz. Furthermore, while the optimized pulse widths within this frequency range may vary more significantly among the optimized stimulation parameters 420', subset 425a may be constrained to the smaller of these pulse widths, such as the lower half of such pulse widths, as shown. Fig.25A Similarly, using a lower pulse width will result in lower power draw. Fig.25B As shown in , subset 425a can be constrained to a lower amplitude within the optimized stimulation parameters 420' for the relevant frequency range, again resulting in lower power draw. In short, subset 425a can include a smaller volume of stimulation parameters within the entire volume of optimized stimulation parameters 420' that provide sufficient subthreshold stimulation for the patient, while providing lower power draw from the IPG battery 14. Not all stimulation modes ( Fig.24 ) The corresponding subsets 425x all contain stimulation parameters, which must be completely within the determined optimized stimulation parameters 420', as shown in some subsequent examples.

[0219] When the economy mode 500 is selected, the external controller 45 may simply send a single low power optimized parameter (F, PW, A) within the subset 425a to the IPG for execution. However, and more preferably, the user interface will include means for allowing the patient to adjust the stimulation parameters to those within the subset 425a. In this regard, the user interface may include a slider interface 550 and a parameter interface 560. The slider interface 550 may be as described previously (see Fig. 22 ), and a cursor may be included to allow the patient to slide through the parameters in subset 425a. In the example shown, the slider interface 550 may not adjust the pulse width, which is set to a specific value (e.g., 325 μs), but the frequency and amplitude may vary. This is just one example, and in other examples, all three of the frequency, pulse, and amplitude may be changed via sliders, or other parameters may be held constant. Note that a more sophisticated user interface may be used to allow the patient to navigate through subset 425a. For example, although not shown, a user interface element with a higher three-dimensional quality, such as previously described in Fig.23E and Fig.23F Those discussed in can be used to navigate the volume of subset 425a. Parameter interface 560 can also allow the patient to navigate parameters within subset 425a, and is simply shown as having selectable buttons to increase or decrease parameters within determined subset 425a. Parameter interface 560 can also include fields showing current values ​​of frequency, pulse width, and amplitude. Initially, these values ​​can be populated with parameters roughly in the center of determined subset 425a, allowing the patient to adjust stimulation near the center.

[0220] Fig.26A and Fig.26B 5 shows a selection of sleep mode 502, and a subset 425b of optimized stimulation parameters 420' that is generated when the selection is selected. In this example, subset 425b is determined using optimized stimulation parameters 420' in a manner such that subset 425b is only partially constrained by optimized stimulation parameters 420'. Subset 425b may include low to medium frequencies within optimized stimulation parameters 420' (e.g., 40 to 200 Hz), and may include medium pulse widths within this frequency range that are allowed by 420', such as by Fig.26A shown.

[0221] Because the intensity of stimulation during sleep may not need to be that high, the amplitudes within subset 425b may fall outside the amplitudes suggested by optimized parameters 420', such as Fig.26B. For example, while the optimized parameters 420' may, for example, suggest that for the frequency and pulse width range of interest, the amplitude based on earlier modeling would fall within the range of 3.6 to 4.0 mA, the amplitude within subset 425b is set to an even lower value in this example. Specifically, as shown in the slider interface 550, the amplitude may be set between 1.5 mA and 4.0 mA. In order to know where the lower boundary of the amplitude should be set, the modeling information may include an additional model 422, which may be determined separately from the optimized stimulation parameters 420' based on patient testing. In the case of sleep, due to the expected changes in the position of the electrode leads within the patient's spine when the patient lies down, it may be warranted to allow the use of amplitudes lower than those suggested by the optimized parameters 420'. Additionally, the patient may be less bothered by pain while sleeping, and thus lower amplitudes may still be reasonably effective. Having said that, subset 425b may also include values ​​(including amplitudes) that are completely within and constrained by the optimized stimulation parameters 420', similar to Fig.25A and Fig.25B The values ​​shown for subset 425a in .

[0222] Fig.27A and Fig.27B A selection of a sensory mode 504 is shown, along with the resulting subset 425c that may be used for a given patient during that mode. The purpose of this mode is to allow the patient to sense the stimulation provided by their IPG at their discretion. In other words, the stimulation provided to the patient in this mode is super-sensory. The optimized stimulation parameters 420' preferably define a volume of stimulation parameters in which sub-perceptual stimulation is optimized for the patient. However, as previously described, as part of the determination of the optimized sub-threshold stimulation parameters 420', the perception threshold pth is measured and modeled. As such, the perception threshold pth as determined earlier is useful for selecting the amplitude that the patient will feel during this mode, i.e., an amplitude (particularly pulse width) that is higher than pth for other stimulation parameters. Therefore, the sensory mode 504 is an example in which it is beneficial to include a pth value (or pth range) within the optimized stimulation parameters 420'.

[0223] like Fig.27A As shown in , patients are generally more likely to feel stimulation at lower frequencies, and therefore, the selection of a sensory pattern may constrain stimulation in subset 425c to lower frequencies (e.g., 40 to 100 Hz). Control of pulse width may not be a major issue, and therefore, for this frequency range, pulse width may have a moderate range allowed by 420', as also shown in Fig.27A As shown.

[0224] However, because the patient in this mode wants to feel the stimulation, the amplitudes in subset 425c are set to higher values, such as Fig.27BSpecifically, the amplitudes of the relevant frequencies and pulse widths are not only set above the upper limit of the amplitude as determined for the optimized stimulation parameters 420'; they are also set at or above the perception threshold pth. As mentioned previously, the perception threshold pth and the more particularly important range of pth determined for the patient (when taking into account statistical variations) can be included in the optimized stimulation parameters 420' (see Fig.23D ) to produce a useful effect in that mode. Therefore, based on earlier measurements and modeling, subset 425c is defined to set the amplitude within a value or range that should provide extrasensory stimulation. If pth is defined by a range based on statistical variance, the allowed range of amplitudes for the sensory pattern 504 can be set to an upper limit value beyond that range, such as Fig.27B . Thus, while the optimized (sub-perceptual) amplitude (per 420') for the frequency range of interest may range from about 3.7 to 4.5 mA, the amplitudes within subset 425c are set to about 5.8 to 7.2 mA, exceeding the upper limit of the pth range to ensure that the stimulation produced for the patient in question is super-perceptual. In this example, note that subset 425c was determined using optimized stimulation parameters 420', but is only partially constrained by such optimized parameters. Frequency and pulse width are constrained; amplitude is not, because the amplitudes in this subset 425c are set to exceed 420', and more particularly exceed pth.

[0225] Fig.28A and Fig.28B The selection of comfort mode 506 is shown, and the resulting subset 425d of stimulation parameters for that mode. In this mode, the stimulation parameters are set via subset 425d to nominal values ​​within optimized stimulation parameters 420': medium frequencies (such as 200 to 400 Hz), and medium pulse widths for these frequencies (such as 175 to 300 μs as shown in slider interface 550), as shown in FIG. Fig.28A As shown in Fig.28B , the amplitudes within subset 425d may also be medium amplitudes within the optimized stimulation parameters 420' for the frequency and pulse width in question. In this example, the stimulation parameters in subset 425d are completely constrained by the optimized stimulation parameters 420', although as mentioned previously, this is not necessarily the case for every subset.

[0226] Fig.29A and Fig.29B The selection of an exercise mode 508 and a subset of stimulation parameters 425e associated with the mode are shown. In this mode, medium and high frequencies (e.g., 300-600 Hz) may be used, but for these frequencies, the pulse widths are higher than those specified by the optimized stimulation parameters 420', such as Fig.29AThis is because the position of the electrode leads within the patient may vary more as the patient moves, and therefore it may be useful to provide a higher charge injection into the patient, which a higher pulse width can achieve. Fig.29B As shown in , the amplitudes used can span a moderate range of frequencies and pulse widths involved, but higher amplitudes (not shown) beyond 420' can also be used to provide additional charge injection. Subset 425e shows an example in which the frequency and amplitude are constrained by the optimized stimulation parameters 420', while the pulse width is not constrained; thus, subset 425e is only partially constrained by the optimized stimulation parameters 420'. In other examples, subset 425e can be fully constrained within the optimized stimulation parameters 420' determined earlier.

