Parametric techniques for sub-perception and paraesthesia therapy

Abstract

A system for closed-loop control of sub-perception spinal cord stimulation therapy is provided. The system includes an implantable pulse generator (IPG) with an evoked compound action potential (ECAP) recording module and an accelerometer. A regression model maps accelerometer data to optimal stimulation parameters for consistent neural activation. This enables closed-loop therapy in the sub-perception domain where ECAP features are visually undetectable. A clinician programmer applies the model and sub-perception dose to a therapy program deployed to the IPG. The model can be customized for each patient by collecting ECAP data and accelerometer data. The mapping reconciles ECAP-based control between the perception domain and the sub-perception domain by supplementing unavailable ECAP data with accelerometer data. This improves stimulation therapy for patients receiving sub-perception therapy and multi-sensor paresthesia therapy.

Description

Priority requirements

[0001] This application claims the benefit of U.S. Provisional Application No. 63 / 543,139, filed October 9, 2023, which is incorporated herein by reference in its entirety. Technical Field

[0002] This document generally relates to healthcare systems, and more specifically, but not in a limited manner, to systems, devices, machine-readable media, and methods for using healthcare-related data to improve patient monitoring, streamline patient triage, and provide treatment options for implantable electrical stimulation for pain treatment and / or management. Background Technology

[0003] Chronic pain, such as pain that has been present for most of a period of six months or longer during the previous year (or other definitions of chronic pain), is an extremely common complaint and is always associated with mental, physical, and / or psychological disorders. Chronic pain can originate from trauma, injury, or infection, or there may be a persistent cause of pain. Chronic pain can also occur without any evidence of prior injury or physical damage. Common types of chronic pain can include headache, lower back pain, cancer pain, arthritis pain, neurogenic pain (pain caused by damage to the peripheral or central nervous system), somatic pain, psychogenic pain (pain not caused by a prior illness or injury or any visible signs of damage inside or outside the nervous system), or other types of pain.

[0004] However, chronic pain is far more than just a physical sensation or feeling of pain. It's not merely a numerical score; it can be an exhausting change to a person's life every single day. Chronic pain has a significant impact on an individual's quality of life, affecting not only physical health but also emotional and social well-being. It can limit a patient's physical capacity, cause emotional and psychological distress, increase social isolation, create financial stress, and reduce overall quality of life, among other things.

[0005] Implantable devices can be configured to sense neural activity, such as evoked compound action potentials (ECAPs). The neural sensor can be an integral part of the device or part of a therapy delivery device. The delivered therapy can include electrotherapy or pharmacological therapy, for example. By way of example, implantable, attachable, wearable, or similar devices can be neuromodulators capable of delivering neuromodulation therapies. Examples may also include cardiac stimulators that monitor neural activity.

[0006] Neurostimulation (also known as neuromodulation) has been proposed as a therapy for a variety of conditions, injuries, and causes of chronic pain. Examples of neurostimulation include spinal cord stimulation (SCS), deep brain stimulation (DBS), peripheral nerve stimulation (PNS), functional electrical stimulation (FES), and similar stimulation. Implantable neurostimulation systems have been used to deliver such therapies. Implantable neurostimulation systems may include an implantable neurostimulator (also known as an implantable pulse generator (IPG)) and one or more implantable leads, paddle electrodes, or the like, each comprising one or more electrodes. An external experimental stimulator (ETS) is an external neurostimulation device used for a temporary trial period (typically prior to receiving an IPG). Implantable neurostimulators deliver neurostimulation energy through one or more electrodes placed on or near a target site in the patient's system.

[0007] Remote patient monitoring platforms for patients with neurostimulation devices allow clinicians, device representatives, patients, and caregivers to monitor large patient populations. However, checking on each patient individually to determine if they need assistance can be difficult and time-consuming. Improvements in management, triage, and notification of remotely managed patient populations are desired. Summary of the Invention

[0008] Various embodiments of this subject matter classify pain patients into one of a number of therapeutic categories based on parameterization techniques used for subsensory therapy and multisensor sensory abnormality therapy, remotely providing patients with treatment changes, monitoring parameters, or combinations thereof.

[0009] The implementation described in this subject matter may include one or more features, individually or in combination, as shown by way of example below.

[0010] An example of the subject (e.g., “Example 1”) may include a system for parameterizing a closed-loop algorithm for delivering subsensory therapy, the system comprising: a stimulation output circuit configured to deliver neural stimulation; a sensing circuit configured to sense neural signals indicative of neural responses, each of which is a response to the delivery of the neural stimulation; a stimulation control circuit coupled to the stimulation output circuit and the sensing circuit, the stimulation control circuit configured to control the delivery of the neural stimulation using a plurality of stimulation parameters; and one or more memories storing instructions that, when executed by one or more processors, cause the one or more processors to perform the following operations: record evoked compound action potential (ECAP) characteristics received from the sensing circuit; adjust the stimulation parameters at least in part based on the recorded ECAP characteristics to maintain consistent neural activation; record accelerometer data; generate a regression model for mapping the accelerometer data to optimal stimulation parameters to achieve consistent neural activation, the mapping enabling closed-loop control of the subsensory therapy; apply the regression model and subsensory dose levels to a plurality of subsensory therapy procedures; and deploy the regression model to a patient via the stimulation output circuit to deliver neural stimulation.

[0011] In Example 2, the subject matter described in Example 1 may optionally be configured such that the instruction performs the following operations: identifying a target range for the ECAP feature, the target range including an upper limit and a lower limit of the ECAP range; and determining consistent neural activation by maintaining the value of the ECAP feature within the target range.

[0012] In Example 3, the subject matter described in any one or more of Examples 1 to 2 may optionally be configured such that the regression model is selected from at least one of the following: a linear regression model, a support vector machine model, a neural network model, an autoregressive model, or a moving average model.

[0013] In Example 4, the subject matter described in any one or more of Examples 1 to 3 may optionally be configured such that the instruction performs the following actions: prompting the patient to perform one or more physical movements; and receiving patient input via a user interface that takes into account environmental factors affecting sensor readings.