[0227] Fig. 30A and Fig. 30B Selection of an intense mode 510 of stimulation is shown. In this mode, stimulation is more aggressive and the stimulation parameter subset 425f may occur at a higher frequency (e.g., 500 to 1000 Hz). However, the pulse width and amplitude at these frequencies may be moderate for the frequencies involved, such as Fig. 30A and Fig. 30B In this example, the subset of stimulation parameters in subset 425f may be completely constrained by (contained in) optimized stimulation parameters 420'. As in the previous examples, the patient may use interfaces 550 or 560 or other interface elements not shown to adjust the stimulation parameters in subset 425x that correspond to the patient's stimulation mode selection ( Fig.24 ). Less preferably, selection of a stimulation mode may cause the external controller 45 to send a single set of stimulation parameters (F, PW, A) determined using the optimized stimulation parameters 420' (or 420).

[0228] Note that the stimulation parameters in subsets 425x may overlap; some F, PW, and A values ​​in one subset (e.g., 425a) may also exist in another subset (e.g., 425b). In other words, although this may also be the case, it is not strictly necessary that the stimulation parameters in a given subset are unique to the subset or the stimulation mode represented by the subset. In addition, the boundaries of the individual subsets 425x may be adjustable. For example, although not shown, the external controller 45 may have an option to change the boundaries of the individual subsets. Using such an option, a patient or clinician can, for example, change one or more of the stimulation parameters (e.g., frequency) in the subset (e.g., by increasing the frequency within subset 425a from 10 Hz to 100 Hz to 10 Hz to 150 Hz). Adjustment of subset 425x may also be affected in response to certain feedback, such as a patient's pain level that may be input into the external device 45, or detection of patient activity or posture. More complex adjustments can be locked to the patient and only accessible to the clinician, such as by providing such accessibility through a password entered into the external controller 45. Behind such password protection, subset 425x may be adjustable and / or may enable other stimulation modes (e.g., beyond Fig.24 ) are only accessible to clinicians. As previously described, clinicians can also use clinician programmer 50 to make such clinician adjustments.

[0229] Subset 425x may also be automatically updated from time to time. This may be advantageous because as data is collected for more patients, the underlying modeling that led to the generation of optimized stimulation parameters 420' may change or become better informed. It may also be learned later that different stimulation parameters may better produce the desired effect of the stimulation pattern, and thus warrant adjustment of which parameters are included in the subset. Different stimulation patterns that are provided for different reasons or to produce different effects may also become apparent later, and thus such new patterns and their corresponding subsets may be programmed into external controller 45 later and used in conjunction with the external controller 45. Fig.24 The stimulation mode user interface is presented to the patient. Updating of subsets and / or stimulation modes can occur wirelessly by connecting the external controller 45 to a clinician's programmer or to a network such as the Internet. It should be understood that the disclosed stimulation modes and subsets of stimulation parameters 425x corresponding to such modes are merely exemplary and that a different mode or subset may be used.

[0230] Reference again Fig.24, the stimulation mode user interface may include an option 512 to allow the patient or clinician to define a custom mode of stimulation. The custom mode 512 may allow the user to select frequency, pulse width, and amplitude, or define a subset at least partially defined by the optimized stimulation parameters 420'. Selection of this option may provide a user interface that allows the patient to navigate different stimulation parameters within the optimized stimulation parameters 420', such as previously described in Fig.23E and Fig.23F If the patient finds stimulation parameters that appear to work well as a simulation mode through this option, the user interface may allow the stimulation mode to be stored for future use. For example, and with reference to Fig.23E , the patient may have found stimulation parameters within the optimized stimulation parameters 420' that are beneficial when the patient is walking. Such parameters can then be saved by the patient and appropriately labeled, such as Fig.23D This newly saved stimulation pattern can then be presented to the patient as a selectable stimulation pattern ( Fig.24 ). Logic in the external controller 45 may additionally define a subset 425 of stimulation parameters (e.g., 425g) through which the patient may navigate when later selecting the user-defined stimulation mode. Subset 425g may include, for example, stimulation parameters that constrain the patient's selected parameters (e.g., + / - 10% of the patient's selected frequency, pulse width, and amplitude), but are still constrained in whole or in part by selectable stimulation parameters 420'.

[0231] like Fig.24 As shown in FIG. 5 , the stimulation mode user interface may also include an option 514 for automatically selecting and adjusting a stimulation mode for the patient based on various factors that the IPG 10 may detect. Fig.31 514. Preferably, the selection of the automatic mode 514 allows the patient to select 570 which of the stimulation modes he wants to detect and will be automatically used by his IPG 10. In the depicted example, the user has selected the sleep mode 502, the comfort mode 506, and the exercise mode 508. The IPG 10 will attempt to automatically detect when these stimulation modes should be entered, and in this regard the IPG 10 may include a stimulation mode detection algorithm 610. As shown, the algorithm may be programmed into the control circuit 600 of the IPG 10. The control circuit may include a microprocessor, a microcomputer, an FPGA, other digital logic structures, etc., which are capable of executing instructions of an electronic device. Alternatively, the algorithm 610 in the IPG 10 may attempt to detect and adjust stimulation for all stimulation modes (e.g., 500-510) supported by the system without the user selecting 570 the stimulation mode of interest.

[0232] The algorithm 610 can receive different inputs related to detecting stimulation patterns, and therefore receive a subset 425x that should be used for a patient at any given time. For example, the algorithm 610 can receive inputs from various sensors such as an accelerometer 630 indicating the patient's posture and / or activity level. The algorithm 610 can also receive inputs from various other sensors 620. In one example, the sensor electrode 620 can include an electrode Ex of an IPG 10, which can sense various signals related to stimulation pattern determination. For example, and as discussed in USP9,446,243, the signal sensed at the electrode can be used to determine the (complex) impedance between each pair of electrodes, which can be related to various impedance features indicating the patient's posture or activity in the algorithm 610. The signal sensed at the electrode can include those signals generated by the stimulation, such as an induced compound action potential (ECAP). As disclosed in U.S. patent application serial number 16 / 238,151 filed on January 2, 2019, a review of the various features of the detected ECAP can be used to determine the patient's posture or activity. The signals sensed at the electrodes may also include stimulation artifacts resulting from the stimulation, as disclosed in U.S. Provisional Patent Application Serial No. 62 / 860,627 filed on June 12, 2019, which may also indicate the patient's posture or activity. The sensed signals at the electrodes may also be used to determine the patient's heart rate, as disclosed in U.S. Patent Application Serial No. 16 / 282,130 filed on February 21, 2019, which may also be related to the patient's posture or activity.

[0233] The algorithm 610 may receive other information relevant to determining the stimulation mode. For example, the clock 640 may provide time information to the algorithm 610. This may be relevant to determining or confirming whether the patient is participating in activities that occur during certain times of the day. For example, it may be expected that the patient may sleep at night, or exercise in the morning or afternoon. Although not shown, the user interface may allow programming of the time range of the expected activities, such as whether the patient prefers to exercise in the morning or afternoon. The algorithm 610 may also receive input from the battery 14, such as the current state of the battery's voltage Vbat, which may be provided by any number of voltage sensors, such as an analog-to-digital converter (ADC; not shown). For example, this may be useful in deciding when the economy mode 500 or other power-based stimulation mode should be automatically entered (i.e., if Vbat is low).