[0014] In Example 5, the subject matter according to Example 4 can be optionally configured such that one or more body movements include one or more of a plurality of postural changes performed by the patient, including at least one of lying down, standing up, bending over, lateral twisting, performing activities of daily living, and leaning back.

[0015] In Example 6, the subject matter described in any one or more of Examples 1 to 5 may optionally be configured such that the instruction performs the following: identifies one or more settings in which changes are proposed for delivering neural stimulation to a patient, the neural stimulation being defined at least in part based on accelerometer data.

[0016] In Example 7, the subject matter described in any one or more of Examples 1 to 6 may optionally be configured such that the instruction performs the following operations: providing a monitoring platform; and receiving information about a plurality of accelerometer data representing overall patient health based on a combination of neural stimulation parameters, including: an electrical waveform; and the selection of an electrode through which the electrical waveform is delivered.

[0017] In Example 8, the subject matter described in Example 7 can optionally be configured to cause the instruction to perform the following operations: generate a suggestion for one or more additional combinations of neural stimulation parameters; and provide the suggestion to a monitoring platform.

[0018] In Example 9, the subject matter described in any one or more of Examples 1 to 8 may optionally be configured such that multiple accelerometer data provide supplementary data to the ECAP feature to optimize the stimulus in the sub-sensory domain.

[0019] In Example 10, the subject matter described according to any one or more of Examples 1 to 9 may optionally be configured such that the system is also configured to analyze data used for neural stimulation programming.

[0020] In Example 11, the subject matter described according to any one or more of Examples 1 to 10 may optionally be configured such that the instruction performs the following operations: initializing the closed-loop algorithm based at least in part on historical patient data; and adjusting the closed-loop algorithm using patient-specific data.

[0021] In Example 12, the subject matter described according to any one or more of Examples 1 to 11 may optionally be configured such that the instruction performs the following operation: recalibrates the mapping to take into account the changes identified in the implantable pulse generator (IPG), which is configured to deliver electrical stimulation via leads and electrodes.

[0022] In Example 13, the subject matter described in any one or more of Examples 1 to 12 may optionally be configured such that the regression model is also configured to generate a custom mapping for each patient by collecting ECAP data and accelerometer data in a clinical setting.

[0023] In Example 14, the subject matter described in Example 13 can optionally be configured such that the regression model collects ECAP factors during the mapping period, the ECAP factors being selected from at least one of the following: amplitude, detection threshold, perception level, linearity with stimulus intensity, stimulus parameters, electrode location, or neural activation level.

[0024] In Example 15, the subject matter described according to any one or more of Examples 1 to 14 may optionally be configured to further include: an anomaly detector configured to identify divergent outputs between the ECAP algorithm and the acceleration calculation method.

[0025] Example 16 includes topics such as methods, means for performing actions, machine-readable media including instructions that, when executed by a machine, cause the machine to perform an action, or means for performing actions. This topic may include parameterizing a closed-loop algorithm for providing subsensory therapy, the method comprising: delivering neural stimulation; sensing neural signals indicative of neural responses, each of which is a response to the delivery of the neural stimulation; using multiple stimulation parameters to control the delivery of the neural stimulation; recording evoked compound action potential (ECAP) characteristics received from sensing circuitry; adjusting the stimulation parameters, at least in part, based on the recorded ECAP characteristics, to maintain consistent neural activation; recording accelerometer data; generating a regression model for mapping the accelerometer data to optimal stimulation parameters for consistent neural activation, the mapping enabling closed-loop control of the subsensory therapy; applying the regression model and subsensory dose levels to multiple subsensory therapy procedures; and deploying the regression model to a patient via stimulation output circuitry to deliver neural stimulation.

[0026] In Example 17, the subject matter according to Example 16 may optionally be configured to further include: identifying a target range for the ECAP feature, the target range including an upper limit and a lower limit of the ECAP range; and determining consistent neural activation by maintaining the value of the ECAP feature within the target range.

[0027] In Example 18, the subject matter described according to any one or more of Examples 16 to 17 may optionally be configured to further include: prompting the patient to perform one or more physical movements; and receiving patient input via a user interface that takes into account environmental factors affecting sensor readings.

[0028] In Example 19, the subject matter according to Example 18 can be optionally configured such that one or more body movements include one or more of a plurality of postural changes performed by the patient, including at least one of lying down, standing up, bending over, lateral twisting, performing daily living activities, and leaning back.

[0029] In Example 20, the subject matter described according to any one or more of Examples 16 to 19 may optionally be configured to further include: identifying one or more settings in which changes for delivering neural stimulation to a patient are proposed, the neural stimulation being defined at least in part based on accelerometer data.

[0030] In Example 21, the subject matter described according to any one or more of Examples 16 to 20 may optionally be configured to further include: identifying one or more settings in which changes for delivering neural stimulation to a patient are proposed, the neural stimulation being defined at least in part based on accelerometer data.

[0031] In Example 22, the subject matter described in any one or more of Examples 16 to 21 may optionally be configured to further include: providing a monitoring platform; and receiving information about a plurality of accelerometer data representing overall patient health based on a combination of neural stimulation parameters, including: an electrical waveform; and selection of electrodes through which the electrical waveform is delivered.

[0032] In Example 23, the subject matter described in Example 22 may optionally be configured to further include: generating suggestions for one or more additional combinations of neural stimulation parameters; and providing such suggestions to a monitoring platform.

[0033] In Example 24, the subject matter described in any one or more of Examples 16 to 23 may optionally be configured such that multiple accelerometer data provide supplementary data to the ECAP feature to optimize stimulation in the sub-sensory domain; and also includes: analyzing data for neural stimulation programming.

[0034] In Example 25, the subject matter described in any one or more of Examples 16 to 24 may optionally be configured to further include: initializing the closed-loop algorithm based at least in part on historical patient data; and adjusting the closed-loop algorithm using patient-specific data.