[0234] In any case, the stimulation pattern detection algorithm 610 can wirelessly receive an indication that the automatic mode 514 has been selected, as well as any selected modes 570 of interest to the patient. The algorithm 610 can then use its various inputs to determine when those modes should be entered, and will therefore enable the use of the subset 425x corresponding to the detected stimulation pattern at the appropriate time. For example, in Fig.31 In the example of FIG. 6 , the algorithm 610 may use the accelerometer 630, the sensor 620, and the clock 640 to determine that a person is still at night, lying on his back or prone, and / or that his heart rate is slow, and therefore determines that the person is currently sleeping. The algorithm 610 may then automatically activate the sleep mode 502 and activate the subset 425b corresponding to the mode ( Figure 26A-26B ) within the stimulation parameters. In addition, the IPG 10 may send a notification of the current stimulation mode determination back to the external controller 45, which may be displayed at 572. This is useful to allow the patient to review that the algorithm 610 has correctly determined the stimulation mode. In addition, notifying the external controller 45 of the current determined mode may allow the external controller 45 to use an appropriate subset 425x of the mode to allow the patient to adjust the stimulation. That is, the external controller 45 may use the determined mode (sleep) to adjust the ( Figure 26A-26B ) is constrained to be a corresponding subset of the pattern (425b).

[0235] If the algorithm 610 uses one or more of its inputs to determine that a person is changing position rapidly, standing upright, and / or that his heart rate is high, it can be determined that the person is currently exercising, which is the stimulation mode of interest selected by the patient. The algorithm 610 can then automatically activate the exercise mode 508 and activate the subset 425e corresponding to that mode ( Figure 26A-26B ) within the stimulation parameters. Again, the IPG 10 may send a notification of the current stimulation mode determination back to the external controller 45 to adjust ( Figure 29A-29B ) to the corresponding subset of that mode (425e). If the algorithm 610 cannot determine that the patient is sleeping or exercising, it can default to the selection of comfort mode 506 and provide stimulation, notifications, and constraint adjustments accordingly (subset 425d, Figure 28A-28B ).

[0236] The external controller 45 may also be useful in determining the relevant stimulation mode to be used during automatic mode selection. Fig.31, but the external controller 45 may include a sensor, such as an accelerometer, for determining patient activity or posture. The external controller 45 may also include a clock, and may wirelessly receive information about its battery voltage from the IPG 10, and wirelessly receive information about signals detected at the electrodes of the IPG from the sensor 620. Therefore, the external controller 45 may also include a stimulation pattern detection algorithm 610' responsive to such inputs. The algorithm 610' may replace the algorithm 610 in the IPG 10, or may supplement the information determined from the algorithm 610 to improve the stimulation pattern determination. In short, and as facilitated by two-way wireless communication between the external controller 45 and the IPG 10, the stimulation pattern detection algorithm may be effectively split between the external controller and the IPG 10 in any desired manner.

[0237] In addition, the external controller 45 can receive relevant information from various other sensors to determine which stimulation mode should be input. For example, the external controller 45 can receive information from an external device 612 (such as a smart watch or smart phone) worn by the patient. Such a smart device 612 contains sensors indicating motion (e.g., accelerometers) and may also include biosensors (heart rate, blood pressure), which can help understand different patient states and therefore different stimulation modes should be used. More generally, other sensors 614 can also provide relevant information to the external controller 45. Such other sensors 614 can include other implantable devices that detect various biological states (glucose, hearing rate, etc.) of IPG patients. Such other sensors 614 can provide other information. For example, since it has been shown that cold or bad weather can affect the stimulation treatment of IPG patients, the sensor 614 may include a weather sensor that provides weather information to the external controller 45. Please note that the sensor 614 may not need to communicate directly with the external controller 45. Information from such a sensor 614 can be sent by a network (e.g., the Internet) and provided to the external controller 45 via various gateway devices (routers, WiFi, Bluetooth antennas, etc.).

[0238] Fig.32Another example of a user interface on a patient external controller 45 is shown that allows the patient to select from different stimulation modes. In this example, different stimulation modes (consistent with the optimized stimulation parameters 420' determined for the patient) are displayed in a two-dimensional representation. In the example shown, the two-dimensional representation includes a graph of pulse width (Y-axis) versus frequency (X-axis), but any two stimulation parameters (amplitude versus frequency or pulse width versus amplitude) may be used. However, please note that these X and Y axes may not be labeled, nor may specific pulse width or frequency values ​​be labeled, if the goal is to provide the patient with a simple user interface that is not hindered by technical information that the patient may not understand.

[0239] Labeled in this two-dimensional representation are the different stimulation modes discussed previously, with boundaries showing the range of the subset 425x for each stimulation mode. Using this representation, the patient can position a cursor 430 to select a particular stimulation mode, and in doing so, select a frequency and pulse width and its corresponding subset 425x. Because the subsets 425x can overlap, selection on a particular frequency and pulse width can select more than one stimulation mode and more than one subset 425x, thereby allowing the patient to browse more than one subset of stimulation parameters. Because the amplitude is not represented in the two-dimensional representation, the amplitude can be automatically adjusted to an appropriate value if a stimulation mode / subset 425x or a particular frequency / pulse width is selected. Alternatively, a separate slider can be included to allow the patient to additionally adjust the amplitude for each of the stimulation modes according to the subset 425x. As explained above, the amplitude can be completely constrained within the optimized stimulation parameters 420' by the selected mode / subset, or can be allowed to range beyond 420' (e.g., Fig.26B , Fig.27B ). In a more complex example, the representation may include a three-dimensional space (F, PW, A) in which the patient may move a cursor 430, similar to Fig.23E Shown in FIG. 4 is a three-dimensional subset 425x with a displayed stimulation pattern.

[0240] Fig.33 Another GUI aspect is shown that allows the patient to adjust stimulation according to the model developed for the patient. In these examples, a suggested stimulation area 650 for the patient is shown, which is overlaid on a user interface element that otherwise allows the user to adjust stimulation. Fig.33 The example in Fig. 22 and Fig.32, but may also be applied to other user interface examples. In these examples, the suggested stimulation areas 650 provide a visual indicator for the patient, where the patient may wish to select (e.g., using cursor 430) stimulation settings that are consistent with the optimized stimulation parameters 420 or 420' or subset 425x. These areas 650 can be determined in different ways. They can be mathematically determined using the optimized stimulation parameters 420 or 420' or subset 425x, such as by determining the center or "center of mass" of these areas. They can also be determined with a particular emphasis on providing the patient with stimulation parameters with appropriate amplitude, intensity, or total charge. This may be particularly useful if the patient's previous selection has been far from such an ideal value. The areas 650 can also be determined during the fitting procedure by determining the area or volume within the optimized stimulation parameters 420 or 420' or subset 425x that the patient likes best.

[0241] In addition, the area 650 can be determined for the patient over time based on the previously selected stimulation parameters. Therefore, the area 650 can be associated with the settings that the patient uses most often. In an improved example, the patient can also provide feedback related to the location of the determined area 650. For example, the external device 45 may include option 652 to allow the patient to provide an indication of their symptoms (e.g., pain) using the rating scale shown. Over time, the external controller can track the pain level entered at 652 and associate it with the selected stimulation parameters, and draw or update the area 650 to the appropriate position of the stimulation adjustment that covers the patient has experienced the best symptom relief. Again, a weighted mathematical analysis of the stimulation parameters relative to their pain level, or a centroid method, can be used.

[0242] It should be noted that the use of the disclosed technology should not necessarily be limited to the specific frequencies tested.Other data suggest the applicability of the disclosed technology to provide pain relief without paresthesia at frequencies as low as 2 Hz.

[0243] To summarize this point, modeling and patient fitting allow for the determination of optimized and preferably sub-perceptual stimulation parameters (in the form of range 420, volume 420', or subset 425) for a given patient. However, once such optimized stimulation parameters are found, it may be desirable to vary such parameters over time when stimulation is applied to the patient. This is because providing the same, unvarying stimulation to neural tissue - even if ideal - can lead to habituation of such tissue, such that stimulation may not be as effective as before.

[0244] Therefore, once the optimized stimulation parameters are determined, it may be useful to automatically vary the stimulation applied by the IPG 10 or ETS 40 within these parameters over time. Fig.34AIn the first example shown in FIG. 4 , it is assumed that a subset 425 (particularly 425e, see FIG. 4 ) of the volume of the optimized stimulation parameters 420 ′ is Fig.27A and Fig.27B ) has been determined for use with the patient. However, although not shown, a mean charge per second (MSC) model (e.g., 380, 381) may also be used to select optimized stimulation parameters, as previously described with respect to Figures 17A-17F described.