[0035] In Example 26, the subject matter described in any one or more of Examples 16 to 25 may optionally be configured to further include: recalibrating the mapping to take into account changes identified in an implantable pulse generator (IPG) configured to deliver electrical stimulation via leads and electrodes.

[0036] In Example 27, the subject matter described in any one or more of Examples 16 to 26 may optionally be configured such that the regression model is also configured to generate a custom mapping for each patient by collecting ECAP data and accelerometer data in a clinical setting, wherein the regression model collects ECAP factors during the mapping period, the ECAP factors being selected from at least one of: amplitude, detection threshold, perception level, linearity with stimulus intensity, stimulus parameters, electrode location, or neural activation level.

[0037] An example of the subject (e.g., “Example 28”) may include a machine storage medium containing instructions that, when executed by the machine, cause the machine to perform the following operations: delivering a neural stimulus; sensing neural signals indicative of neural responses, each of which is a response to the delivery of the neural stimulus; controlling the delivery of the neural stimulus using a plurality of stimulation parameters; recording evoked compound action potential (ECAP) characteristics received from sensing circuitry; adjusting the stimulation parameters, at least in part, based on the recorded ECAP characteristics, to maintain consistent neural activation; recording accelerometer data; generating a regression model for mapping the accelerometer data to optimal stimulation parameters for consistent neural activation, the mapping enabling closed-loop control of subsensory therapy; applying the regression model and subsensory dose levels to a plurality of subsensory therapy procedures; and deploying the regression model to a patient via stimulation output circuitry to deliver the neural stimulus.

[0038] In Example 29, the subject matter according to Example 28 may optionally be configured to further include: identifying a target range for the ECAP feature, the target range including an upper limit and a lower limit of the ECAP range; and determining consistent neural activation by maintaining the value of the ECAP feature within the target range.

[0039] In Example 30, the subject matter described according to any one or more of Examples 28 to 29 may optionally be configured to further include: prompting the patient to perform one or more physical movements; and receiving patient input via a user interface that takes into account environmental factors affecting sensor readings.

[0040] In Example 31, the subject matter according to Example 30 may optionally be configured such that one or more body movements include one or more of a plurality of postural changes performed by the patient, including at least one of lying down, standing up, bending over, lateral twisting, performing activities of daily living, and leaning back.

[0041] In Example 32, the subject matter described according to any one or more of Examples 28 to 31 may optionally be configured to further include: identifying one or more settings in which changes for delivering neural stimulation to a patient are proposed, the neural stimulation being defined at least in part based on accelerometer data.

[0042] In Example 33, the subject matter described in any one or more of Examples 28 to 32 may optionally be configured to further include: providing a monitoring platform; and receiving information about a plurality of accelerometer data representing overall patient health based on a combination of neural stimulation parameters, including: an electrical waveform; and selection of electrodes through which the electrical waveform is delivered.

[0043] In Example 34, the subject matter described in Example 33 may optionally be configured to further include: generating suggestions for one or more additional combinations of neural stimulation parameters; and providing such suggestions to a monitoring platform.

[0044] An example of the subject (e.g., “Example 35”) may include a system for parameterizing a closed-loop algorithm for delivering subsensory therapy, the system comprising: one or more hardware processors of a machine; and at least one memory storing instructions that, when executed by the one or more hardware processors, cause the system to perform the following operations: recording evoked compound action potential (ECAP) characteristics received from sensing circuitry; adjusting stimulation parameters, at least in part, based on the recorded ECAP characteristics, to maintain consistent neural activation; recording accelerometer data; generating a regression model for mapping the accelerometer data to optimal stimulation parameters for consistent neural activation, the mapping enabling closed-loop control of the subsensory therapy; applying the regression model and subsensory dose levels to multiple subsensory therapy procedures; and deploying the regression model to a patient via stimulation output circuitry to deliver neural stimulation.

[0045] Example 36 is at least one machine-readable medium comprising instructions that, when executed by a processing circuitry system, cause the processing circuitry system to perform operations to implement any one or more of Examples 1 to 35.

[0046] Example 37 is an apparatus that includes means for implementing any one or more of Examples 1 to 35.

[0047] Example 38 is a system for implementing any one or more of Examples 1 through 35.

[0048] Example 39 is a method for implementing any one or more of Examples 1 through 35.

[0049] This invention provides an overview of some of the teachings of this application and is not intended to be exclusive or exhaustive of the subject matter. Further details regarding the subject matter are found in the detailed description, drawings, and appended claims. Other aspects of this disclosure will be apparent to those skilled in the art upon reading and understanding the following detailed description and viewing the drawings, which form a part thereof, and none of these detailed descriptions and drawings should be considered limiting. The scope of this disclosure is defined by the appended claims and their legal equivalents. Attached Figure Description

[0050] This document generally relates to medical systems, and more specifically, but not in a limiting way, to systems, devices, machine-readable media, and methods for using healthcare-related data to improve patient monitoring, streamline patient triage, provide therapeutic options for implantable electrical stimulation for pain treatment and / or management, as well as parameterization techniques for subsensory therapy and multisensor sensory abnormality therapy. Various embodiments are illustrated by way of example in the accompanying figures. Such embodiments are exemplary and are not intended to be exhaustive or exclusive embodiments of the subject matter.

[0051] Figure 1 An embodiment of a monitoring system according to one embodiment is shown, which may include devices, external systems or other healthcare-related data sources configured to collect healthcare-related data to characterize parametric techniques for subsensory therapy and multisensor sensory abnormality therapy.

[0052] Figure 2 An example embodiment of data in a feature map according to one embodiment is shown.

[0053] Figure 3 A series of accelerometer graphs according to one embodiment are shown.

[0054] Figure 4 A diagram is shown illustrating the definition of a subsensory scaling factor according to one embodiment.

[0055] Figure 5 A series of accelerometer graphs according to one embodiment are shown by way of example rather than limitation.

[0056] Figure 6 The stimulus amplitude curve according to one embodiment is shown by way of example rather than limitation.