[0245] To prevent habituation, the simulation applied to the patient varies over time within subset 425, as represented by adjustment 700. Adjustment 700 may vary any simulation parameter within subset 425, including frequency, pulse width, and amplitude, and any one or more of these parameters may be varied at any given time. Fig.34A In the example shown at the top of FIG. , the frequency and pulse width are changed at different times (t1, t2, tec.) within the subset 425, while the amplitude remains constant. Fig.34A In the example shown at the bottom of , the frequency and amplitude are varied within subset 425 at different times, while the pulse width remains constant.

[0246] Fig.34B An example of how an adjustment 700 may be formulated and how a program for an IPG including instructions may be generated is shown. It is shown in a graphical user interface (GUI) 710, which is described in more detail in U.S. Provisional Patent Application Serial No. 62 / 897,060 filed on September 6, 2019, and it is assumed that the reader is familiar with it. The GUI 710 can be operated on a clinician programmer 50 or an external controller 45, and allows a user to specify pulses in a manner that achieves the desired changes in the adjustment 700. The GUI 710 includes a plurality of blocks 711, in which a user can specify a time-sequential pulse sequence. The first block (1) specifies a pulse to be formed at t1, at electrodes (E1 and E2) as specified by Guide Program A, and whose frequency (200 Hz), pulse width (225 μs) and amplitude (4 mA) are within the volume of the subset 25. The frequency, pulse width and amplitude can be specified by Pulse Program I, as explained in further detail in the '060 application. During time period t1 (and all other time periods in this example), ten of these pulses will be formed, although the number of pulses may vary and may be set in GUI 170. The second block (2) specifies that 10 pulses be formed at t2, at the same electrodes, with a frequency (200 Hz), pulse width (325 μs), and amplitude (4 mA) as specified by pulse program J. Thus, only the pulse width changes from time period t1 to t2. Fig.34B The pulse width and frequency in the example are changed to affect Fig.34AThe modulation 700 shown at the top. In this example, the pulse width and frequency are modulated in a serpentine manner between different time periods, but this is only an example and the modulation 700 within the optimized stimulation parameters may be performed in a different manner or even randomly. The amplitude determined within the optimized stimulation parameters may also be varied, as will be described later with respect to Fig.34C Explained.

[0247] Even though certain stimulation parameters are changed via adjustment 700, they are still within the previously determined optimized stimulation parameters, and in particular within subset 425. Note, however, that adjustment 700 need not occur within subset 425. More generally, adjustments to prevent habituation may occur within optimized stimulation parameters 420 or 420' as determined earlier for the patient.

[0248] The GUI 710 may include an option 712 to allow the user to import previously determined optimized stimulation parameters (a model of the patient) into the GUI 170, which may reside in the clinician programmer 50 or the external controller 45. Once imported, another option 714 may be used to automatically form an adjustment 700 within those optimized stimulation parameters. Selecting option 714 may cause the GUI 710 to automatically fill in blocks 711 in a manner that changes one or more stimulation parameters as needed to produce the desired adjustment 700. Although not shown, option 714 may allow the user to select which of one or more stimulation parameters (e.g., frequency, amplitude, pulse width) should be varied within the optimized stimulation parameters, and may further allow the user to determine an order or pattern within the optimized stimulation parameters in which the stimulation parameters may be varied. Option 714 may also allow the user to select to randomly change the stimulation parameters during the adjustment 700. If necessary, the user may adjust the stimulation parameters in the independent blocks 711 after they have been automatically created to best affect a particular adjustment 700.

[0249] Fig.34C 700 within the optimized stimulation parameters can be affected. As shown, the pulse width, amplitude, and frequency of the stimulation pulses can be adjusted between maximum and minimum values ​​(e.g., PW(min), PW(max)) within the optimized stimulation parameters. Note that these maximum and minimum values ​​may not be constant, but may be affected by other stimulation parameter values. For example, and as Fig.34A As shown, PW(max) and PW(min) may include 325 and 225 μs when the frequency is 200 Hz, but may include 225 and 150 μs when the frequency is 400 Hz. Fig.34CAn example is shown at the bottom of FIG. 700 where the adjustment 700 varies the amplitude, pulse width, and frequency within a volume of optimized stimulation parameters 420' or within a subset 425 of such parameters. Again, this is shown simply as a cube with maximum and minimum values ​​for the parameters, but the resulting volume may actually have a more random shape. In another modification not shown, a mean charge per second (MSC) model (e.g., 380, 381) may be used to select different optimized stimulation parameters, as previously described with respect to FIG. Figures 17A-17F described, and applied at different times.

[0250] In addition to helping prevent tissue habituation, adjustment 700 is expected to be beneficial because stimulation is adjusted over time within a range or volume of optimized stimulation parameters, thereby making it more likely that optimized stimulation parameters (or parameter combinations) for the patient within this range will be provided at least occasionally during adjustment 700. This can be important because the leads may move within the patient, such as with activity, which may cause the optimized stimulation parameters to change from time to time. Thus, adjusting the stimulation parameters helps ensure that the optimal parameters within the range or volume will be applied during at least some time periods of adjustment 700. In addition, when adjusting stimulation parameters, it may not be necessary to spend time fine-tuning the stimulation to determine a single, unchanging set of optimized stimulation parameters for the patient.

[0251] Fig.35 Other adjustments 700 that can be used are shown that affect the electric field formed in the patient's tissue and can also be used to prevent habituation of the tissue. A specific pole configuration 730 that has been selected for the patient is shown. In the depicted example, the pole configuration 730 includes a virtual bipole with a virtual anode pole (+) and a virtual cathode pole (-). By way of review, virtual poles are further discussed in U.S. Patent Application Publication 2019 / 0175915 and previously referenced. Figure 7B As previously discussed, the locations of the anode pole and cathode pole do not necessarily correspond to the locations of the physical electrodes 16 in the electrode array. Also as previously discussed, the pole configuration 730 may have a different number of poles and may include three poles or other configurations, although for simplicity the pole configuration 730 is shown in FIG. Fig.35 The double poles are depicted in .

[0252] Fig.35 The top of shows different ways in which the pole configuration 730 can be moved in the electrode array 17. Fig.35The upper left corner of shows how the dual pole 730 can be moved to different xy positions in the electrode array 17 while still maintaining the relative positions of the poles with respect to each other. The upper right corner shows how the focal length (i.e., the distance d between the poles) is changed. Both of these means of adjusting the pole positions can be achieved by changing the activation of the electrodes 16 in the array, such as by providing specific polarities and current percentages to selected electrodes, as discussed previously.

[0253] Fig.35 The bottom of the figure shows how the adjustment of the position of the poles can be combined with the adjustment consistent with the previously determined optimized stimulation parameters. The left figure shows how one of the stimulation parameters (in this case, the pulse width) can be changed while also changing the xy position of the pole configuration 730. This adjustment 700 can change the pulse width between maximum and minimum values ​​(PW(max) and PW(min)) over time, as determined for the optimized stimulation parameters 420' or 425. This adjustment can also change the xy position of the pole configuration 730 over time. Preferably, the (x, y) position of the pole configuration 730 is previously determined (using a sweet spot search, as explained above), but the maximum and minimum values ​​are changed from that position during the adjustment 700. For example, the position (x, y)(min) can include a position where both x and y are 1 mm smaller, and the position (x, y)(max) can include a position where both x and y are 1 mm larger. In other words, if the optimal position (x, y) is at (5 mm, 6 mm) in the electrode array, then (x, y)(min) will include a position at (4 mm, 5 mm), and (x, y)(max) will include a position at (6 mm, 7 mm). Thus, adjustment 700 can move the position of pole configuration 730 anywhere within the two-dimensional region bounded by these maximum and minimum positions. Pulse width is also varied during adjustment 700, and other stimulation parameters (frequency, amplitude) may also be varied, again between maximum and minimum values ​​determined using optimized stimulation parameters 420' or 425. The stimulation parameters and poles may be adjusted over time according to the pattern shown, or randomly adjusted between maximum and minimum values.

[0254] The middle diagram shows how one of the stimulation parameters (again, pulse width) can be changed while also changing the focal length d of the pole configuration 730. Preferably, the focal length d is previously determined (e.g., during the sweet spot search), but the maximum and minimum values ​​are changed from that distance during adjustment 700. For example, if d equals 10 mm, then d(max) may be 12 mm, while d(min) is 8 mm. Thus, adjustment 700 can move the focus of the pole configuration 730 anywhere within these maximum and minimum distances. The pulse width (and / or at least one other parameter, such as amplitude or frequency) is also changed during adjustment 700 in concert with the previously determined optimized stimulation parameters 420' or 425. Likewise, the stimulation parameters and focal length can be adjusted over time according to a pattern, or randomly adjusted between maximum and minimum values.

[0255] The right figure shows that both the xy position of the pole configuration 730 and the focal length d of the pole configuration can be varied with at least one other stimulation parameter (e.g., pulse width) during adjustment 700. In all of these examples, adjustment 700 including slight adjustments to the position of the poles in the pole configuration 730 is expected to be useful in preventing tissue habituation.

[0256] Adjustment 700 may prioritize adjustment of certain parameters over other parameters, and such prioritization may be based on the patient's state or symptoms. For example, if the patient is noted to be particularly sensitive to the location of stimulation, it may be desirable for adjustment 700 to prioritize changes in the location of the poles, either by changing the xy location of the pole configuration 730 in the electrode array and / or by changing the focal length d. In contrast, if the patient is particularly sensitive to the amount of stimulation (e.g., the neural dose received), it may be desirable for adjustment 700 to prioritize changes in one or more of the stimulation parameters (pulse width, frequency, amplitude). Patient sensitivity useful in prioritizing adjustment 700 may be determined using subjective or objective measurements, such as by receiving patient feedback, or by making measurements that indicate stimulation efficacy (e.g., by measuring ECAP as described above).

[0257] Fig.36Another example of adjustment 700 of stimulation parameters within optimized stimulation parameters 420' is shown. In this example, pulse width and frequency are adjusted between different time periods. Unlike the previous example, the stimulation duration at each time period is longer, on the order of hours. In addition, the pole configuration is changed at different time periods to achieve different beneficial effects. For example, at time periods t1 and t2, a bipole 740 that forms a relatively small field in the tissue is used. This may be useful because, as described previously, such a sub-perceptual bipole 740 can provide rapid relief and short-term washout, and especially at lower frequencies present during these time periods. However, a smaller bipole such as 740 may be sensitive: it only produces a small field in the tissue, and therefore if the lead in the electrode array 17 migrates within the tissue, the bipole 740 may migrate away from the patient's pain site and become less effective. Therefore, in subsequent time periods (e.g., t3-t5), the pole configuration is changed to a larger bipole 745, which provides a larger field in the tissue. The larger field makes it less likely that lead migration will result in effective stimulation away from the patient's pain site, and thus more easily recruits such pain site. The higher frequencies used with the larger bi-pole 745 can additionally more easily recruit the patient's tissue and be less susceptible to lead migration. Therefore, during adjustment 700, the bi-pole is increased in size to promote a larger field in the tissue, and the frequency of the pulses is increased until a constant (but still optimized) stimulation therapy is provided at time t5.

[0258] like Fig.37A and Fig.37B As shown in , when stimulation is provided as a bolus, adjustments 700 within previously determined optimized stimulation parameters may also be used. Providing stimulation boluses is described in more detail in U.S. Provisional Patent Application Serial No. 62 / 916,958, entitled “Prescribed Neuromodulation Dose Delivery,” filed concurrently on October 18, 2019, the entirety of which is incorporated herein by reference. A bolus includes stimulation provided within a set time unit, such as ten minutes, thirty minutes, one hour, two hours, or any other effective duration, with time intervals without stimulation between bolus administrations. It has been observed that some patients respond well to “bolus mode” treatment. When a patient feels pain coming on (shown as lightning), they may initiate a stimulation bolus (shown as Fig.37A capsules in the Fig.37AThree days are shown during which nine stimulation boluses have been administered. As discussed in the '958 Provisional Application, administration of the boluses may also be automated. Providing simulations during the boluses may be beneficial because some patients experience prolonged pain relief for up to several hours or longer after receiving a bolus stimulation, i.e., during the interval between boluses during which no stimulation occurs. In addition, providing stimulation boluses saves energy in the IPG because the simulations are not continuous, and also helps prevent tissue overstimulation and habituation.

[0259] like Fig.37B As shown in FIG. 7 , the stimulation parameters used during each stimulation bolus can be adjusted. This adjustment 700 can be similar to Fig.36 , but can occur on a shorter time scale. For example, the stimulation bolus shown is 100 minutes in length and consists of five different time periods t1-t5, each lasting 20 minutes. As previously described, one or more stimulation parameters (e.g., pulse width and frequency) are adjusted during different time periods within the optimized stimulation parameters 420' determined earlier. Preferably, the simulation parameters initially used (e.g., during t1) are designed to bring about rapid symptom relief. As previously described, the position, size, or focal length of the pole configuration can also be changed during different time periods including the adjustment 700.

[0260] Figure 38-41B A fitting algorithm 740 is shown that can be used to determine the best 750 of the optimized stimulation parameters 420 or 420' for use with a given patient. In this example, a range or volume of preferred optimized stimulation parameters 420 or 420' is determined for the patient, which preferably prescribes sub-perceptual stimulation for the patient, as previously described. Although not shown, a mean charge per second (MSC) model (e.g., 380, 381) can be used to select the optimized stimulation parameters, as previously described with respect to Figures 17A-17F described.

[0261] The fitting algorithm 740 then uses the fitting information 760 to determine the best one or more of the optimized stimulation parameters for use with the patient. The best optimized stimulation parameters 750 may include a single set of stimulation parameters—e.g., a single frequency, pulse width, and amplitude value 750a—or a subset 750b of a set of parameters similar to the previously described subset 425. In short, by using the additional information included in the fitting information 760, the fitting algorithm 740 may determine the most logical one or more stimulation parameters within the optimized stimulation parameters 420 or 420' for the patient, and may set sub-perceptual stimulation in the patient's IPG accordingly.

[0262] The fitting information 760 is preferably acquired during a fitting procedure after implantation, which typically occurs in a clinical setting. Therefore, the fitting algorithm 740 is preferably implemented as a clinician programmer software 66 (executable on the clinician programmer 50). Figure 4 ). However, the fitting algorithm 740 may also be used with any device or system capable of communicating with the patient's IPG, including the patient's external controller 45 ( Figure 4 ). Aspects of the fitting algorithm 740 may be presented as part of a clinician programmer GUI. The fitting algorithm 740 may also include instructions in a computer-readable medium, as described elsewhere. The fitting algorithm 740 may be executed in conjunction with other operations that logically occur during the fitting procedure and the fitting information 760 is received. For example, when the fitting information 760 is received, the tests that occur during the algorithm 400 - such as different pulse widths (404, 405) used during the determination of the optimized stimulation parameters 420 or 420' may be performed. Fig.21A )—can occur at the same time and during the same procedure.

[0263] The fitting information 760 may include various data indicating the patient, his symptoms, and the stimulation provided during the fitting procedure. For example, the fitting information 760 may include pain information 770 characterizing the pain of the patient without stimulation. The fitting information 760 may also include mapping information 780 indicating the effectiveness of the stimulation used during the fitting procedure. The fitting information 760 may also include spatial field information 790 indicating the stimulation used during the fitting procedure, and the electric field it produced in the patient's tissue. The fitting information 760 may also include phenotypic information 800, such as the patient's age, gender, and other patient-specific details.