[0057] Figure 7 A process flowchart according to one embodiment is shown by way of example rather than limitation.

[0058] Figure 8An embodiment of a stimulation device and lead system including an ECAP factor and sensing control is shown by way of example and not limitation, according to one embodiment.

[0059] Figure 9 An example is illustrated by way of example of a data analysis computing system according to one embodiment interacting with a clinical physician and patient interactive computing device for operating or monitoring a neurostimulation device based on text input analysis.

[0060] Figure 10 A block diagram of an embodiment of a system (e.g., a computing system) for implementing a neurostimulation programming circuitry system to program an implantable electroneurostimulation device is shown by way of example.

[0061] Figure 11 A block diagram of an embodiment of a system (e.g., a computing system) for performing patient data analysis in conjunction with programmed operations is shown by way of example.

[0062] Figure 12 An embodiment of a neural stimulation system according to one example is illustrated by way of example.

[0063] Figure 13 An example of a patient system according to one embodiment and an example of devices that may comprise the components of the patient system are shown by way of example and not limitation.

[0064] Figure 14 The invention illustrates, by way of example and not limitation, a neural stimulation system (such as...) according to one embodiment. Figure 17 Examples of programming systems and data analysis systems used together with implantable neuromodulation systems.

[0065] Figure 15 An example is illustrated by way of example of a data analysis computing system according to one embodiment interacting with a clinical physician and patient interactive computing device for operating or monitoring a neurostimulation device based on text input analysis.

[0066] Figure 16 An embodiment of a data processing flow for influencing neurostimulation therapy of human patients based on text and device data processing is illustrated by way of example.

[0067] Figure 17 The implementation of a spinal cord stimulation system or deep brain stimulation system according to one embodiment is shown by way of example and not limitation. Figure 12 The neural regulation system.

[0068] Figure 18This is a block diagram illustrating a machine in the form of an example computer system according to one embodiment, within which a set of instructions or a sequence of instructions can be executed to cause the machine to perform any of the methods discussed herein. Detailed Implementation

[0069] The following detailed description of this subject matter takes into account the accompanying drawings, which illustrate by way of illustration specific aspects and embodiments in which the subject matter may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the subject matter. Other embodiments may be utilized, and structural, logical, and electrical changes may be made without departing from the scope of the subject matter. References to “a,” “an,” or “various” embodiments in this disclosure do not necessarily refer to the same embodiment, and such references contemplate more than one embodiment. Therefore, the following detailed description should not be considered limiting, and its scope is defined only by the appended claims and the full scope of their legal equivalents.

[0070] Overview

[0071] Example embodiments of systems, methods, machine-readable media, software, and artificial intelligence provide improved approaches for enhancing therapies (such as deep brain stimulation and / or spinal cord stimulation) based on recorded ECAP signals and / or additional indicators to adapt stimulation therapies to patients receiving subsensory therapy, multisensor sensory impairment therapy, and similar therapies. Example embodiments of this disclosure provide systems for reconciling the use of one or more ECAP signals for parameterizing closed-loop algorithms for subsensory therapy and multisensor sensory impairment therapy, without visualizing, recording, and / or accessing the ECAP signals during or before / after periods of implementing changes to the therapy (e.g., treatment procedures).

[0072] Existing methods for treating chronic pain include routine in-person clinic visits, telemedicine appointments, or other scheduled physician visits, all of which require face-to-face pain assessments in a clinical setting. Often, patients will ultimately need to provide detailed feedback to clinicians before treatment problems can be identified and changes to neurostimulation therapy can be implemented; this process, from onset to triage to treatment implementation, can take weeks or even months. Conventional neurostimulation systems are insensitive to the possibility that, despite chronic pain potentially limiting a patient's functional abilities and causing mobility problems, improvements in pain perception (e.g., lower pain scores) may not always be accompanied by, or synchronized with, improvements in a patient's functional abilities, mobility status, and overall quality of life. For example, a chronic pain patient may report a significant reduction in pain scores but remain bedridden without any improvement in their physical or functional abilities. Similarly, a chronic pain patient may not report significant pain relief even if they have begun to walk normally, sleep, participate in more activities, change their lifestyle, or similar activities. Such conventional systems fail to provide support, monitoring, and triage for pain patients in their daily lives. Furthermore, existing chronic pain monitoring mechanisms fail to take into account or incorporate the vast amount of mental, physical, and functional pain data available to patients throughout their daily lives.

[0073] Further prior art solutions for sensing neural activity, such as evoked compound action potentials (ECAPs), can utilize implanted electrodes. However, implanted electrodes can be adversely affected by changes over time, such as impedance variations, lead migration, scar tissue, similar physical or physiological problems, or combinations thereof. In prior art solutions, based on current ECAP recording devices, ECAP signals are limited to being recorded only at or above the level of sensory stimulation in the patient (e.g., to provide data for visualization by the user and further analysis by a computer or other computing device, such as a microcontroller). When a patient receives a therapy based on sensory abnormalities (e.g., sensory-based, above-sensory, sensory therapy, etc.), this ECAP signal (e.g., level) can typically be recorded using electrophysiological measurements and identified through analysis based on computer codes (or, for example, through visual inspection by a human operator or a machine learning model). However, due to the lack of ECAP signal recording, when subsensory therapy (e.g., Fast-Onset Subsensory Therapy (FAST), nonsensory therapy procedures, etc.) is applied to a patient, ECAP may not be recorded using electrophysiological measurements or otherwise available for visual examination at any given time period and / or during any given patient action.

[0074] Existing methods for neurostimulation systems rely heavily on their post-manufacturing programmability. One limiting factor in the application of neurostimulation therapy is the frequent delays in implementing new or improved neurostimulation therapies, even when multiple advanced procedures are available from neurostimulation devices. Such delays may stem from the infrequent provision of care by supervising clinicians or other healthcare professionals, and the lack of clear information regarding treatment outcomes (e.g., whether subsensory therapy is providing relief to the patient). Various approaches to neurostimulation programming and customization have attempted more dynamic forms of open-loop and closed-loop programming to allow new neurostimulation parameters or procedures to be introduced, deployed, tested, and tuned by clinicians, patients, software programs, models, or the like. While some neurostimulation devices offer the ability for patients to switch between procedures or change the level of a particular stimulation effect, it is often unclear whether such changes (or which changes) are beneficial to the patient and lead to improvements in their medical condition.

[0075] The exemplary embodiments of this disclosure improve upon existing models by providing evoked response-based subsensory therapy, overcoming current technical challenges that include a substantial and growing portion of stimulation therapy for patients with chronic pain and other users. In many patients, evoked responses may be unavailable below the perception threshold, particularly at fractional stimulation amplitudes (e.g., 20-50% of the perception threshold) typically programmed for subsensory therapy. Other sensors (including onboard accelerometers) are available in both domains (e.g., subsensory and hypersensory), thus the mapping between the two domains enables closed-loop therapy in evoked response-based subsensory therapy. By collecting data and constructing regression models, a clinical mapping between ECAP and accelerometer data will be generated for each patient; in other examples, a mapping between stimulation parameters controlled by the ECAP level and accelerometer data will be generated. In additional examples, a second sensor (or one or more additional sensors) for performing the mapping to the closed-loop output can enhance the overall algorithm performance and make it robust to noisy and uncertain environments, and assist in anomaly detection. For example, closed-loop functionality may include models and / or algorithms used by the neurostimulator to update therapeutic parameters (e.g., amplitude) based on sensor data (e.g., ECAP signals, accelerometer readings, etc.).

[0076] According to example embodiments, both subsensory and sensory therapy are effective (e.g., therapeutically effective for the patient, providing relief to the patient, delivering pain relief, etc.). While not all sensory and / or subsensory levels record ECAPs, therapeutic properties (such as activation of neural activity on one or more patient nerves) can still be provided to the patient receiving the therapy. In some examples, once the evoked compound action potential (ECAP) recording is visualized, the ECAP recording begins to grow in size; for example, the ECAP recording may grow linearly with the intensity of the stimulus input in the patient therapy. However, below a certain stimulus intensity, the ECAP level may not be recorded (or otherwise seen or visualized), while nerve fiber activation still occurs, although it is below the sensitivity level (or the noise level of the recording system). For example, the availability of the ECAP signal can vary during sensory and subsensory therapy, and although therapeutic effects have been established at both levels, the ECAP level may appear below its unobservable detection threshold for technical rather than physiological reasons. Once detectable, its amplitude grows approximately linearly with the stimulus intensity; for example, the system may empirically observe ECAPs slightly below the patient-reported sensory level.

[0077] However, stimulation therapy typically targets amplitudes of 20-40% of perception, usually below the ECAP detection threshold. Example embodiments use accelerometers to supplement data to achieve ECAP-based control within this sub-sensory domain. For example, as intensity increases (as the patient receives higher stimulation intensity), the ECAP recording can represent a large (or even larger) number of data points of the individual activated fibers (in some examples, all fibers are activated simultaneously). In some examples, the ECAP level can be recorded at or below the level reported by the patient (e.g., the patient's perceived or interpreted level of sensation). For example, ECAP recordings can be expected anywhere between 75% and 80% of the perception threshold for, for example, a patient-programmed therapy (e.g., sub-sensory therapy, multisensor sensory abnormality therapy, etc.). For example, stimulation amplitudes at approximately 20% of the perception threshold can be recorded by an ECAP recorder. In some exemplary embodiments, the therapeutic effect has been well established, and the underlying neural mechanisms being engaged are present, but these mechanisms can occur at different levels.

[0078] According to an example embodiment of this disclosure, when ECAP is measured, it is measured by a sensor that exists or is currently available under existing technology. According to an example embodiment of this disclosure, the device will have a circuitry capable of capturing electrophysiological signals from, for example, one or more non-stimulating contacts on a lead (or paddle electrode), such that the non-stimulating contacts on one or all of the leads will have distinct paths (e.g., two or more paths) returning to a stimulator configured to connect the lead. For example, if two distinct paths exist, one path is a current source that will provide therapy; and the other path will conform to analog-to-digital converter and amplifier circuitry, which can modulate the signal so that it can be measured and processed by a stimulator (e.g., stimulator hardware, software, or a combination thereof). The example embodiments of this disclosure can be applied to various versions of stimulation systems, devices, and / or methods; for example, embodiments using leaded stimulators or embodiments using paddle electrodes, which can interface with the stimulator in the same, similar, or different ways (e.g., circuitry, connections, etc.).

[0079] Then, examples of sub-sensory spaces can be supplemented with additional circuitry configured or programmed to perform mapping. According to the example embodiments presented herein, these examples provide spinal cord stimulation, and depending on the intensity of the stimulation, the patient (and applicable programming) can observe an evoked response. Embedded in the circuitry of the stimulation device is an accelerometer configured to (e.g., be able to) measure orientation, motion, posture changes, or similar information. For example, a triaxial accelerometer can be configured to measure the patient's motion along the x, y, z axes, and combinations thereof. Based at least in part on triaxial accelerometer recordings, example embodiments of this disclosure provide a system, method, machine-readable medium, etc., for initiating at a higher intensity level, wherein the embodiments provide visualization of data relating to both ECAP and accelerometer recordings. Using the relationship between accelerometer and ECAP recordings, or the relationship between accelerometer and stimulation parameters (if closed-loop control of ECAP is being performed), examples can directly link (e.g., correlate, correlate, etc.) the accelerometer and stimulation amplitude recordings.