[0264] The fitting algorithm 740 also receives or includes training data 810. In essence, the training data 810 is used to associate the fitting information 760 with the best results, as will be described in further detail below. For example, the training data 810 may suggest that the fitting information 760 for a particular patient warrants the use of a lower frequency treatment (e.g., 10-400 Hz) for the patient, and the fitting algorithm 740 will therefore select the lower frequency stimulation when selecting the best optimized stimulation parameters 750 from the optimized stimulation parameters 420 or 420' for the patient. Alternatively, the training data 810 may suggest that the fitting information 760 for a particular patient warrants the use of a higher frequency treatment (e.g., 400-1000 Hz) for the patient, and the fitting algorithm 740 will therefore select the higher frequency stimulation when selecting the best optimized stimulation parameters 750 for the patient. The training data 810 may arrive over time and may be derived from the treatment of previous patients, and in this regard, the training data 810 will improve over time as patients are further treated and data is received from more patients. In this regard, the information including the training data 810 may be received at the fitting algorithm 740 from a source other than an external device, such as from a server that may receive data from different patients to develop or update the training data 810 over time. The training data 810 may also include or contain historical data taken from the current patient. In one example, the training data 810 may be obtained using machine learning techniques and may include weights or coefficients to be applied to various segments of the fitting information 760, as further explained below.

[0265] Figure 39A-Figure 39C A GUI of an external system (e.g., a clinician programmer) is shown that can be used to receive various pieces of fitting information 760 during a fitting procedure, wherein Fig.39A shows the receipt of pain information 770, Fig.39B The receipt of mapping information 780 is shown, and Fig.39C The receipt of spatial field information 790 and patient phenotype information 800 is shown. It is not required that the fitting algorithm 740 receive all of the pieces of fitting information shown in these figures, and the algorithm 740 may receive additional pieces of information not shown that may be relevant to predicting the best optimized stimulation parameters 750. In short, Figure 39A-Figure 39C Only examples of potentially relevant fit information 760 are provided. Furthermore, while it is sensible to divide the illustrated relevant fit information 760 into categories of pain information 770, mapping information 780, spatial field information 790, and patient phenotype information 800, the fit information 760 may be subdivided into more or fewer categories. Alternatively, the fit information 760 may not be subdivided into categories at all, but may instead include one, more, or all of the information segments within these categories.

[0266] First reference Fig.39A , pain information 770 is received at the GUI, which, as previously mentioned, includes information segments characterizing the pain of the patient in the absence of stimulation. In a preferred example, pain information 770 is provided for an independent body region Xx. In this regard, the GUI may include a graphic or image 771 showing different body regions where pain may occur. For example, at Fig.39A , body region X1 represents the upper portion of the lower back, and body region X2 represents the lower portion of the lower back, wherein both regions X1 and X2 appear on the right side of the body. Body portion X3 represents the right gluteus maximus, and region X4 represents the upper portion of the right thigh. Other body regions are not labeled in graphic 771 and may also appear on the left side of the body.

[0267] For each body region Xx, multiple different pain measurements are recorded and can be entered into the GUI by the patient or clinician. For example, and considering body region X1, it can be recorded whether pain is present in that region (e.g., no (0), yes (1)), the intensity of pain in that region (e.g., 3 out of 10), how the patient perceives the pain in that region (e.g., as burning (1), numbness (2), sharp (3), etc.). The type of pain can also be classified; for example, 1 may represent neuropathic pain that can be treated well by SCS, while 0 may represent pain originating from other mechanisms (bruises, arthritis, etc.) that may not be treated well by SCS. Such pain information 770 can be entered into the GUI for each body region as shown, resulting in a pain matrix P, which can also be viewed as multiple pain vectors, each of which contains information about the patient's pain in a different body region Xx.

[0268] Fig.39B The receiving of mapping information 780 at the GUI, as previously mentioned, indicates the effectiveness of the stimulus used during the fitting procedure. (The details of the stimulus provided during the fitting procedure are as follows Fig.39C790 in the mapping information). The mapping information 780 can again be specified by body region Xx, and the GUI can again provide a graphic or image 771 showing different body regions where the effects of stimulation can be felt. For example, and considering body region X1, it can be recorded whether stimulation is felt in that region (e.g., no (0), yes (1)), the intensity of stimulation perceived in that region (e.g., 7 out of 10), and the degree to which the patient feels that the stimulation "masks" their pain (e.g., 60%). In addition, the mapping information 760 can include a pain intensity rating that is similar to the pain intensity previously provided in the pain information 770, but is affected by the stimulation; if the stimulation treatment is effective, then the pain intensity in the mapping information 780 can be expected to improve (or at least not worsen) when compared to the pain intensity received during the pain information 770 when the stimulation was not present. The mapping information 780 can also include a characterization of the stimulation sensations perceived by the patient. For example, a patient may report that the stimulation feels like a constant tingling (1), a vibration (2), a massage (3), light pressure, pulsation, a diffuse field, etc.

[0269] Other mapping information 760 can quantify the intensity of the stimulation as perceived by the patient. For example, a sensory abnormality threshold can be determined. As previously discussed, this threshold (also useful during algorithm 400) can include, for example, the lowest amplitude of the stimulation that the patient can perceive. Similarly, mapping information 780 can also include a discomfort threshold, which can include, for example, the maximum amplitude of the stimulation that the patient can tolerate. Other objective metrics (such as various ECAP features recorded in response to stimulation) can also be included in the mapping information 780. Mapping information 780 can produce a mapping matrix M, which can also be viewed as a plurality of mapping vectors, each of which contains information characterizing the effectiveness of stimulation in different body regions Xx.

[0270] Fig.39C Receiving spatial field information 790 and patient phenotype information 800 at the GUI is shown. The spatial field information 790 includes information indicating the stimulation used during the fitting procedure, such as the shape, size, and location of the electric field created by such stimulation in the patient's tissue, and may also include information indicating the physiological location at which the stimulation is applied, as discussed further below. In this regard, it is noted that different types of stimulation may be tried for the patient during the fitting procedure, such as using Figure 5 The GUI aspects shown in .

[0271] The spatial field information 790 may include the type of pulse used during the fitting. For example, the GUI may receive an indication of using a monophasic pulse followed by passive charge recovery (0), a biphasic pulse for active charge recovery (1), a biphasic pulse with additional passive charge recovery (2), etc. Such pulse types have been previously described, and other pulse types may also be used and received at the GUI. The GUI may also receive information about the pole configuration used to provide the stimulation, including the number and polarity of the poles in the configuration, such as whether a biphasic pulse was used (0; e.g., Figure 6 ), three-pole (1; see U.S. Patent Application Publication 2019 / 0175915), extended two-pole (2; e.g., Fig.7D ). The spatial field information 790 may also include information about the size of the pole configuration and the electric field it generates in the tissue. For example, the focal length between the poles and / or the area defined by the poles (or the estimated area of ​​the electric field generated in the tissue) may be received.

[0272] Some segments of the spatial field information 790 may be associated with physiological coordinates, which the fitting algorithm 740 may determine in conjunction with the use of other techniques. Physiological coordinates describe physiological positions in a common manner between patients and with reference to common physiological structures. For example, in an SCS application, the coordinate (0, 0, 0) may correspond to the center of the T10 vertebra, while (20, 0, 0) corresponds to the center of the T9 vertebra, and (-20, 0, 0) corresponds to the center of the T11 vertebra. In this regard, the physiological coordinates may not necessarily specify actual dimensions; for example, the actual distance between T10 and T9 in a larger patient may be greater than the same distance in a smaller patient. Nevertheless, the physiological coordinates generally describe general anatomical locations. In SCS applications, the position of the electrode array 17, and therefore the physiological coordinates of the electrodes 16, are typically known relative to known physical structures, such as by using fluoroscopic imaging techniques that show the position of the patient array 17 / electrode 16 relative to such structures. Although not shown (e.g., in Figure 5 ), but such physiological structures (e.g., different known vertebrae) can be superimposed on the image of the electrode array. Depending on how they are calculated, physiological coordinates can be two-dimensional (x, y), but also three-dimensional (x, y, z), and as Fig.39C As shown in .

[0273] Knowing the location of anatomical structures within the patient, the physiological coordinates of the electrodes 16 relative to such structures, and the active electrodes forming the pole configuration in the array, the fitting algorithm 740 can determine the physiological coordinates of various spatial field parameters. For example, knowing the current at each anode pole and each cathode pole allows the fitting algorithm 740 to determine the physiological coordinates corresponding to the location of those poles, which, as previously mentioned, may not correspond to the physical location of the electrodes 16. Note that, as previously described, knowing the location of these poles can also allow the calculation of focal length and field area.