[0080] Based on the established relationships, the mapping associated with orientation or location (or both) is the patient on one or all three axes. For example, this mapping can provide data on which ECAP range corresponds or which stimulus amplitude is associated with in order to keep the ECAP range constant. According to example embodiments, this mapping can provide information on how each stimulus parameter (or combination of stimulus parameters) corresponds to subsensory therapy and multisensor sensory abnormality therapy. For example, to achieve a subsensory therapy level, example embodiments may apply a scaling factor that indicates the expected (e.g., desired) intensity level for the specific therapy to be applied to the patient. In some examples, a scaling factor can be applied to multiply in an attempt to predict stimulus amplitude using accelerometer recordings. For example, stimulus amplitude can be predicted in a higher intensity space, where the system or method can track the ECAP to ensure its constancy. The amplitude is then predicted by multiplying the ECAP level by the subsensory scaling factor based on consistent ECAP recordings. For example, a sub-sensory scaling factor can be 20% or 10% or another number or percentage (e.g., a factor) that can reduce the scaled dose to the dose level that the user (e.g., patient, doctor, clinician, programming model, etc.) wants or expects in the sub-sensory range.

[0081] According to examples of this disclosure, the mapping can be created (e.g., generated) based on algorithms or methods established (e.g., constructed, prepared, etc.) in a clinic, home, or other location in the patient's presence. For example, a representative or clinician can configure one or more anticipated therapies (e.g., above-perceived or below-perceived) such as stimulation procedures based on the clinical environment. In some examples, the clinician will begin with an above-perceived level of therapy (e.g., for configuration purposes) where the ECAP can be measured (e.g., recorded, viewed, visualized, etc.). In such an example, the clinician can ensure that the accelerometer is working and available. For example, this example would record ECAP data and / or accelerometer readings while attempting to maintain (e.g., keep constant) the ECAP constant using a control algorithm that adjusts one or more of the multiple stimulation parameters.

[0082] According to examples, the configured model can be generated as a general (e.g., general, exemplary, etc.) model that will be used as a starting point (or initial point) across all or part of the patients on various bases (e.g., per pain level, per symptom, per implementation method, etc.). For example, different patients may have IPGs implanted in different orientations, where different patients move and / or respond to stimuli or other properties in unique ways, which can be used to determine or help identify how the mapping will occur or be visualized. For example, the system can be modeled on a per-patient, per-damage, per-symptom, etc. According to some exemplary embodiments, patients with stimulation devices for receiving subsensory therapy and / or multi-sensor sensory abnormality therapy (capable of measuring ECAP or where ECAP can be measured in other ways (e.g., accelerometer and stimulation capacity)) can receive parameterized closed-loop models to optimize (e.g., improve, adjust, enhance) treatment outcomes.

[0083] Figure 1 A block diagram 100 of a monitoring and processing system 130 is shown. This system 130 may include medical devices, external systems, or other healthcare-related data sources configured to collect healthcare-related data to characterize parameterization techniques for subsensory therapy and multisensor sensory abnormality therapy. An embodiment of the monitoring and processing system 130 is shown by way of example, configured to collect and analyze healthcare-related data in conjunction with the use of neurostimulation procedures and closed-loop neurostimulation programming.

[0084] The monitoring and processing system 130 shown is configured to include at least one data collection platform 101, which operates to collect and provide data input / output 104. The data collection platform 101 is configured to process (e.g., filter, extract, transform) the input data to generate analytical data 140, which can then be used for one or more iterations of programming model training 102. Programming model training 102 can be a closed-loop programming model, or an open-loop system where actions from the controller are independent of the process output (which is a system variable being controlled), or other programming models that provide feedback to adjust system operation. The data collection platform 101 can also generate output data in the form of recommendations, suggestions, and other information related to the collection of input data.

[0085] The monitoring and processing system 130 may be implemented at one or more servers or other systems remotely located relative to the patient. The monitoring and processing system 130 may use various network protocols to communicate and transmit data over one or more networks, which may include the Internet. The monitoring and processing system 130 and the data collection platform 101 may include at least one processor configured to execute instructions stored in memory (e.g., depicted as processor / memory 105) to generate or evaluate data output 106, acquire or evaluate data input 108, and perform data processing 107 on both the inputs and outputs, as well as accompanying training data.

[0086] Data input 108 can be processed to generate analytical data 140, which represents a transformed or refined version of data based on actual patient use, patient feedback (e.g., feedback indicating which settings are effective or not yet effective), patient parameter inputs, and the like. Analytical data 140 may include additional data processing, including, for example, parameter data 141, clustering scheme data 142, accelerometer data 143, ECAP data 144, and alert data 145.

[0087] Data input 108 may include information obtained directly from the patient and may include various types of general health data 110 associated with the patient or treatment. Examples of general health data 110 may include patient data 128, medical device data 112, patient environment data 113, therapy data 114, patient lifestyle data 117, supplementary data 118, or various combinations thereof. Patient data 128 may include objective data 115, such as data collected from physiological sensors; and subjective data 116, such as data collected from user-answered questions (in combination with...). Figure 2 (Detailed description). Other examples of how to collect objective and subjective data from patients are discussed in the more examples below.

[0088] User data input / output system 120 may be implemented at the patient's location or on one or more devices operated by the patient, such as via a smartphone, personal computer, tablet, smart home device, remote programmer, programming device, or another computing device or platform capable of collecting input and providing output. User data input / output system 120 may include at least one processor configured to execute instructions stored in memory (e.g., depicted as processor / memory 105) to provide or generate data inputs and outputs 104 for receiving or collecting general health data 110. One such example is a user interface application 124 implemented as a graphical user interface (GUI). Examples of GUIs for data input and output include smartphone apps, controller apps, or similar applications. Another example is sensor data processing 122, which can be used to determine a patient's physiological state (e.g., activity level, sleep duration) when using a particular procedure.

[0089] User data input / output system 120 may include sensors 126 or interface with sensors 126, such as physiological sensors that can detect information (e.g., activity, sleep, etc.), events, indicators of device usage, patient compliance with data collection and / or therapy, or various combinations thereof. User data input / output system 120 may directly or indirectly capture measurements from sensors 126 or from associated external systems. Examples of external systems include remote controls, programmers, mobile phones, tablets, smartwatches, personal computers, etc. The external system may be configured to provide medical device data 112, patient environment data 113, therapy data 114, patient lifestyle data 117, and / or additional data 118 from associated devices.