[0274] The physiological coordinates of the anode and cathode poles also allow for the determination of further physiological coordinates that are generally indicative of the physiological location of the electric field generated within the patient. For example, stimulation will result in the formation of various voltages V in the patient's tissue that can be estimated in three dimensions, particularly if the resistance of the tissue is known or can be measured. This in turn allows for the determination of the three-dimensional electric field E in the tissue as a first-order spatial derivative, i.e., E=dV / dx, and a second-order spatial derivative, d 2 V / dx 2 . Physiological coordinates indicating the location of any of these derivatives are useful for consideration by the fitting algorithm 740. As described in U.S. Provisional Patent Application Serial No. 62 / 849,642 filed on May 17, 2019, while the fibers in the dorsal column run parallel to the long axis x of the spinal cord (i.e., in a rostrocaudal direction), the fibers in the dorsal horn can be oriented in many directions, including perpendicular to the long axis of the spinal cord. Dorsal horn fibers and dorsal column fibers respond differently to electrical stimulation. The intensity of stimulation (i.e., depolarization or hyperpolarization) of the dorsal column fibers is determined by a so-called "activation function" d along the longitudinal axis (x) of the spine. 2 V / dx 2 The dorsal horn "activation function" is better described by dV / dx (also called the electric field, E), because dorsal horn fibers that propagate past the stimulating electrode are more likely to be activated along the axon. This is partly because the large myelinated axons in the dorsal column fibers are primarily aligned longitudinally along the spine. On the other hand, the likelihood of generating an action potential in dorsal horn fibers and neurons is better described by dV / dx (also called the electric field, E), because dorsal horn fibers and neurons that are generally confined to directly below the electrode may be more likely to respond in dendrites and terminals. Therefore, the dorsal horn "activation function" is not proportional to the second-order derivative, but rather to the first-order derivative of the voltage along the fiber axis.

[0275] The physiological coordinates of these activation functions may include spatial field information 790 calculated and used by the fitting algorithm 740. In particular, and as Fig.39C As shown in the figure, the maximum values ​​of these activation functions (maxdV / dt, maxd 2 V / dx 2) can be determined at physiological coordinates, such as the maximum voltage in the tissue (max V). The activation volume can also be determined at physiological coordinates that indicate the volume of neural tissue that is recruited. See, for example, USP 8,606,360 and 9,792,412 (discussing calculation of the activation volume). The physiological coordinates of the activation volume can include the center point of the volume, such as the center of mass, or any other coordinates that tend to show the physiological location of the activation volume in the patient.

[0276] Providing physiological coordinate information for various relevant field parameters may be important for consideration by the fitting algorithm 740. As just described, such physiological coordinates generally indicate the physiological location at which stimulation is provided in a given patient, and therefore generally indicate the physiological neural location of pain in the patient (see, e.g., 298, Fig. 7A ). Knowing this physiological position of the patient being fitted may be of interest because it may allow the fitting algorithm 740 to determine the best 750 of the optimized stimulation parameters 420 or 420'. For example, the training data 810 may reflect specific field parameters (e.g., maximum d 2 V / dx 2 ) is located at or near specific physiological coordinates (e.g., x7, y7, z7, corresponding to a specific neural structure), the best optimized stimulation parameters 750 of a higher frequency may be warranted. In contrast, the location of the parameters at different physiological coordinates (x11, y11, z11, corresponding to different neural structures) may suggest the use of the best optimized stimulation parameters 750 of a lower frequency. This may be reflected in the training data 810. That is, the training data 810 will reflect from the history of past patients that patients with field parameters close to (x7, y7, z7) respond better when higher frequency stimulation is used, while patients with field parameters close to (x11, y11, z11) respond better when lower frequency stimulation is used.

[0277] Patient phenotypic information 800 includes information about the patient, such as their gender, age, type or indication of the patient's disease, duration of their disease, duration since the patient received their implant. Information about postures and / or activities (referred to as postures) in which the patient's symptoms are particularly problematic (e.g., when sitting (1), when standing (2), etc.) may also be included as patient phenotypic information 800. Although Fig.39C Not shown in , but the GUI may include an option to allow input of all problematic postures, since there may be more than one. Together, the phenotype information 800 may produce a vector Y.

[0278] As previously discussed, different patient postures or activities (postures for short) may also affect the best stimulation for a given patient and thus the selection of the best 750 of the optimized stimulation parameters 420 or 420'. Fig.40 It is shown that fitting information 760 can be received based on posture. For example, fitting information 760 can be received when the patient is sitting (e.g., pain matrix P1, mapping matrix M1, spatial field vector F1), standing (P2, M2, F2), supine (P3, M3, F3), etc., because the fitting information 760 may be different for each of these postures. For example, the patient may experience pain in different body areas in different postures, or may perceive pain in different ways, resulting in a pain matrix Px with different information. Similarly, the effectiveness of stimulation may vary in different postures, resulting in a mapping matrix Mx with different information. In addition, the stimulation used when in different postures may be different, as reflected by different spatial field vectors Fx. (In contrast, the information within the patient phenotype vector Y is agnostic to the patient's posture, as shown in FIG. 1 .) Fig.40 shown).

[0279] Such fitting information 760 - e.g., pain matrix P, mapping matrix M, spatial field vector F, and / or phenotype vector P, or separate pieces of information within each - is useful for fitting algorithm 740 to receive and consider, because such information may suggest optimal stimulation parameters for a given patient, and in particular, optimal stimulation parameters for the optimized stimulation parameters 420 or 420' that have been determined for the patient. Experience will teach which pieces of fitting information 760 will comprise the best predictors of optimal optimized stimulation parameters 750, and such experience may be reflected in the training data 810 ( Fig.38 )middle.

[0280] For example, the percentage of pain coverage - mapping matrix M( Fig.39B )—should be well correlated with the frequency or nerve dose of the best optimized stimulation parameters 750. If the stimulation covers the patient's pain well (a high percentage), meaning that the stimulation recruits the patient's pain well, then stimulation at a lower nerve dose or frequency may be appropriate, and thus the fitting algorithm 740 can select one or more (e.g., a subset) of the best optimized stimulation parameters 750 that have lower frequencies within the optimized stimulation parameters 420 or 420 determined for the patient. In contrast, if the stimulation does not cover the patient's pain well (a low percentage), then stimulation at a higher nerve dose or frequency within the optimized stimulation parameters 420 or 420' can be selected as the best optimized stimulation parameters 750 for the patient. As such, when determining the best optimized stimulation parameters 750, the training data 810 can attribute a high correlation (e.g., assign a high weight) to the percentage of pain coverage, or more generally to the mapping matrix M.

[0281] Fig.41A The flowchart shows how the fitting algorithm 740 uses the fitting information 760 to determine the best optimized stimulation parameters 750 for the patient. Fig.41A Only a simple example of how the fitting algorithm 740 may be performed and how the training data 810 may be applied to the fitting information 760 is provided. As mentioned previously, although those skilled in the art appreciate that the training data 810 may be obtained using machine learning techniques or other statistical techniques that are complex in nature.

[0282] exist Fig.41A , the training data 810 is applied to the fitting information 760 in the form of weights wx, which essentially assign a relevance to each segment of the fitting information. The weights are shown as being applied to the pain matrix P, the mapping matrix M, the spatial field vector F, and the phenotype vector Y. In the example shown, the weights are applied to each of the matrices or vectors, and in this regard, it may be useful to process each matrix or vector so that each of them is represented by a single number. Although not shown, it should be understood that weights can be applied to each of the independent segments of information including the various matrices or vectors, and therefore the fitting information 760 does not have to include matrices or vectors of information. In addition, the fitting algorithm 740 does not have to strictly consider all of the pain information (P), mapping information (M), spatial field information (F), and patient phenotype information (Y), because in actual implementations, these categories or some of the information therein may not be shown to be statistically relevant to selecting the best optimized stimulation parameters 750.