[0090] The programming model training 102 operates to update the programming model by training, retraining, or updating the model (e.g., an artificial intelligence model, such as a neural network) based on the generated analytical data 140. One or more instances of the model can be trained to generate programs and program parameters, including utilizing one or more patient-specific models. The closed-loop programming model training 102 can take into account other aspects of patient-specific data or population data, such as overall health data 110, sensor data, and rules and information from various data sources. Figure 8 More details about the example method are described and depicted.

[0091] In some examples, the data collection platform 101 and the monitoring and processing system 130 can directly interface with one or more medical devices, external systems, or other healthcare-related data sources to collect overall health data 110. One or more of the medical devices, external systems, and / or other patient-related data sources may include technologies used by the monitoring and processing system 130 to collect and / or record data, and thus may form part of the data collection platform 101. Examples of devices or medical devices include implantable devices, wearable devices, and / or operable interconnected devices. Implantable devices can be configured to sense neural activity, such as evoked compound action potentials (ECAPs), which can identify and record neural conduction, sensation, activity, and other patient monitoring information. Implantable devices (such as those combining...) Figure 15 The stimulation device 1502 described and depicted can be configured to use electrodes to sense neural activity (e.g., ECAP) and can also be configured to evaluate sensing electrodes as discussed in this document.

[0092] Implantable devices can also be configured to deliver electrical therapies, such as cardiac rhythm management therapy, neuromodulation therapy, subsensory therapy, multisensor sensory abnormality therapy, similar therapies, or combinations thereof. Therefore, by way of example and not limitation, an implantable device can be configured to deliver electrical therapy via electrodes (such as those combined with…) Figure 12 The described and depicted electrode 1206 is a neuromodulator that delivers neuromodulation energy (such as in the form of electrical pulses) to one or more neural targets (between the upper and lower limits of the ECAP range). The delivery of neuromodulation can be controlled using multiple modulation parameters that specify the electrical waveform (e.g., pulse, pulse pattern, other waveform shapes, etc.) and the selection of the electrode through which the electrical waveform is delivered. Existing waveforms stored in an external storage device can be defined at least in part by one or more parameters, including but not limited to: amplitude, pulse width, frequency, duration, electrode configuration, total charge injected per unit time, cycle (e.g., on / off time), waveform shape, spatial location of the waveform shape, pulse shape, number of phases, phase sequence, inter-phase time, charge balance, and ramp.

[0093] In various example embodiments, at least some of the multiple control parameters can be programmed by a user, such as a physician, caregiver, device representative, patient, hardware system, software system, or a combination thereof. The programming device can provide the user with access to user-programmable parameters, such as those used in parameterization techniques for subsensory therapy and multisensor sensory abnormality therapy.

[0094] Example embodiments of the medical device can be configured to collect data only, deliver therapy only, or collect data and deliver therapy simultaneously. The medical device can be configured to collect and provide medical device data, such as device model, configuration, settings, and similar data. Therefore, the medical device can provide patient data 128, medical device data 112, patient environment data 113, therapy data 114, patient lifestyle data 117, and / or additional data 118, particularly in specialized neurostimulation programming environments.

[0095] Other healthcare-related data sources may include patient data received via a provider's server storing the patient's health records. For example, a patient may use a patient portal to access their health records, such as test results, doctor's notes, prescriptions, and similar records. Other healthcare-related data sources may include data from applications on the patient's smartphone or other computing device, or data accessed by those applications on servers. In another example, applications on the phone or patient's device may include, or can be configured to access, environmental data, such as weather data and air quality information, or location elevation data, such as that determined using cellular networks and / or Global Positioning System (GPS). Weather data may include, but is not limited to, air pressure, temperature, clear or cloud cover, wind speed, and similar data. By way of example, and not limitation, this type of data may include heart rate, blood pressure, weight, and similar data collected from the patient at their home. Environmental factors may be taken into account using patient input or other environmental data.

[0096] In other examples, this type of data may include ECAP characteristics such as amplitude, detection threshold, sensory level, linearity, stimulation parameters, electrode / lead / paddle electrode location, neural activation, or similar features. The amplitude or magnitude of the recorded ECAP signal provides information about the number of nerve fibers activated by electrical stimulation. In other examples, the temporal distribution of the recorded ECAP signal provides information about the type or size of nerve fibers activated by electrical stimulation. A larger ECAP amplitude indicates more nerve fiber recruitment. ECAP exhibits a detection threshold below which it cannot be observed or recorded, even if neural activation is occurring. This is a technical limitation, not a physiological one. For example, in some examples, ECAP may only be recordable at or above the patient's sensory level. In other examples, ECAP signals may be undetectable or barely detectable during subsensory stimulation. Once above the detection threshold, the ECAP amplitude increases approximately linearly with increasing stimulation intensity. Factors similar to stimulation amplitude, pulse width, and frequency affect the ECAP response and must be optimized for target fibers. The recording electrode location affects the ECAP signal and must be carefully placed for effective detection of the response.

[0097] While not always recordable, ECAPs provide objective information about the level of neural activation achieved by stimulation (e.g., the relief provided to the patient). ECAPs serve as feedback biomarkers for a closed-loop system that automatically maintains optimal stimulation levels and neural activation. For example, ECAP characteristics and detection can be based on variations between patients such as electrode placement, neural health, body shape, patient movement, body posture, movement changes, or similar factors. Key ECAP features to consider include, for example, amplitude, detection threshold, perceived level, linearity with stimulus intensity, dependence on stimulus parameters, recording location, relationship to neural activation levels, and variability between patients.

[0098] Data input / output 104 can provide data transmitted via at least one network. Data transmission can utilize various network protocols to communicate and transmit data over one or more networks, which may, but are not necessarily, include the Internet and / or various wireless networks, including short-range wireless technologies such as Bluetooth. Data can be transmitted directly from at least one external system and / or can be transmitted directly from at least one medical device. Furthermore, the external system can be configured to receive data from the associated medical device and / or from other healthcare-related data sources, and then transmit the data to the data receiving system via the network.