[0283] Preferably, applying the training data 810 to the fitting information 760 results in the determination of a fitting variable J. Although not shown, the fitting variable J can have a variance or error associated with it, which can result from the statistical manner in which the training data 810 operates. In this regard, the fitting variable J can include a single variable or a range of variables. The fitting variable J can be used by the fitting algorithm 740 to select one or more optimal 750 of the optimized stimulation parameters 420 or 420'. In one example, the fitting variable J can be related to neural dose. For example, high values ​​of J may correspond to high values ​​of frequency because the optimized stimulation parameters 420 or 420' tend to include higher neural doses at higher frequencies. Fig.41BIn the bottom graph of , a relatively high value of J results in a single point selection of the best optimized stimulation parameters 750a, such as a pulse with a frequency of 600 Hz, a pulse width of approximately 150 microseconds, and an amplitude of approximately 4 mA. Alternatively, fitting variable J (which may include a range of values, or may be associated with an error term) may result in selection of the following best optimized stimulation parameters 750, which include a subset of parameters 750b, such as a range of frequencies (e.g., 400 to 800 Hz) and pulse widths and amplitudes associated with those within 420 or 420'. In contrast, Fig.41A The top portion of illustrative embodiment shows how a lower J value results in the selection of the best optimized stimulation parameter 750 from among the optimized stimulation parameters 420 or 420' having a lower frequency and thus a lower nerve dose.

[0284] When selecting the best 750 of the optimized stimulation parameters 420 or 420', the fitting algorithm 740 may more qualitatively process the fitting variable J. In this regard, and as Fig.41C As shown in , the fitting algorithm 740 can classify the fitting variable J into categories rather than determining J as an absolute value. For example, J can be classified as "1", indicating that stimulation parameters with lower nerve doses should be selected from the optimized stimulation parameters 420 or 420'. Such lower dose parameters just explained can include parameters at lower frequencies, and thus the fitting algorithm 740 can select a subset 750x of stimulation parameters for the patient, including optimized stimulation parameters 420 or 420' at lower frequencies (e.g., 100-200 Hz) and pulse widths and amplitudes consistent with those within 420 or 420'. Similarly, J can be classified as "2" or "3", indicating that a medium or higher nerve dose is used, respectively, which can result in the selection of a suitable optimal stimulation parameter subset 750y (e.g., parameters within 420 or 420' with a medium range frequency of 200-400 Hz) or 750z (e.g., parameters within 420 or 420' with a higher range frequency of 400-1000 Hz). If necessary, the system (e.g., the patient's IPG or an associated external programming device) can limit adjustments to these determined subsets 750x-z, similar to what was explained previously. As previously described, a single set of parameters can also be selected by the fitting algorithm 740 as opposed to a subset of parameters.

[0285] Alternatively, to the extent that the fit information 760 is determined as a function of the patient's posture x, as described earlier in Fig.40As described in , the fitting algorithm 740 can determine fitting variables Jx corresponding to each patient posture x (e.g., J1 sitting, J2 standing, etc.), where each fitting variable is determined using fitting information specific to that posture (or at least information that is not specific to any posture, such as patient phenotype information 800). For example, J1 = w1*P1 + w2*M1 + w3*F1 + w4*Y, and J2 = w5*P2 + w6*M2 + w7*F2 + w4*Y, and so on. Each of these posture-specific fitting variables Jx can be used to determine the best optimized stimulation parameters 750 for a different patient posture. This can be useful because it allows the best optimized stimulation parameters 750 to be adjusted when the patient changes posture. This is similar to what was described above regarding selecting a different subset 425 for a patient depending on the currently detected patient posture: when a new patient posture is detected, the new, best optimized stimulation parameters 750 associated with the detected posture can be applied.

[0286] Preferably, the fitting algorithm 740 uses the previously determined optimized stimulation parameters 420 or 420' for the patient when selecting the best optimized stimulation parameters 750 for the patient. However, this is not strictly necessary, and Fig.42 An alternative fitting algorithm 740' is shown. As previously described, the training data 810 may be applied to the patient's fit information 760 to determine the fit variables J. However, the fit variables J are used to select the best optimized stimulation parameters 750 for the patient from the general model 830. The model 830 may not be specific to the patient providing the fit information 760, and may represent a general modeling of preferred stimulation parameters, such as those noted based on empirical data to provide beneficial results on a larger subset of patients. The model 830 may include a range or volume of stimulation parameters that provide sub-perceptual stimulation, although this is not strictly necessary, and the model 830 may also include a range or volume of stimulation parameters that provide supra-perceptual stimulation. For example, the model 830 may include a previously described set of stimulation parameters 750. Figures 10A-13B Area 100 or relationship 98 discussed, reference Fig.18 The model 390 discussed, or other models developed in the future and indicative of beneficial stimulation parameters. Even if the fitting algorithm 740' does not select the best optimized stimulation parameters 750 from the optimized stimulation parameters 420 or 420' determined to be useful for a particular patient, it is expected that as more patients are successfully treated, the model 830 and training data 810 will evolve over time to allow the best optimized stimulation parameters 750 to be predicted for a given patient using the fitting information 760 for that patient.

[0287] Various aspects of the disclosed technology include processes that can be implemented in an IPG or ETS, or in an external device (such as a clinician programmer or an external controller) to present and operate GUI 64, which can be expressed in a formula and stored as an instruction in a computer-readable medium associated with such a device, such as stored in a magnetic memory, an optical memory, or a solid-state memory. The computer-readable medium with such stored instructions can also include a device that can be read by a clinician programmer or an external controller, such as in a memory stick or a removable disk, and can reside elsewhere. For example, a computer-readable medium can be associated with a server or any other computer device, thereby allowing instructions to be downloaded to a clinician programmer system or an external system or to an IPG or ETS via, for example, the Internet.

[0288] Although specific embodiments of the present invention have been shown and described, it should be understood that the above discussion is not intended to limit the present invention to these embodiments. It will be apparent to those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the present invention. Therefore, the present invention is intended to encompass substitutes, modifications and equivalents that may fall within the spirit and scope of the present invention as defined by the claims.

Claims

1. A spinal cord stimulator device comprising: multiple electrodes that can be inserted into the patient's spine; and A control circuit programmed to provide sub-sensory stimulation pulses to tissue of the patient at at least one of the plurality of electrodes according to a stimulation program defined by a frequency and an average charge per second value above or within one or more linearly bounded regions defined by: (i) (10 Hz, 6 μC / s), (10 Hz, 12 μC / s), (50 Hz, 55 μC / s), and (50 Hz, 27 μC / s); or (ii) (50Hz, 27μC / s), (50Hz, 55μC / s), (100Hz, 88μC / s), and (100Hz, 40μC / s); or (iii) (100 Hz, 40 μC / s), (100 Hz, 88 μC / s), (200 Hz, 151 μC / s), and (200 Hz, 67 μC / s); or (iv) (200Hz, 67μC / s), (200Hz, 151μC / s), (400Hz, 274μC / s), and (400Hz, 118μC / s).

2. The spinal cord stimulator device according to claim 1, wherein The control circuit is configured to provide the stimulation pulses at a constant current amplitude.

3. The spinal cord stimulator device of claim 1, wherein: Each pulse is followed by a charge recovery pulse phase.

4. The spinal cord stimulator device of claim 1, wherein: The spinal cord stimulator is programmable during a programming session, and wherein during the programming session, the subsensory stimulation pulses are washed in over a period of one hour or less.

5. The spinal cord stimulator device of claim 1, wherein: The spinal cord stimulator is programmable during a programming session, and wherein during the programming session, the subsensory stimulation pulses are washed in over a period of ten minutes or less.

6. The spinal cord stimulator device of claim 1, wherein: The control circuit is further configured to cease providing the subperceptual stimulation pulses, wherein effectiveness of the subperceptual stimulation pulses occurs during a washout period of ten minutes or more after the subperceptual stimulation pulses have ceased.

7. The spinal cord stimulator device of claim 1, wherein: The control circuit is further configured to cease providing the subperceptual stimulation pulses, wherein the effectiveness of the subperceptual stimulation pulses occurs during a washout period of one hour or more after the subperceptual stimulation pulses have ceased.

Images