[0099] Figure 2 An example embodiment of data in a feature plot 200 according to one embodiment is shown. According to one example embodiment, feature plot 200 shows an approximately constant neural activation range 250, which is plotted as a function of an evoked compound action potential (ECAP) range 202 over time 204.

[0100] Figure 2 A feature map 200 representing the ECAP level is shown, and the system can extract features (e.g., peak-to-peak amplitudes from the signal region of interest) from the recorded or viewed ECAP level. Based on these features, the system can isolate a specific window of the recorded signal. In some examples, the window of the recorded signal can be based on, for example, when the signal appears (e.g., looks, measures, etc.) largest or most significant. At this time, the system (via clinician programming, algorithmic models, or similar means) can adjust (e.g., make adjustments) the stimulation amplitude and / or current (e.g., Figure 4 The current amplitude diagram 400 describes and depicts the current amplitude.

[0101] Return to Figure 2The approximate constant neural activation range 250 in feature graph 200 includes a series of data points (each represented by a single circle on the graph) between ECAP ranges of 0 to 80 (as shown on the y-axis) over a time 204 ranging from 0 to 200 seconds (as shown on the x-axis). These ranges are non-limiting example ranges. For example, the constant activation range could include a subset of the entire range, such as within an ECAP range of 35 to 55. Feature graph 200 illustrates consistent (or approximately consistent) neural activation via control (e.g., closed-loop control) at stimulus levels that elicit a response. In some examples, the elicited response may be above, at, or below the perceptual level.

[0102] Hundreds of data points (which may range from tens to thousands depending on the feature map example) provide data points starting from position 208 within a time period of approximately 200 seconds (about 3.33 minutes) of a day. However, those skilled in the art will understand that data can be collected at any time period, whether continuous, semi-continuous, distributed, or spanning minutes, hours, days, months, etc. In feature map 200, after position 208, the patient positions movements in various postures within the time period. For example, from position 208, the patient moves to a seated position (first time) 210 for approximately 20 seconds until time t1 212. After time t1 212, the patient moves to a seated position (with neck tilted back) 214 for approximately 10 seconds until time t2 216. After time t2 216, the patient moves to a standing or rising position 218 for approximately 40 seconds until time t3 220. After time t3, the patient moves to a standing position 222 for about 30 seconds until time t4 224. After time t4 224, the patient moves to a backward position (or is in a backward position) 226 for about 40 seconds until time t5 228, at which point the patient moves to a sitting position again (for the second time) 230.

[0103] Based on hundreds of ECAP ranges (ECAPs) 202 data points over time 204 (approximately 0 to 200 seconds), cluster ranges are generated, representing constant neural activation 250 ranges between ECAP ranges. In some examples, clustering data points into dense clusters indicates similarity between neighboring values, a phenomenon known as "clustering" in data visualization. Clustering occurs when a subset of data exhibits high consistency in its measured properties. Identifying such clusters using graphical techniques or algorithmic methods can elucidate the underlying structure within the data based on shared characteristics. For example, constant neural activation 250 ranges can display well-defined clusters that represent distribution patterns corresponding to categories or correlations in underlying phenomena within the ECAP ranges. For instance, discernible cluster counts, densities, shapes, and separations provide insights into distributions and relationships within ECAP range data. Cluster analysis includes formal statistical methods for extracting clusters from data, among which popular techniques include k-means clustering, hierarchical clustering, density-based clustering, other clustering algorithms, or combinations thereof. However, not all data points may exhibit clustering behavior, and there may be outliers that are highly dissimilar to other observations, such as at the beginning of feature map 200; the initial 208 range may not be useful data. Since clustering provides a similar natural tendency for data points to cluster in data visualization, the constant neural activation range 250 allows (in conjunction with below) Figure 4 and Figure 5 Explanation of patterns and categories (descriptions and depictions).

[0104] In some examples, for simplification and pattern analysis purposes, classifying continuous data into discrete groups to demonstrate consistent neural activation via closed-loop control at stimulus levels can map sensor data to stimulus parameters called “bins” (also known as “data binning”). Binning can include dividing a dataset into intervals, segments, value ranges, etc., referred to as “boxes.” These boxes typically divide the data range into equal-sized, continuous, and non-overlapping intervals. Binning thus transforms quantitative data into qualitative categories (e.g., sitting posture (first time) 210, standing posture 218, standing posture 222, etc.). This discretization of continuous variables facilitates understanding of the data distribution by reducing noise and highlighting trends. For example, the ECAP range for each bin can highlight sensor data at least partially or entirely based on stimulus parameters. In some examples, one or more stimulus parameters can be exposed via mapping. However, choosing an appropriate bin size requires a trade-off between overly fine divisions (introducing complexity) and overly coarse divisions (blurring insights). While common bin sizes are 5, 10, or 20 units, the optimal bin width depends on the data distribution. Therefore, depending on the example, data binning can be used (as opposed to or in combination with clustering) as a valuable visualization method for extracting the underlying structure of a broad dataset related to stimulus parameters by carefully aggregating data into discrete bins.

[0105] Based on some examples, a closed-loop machine learning model refers to a model that includes a feedback loop to continuously update and improve its performance based on its own output or predictions. This creates a closed-loop system where the model's predictions are fed back into the model as new input data. The key advantage of this mechanism is that the model can automatically and continuously learn and adapt over real-time, near-real-time, or other fixed or varying time periods without requiring manual updates or retraining. Feedback loops enable continuous learning as new data, including the model's own predictions, is fed back into the model (e.g., algorithms, software, artificial intelligence, machine learning, etc.); the feedback loop can progressively improve itself and optimize its performance. The learning process occurs automatically in a dynamic, iterative manner. As conditions change or new data emerges, the model can adapt and be customized without human intervention.

[0106] For example, closed-loop models are well-suited for applications requiring real-time decision-making, control, optimization, or a combination thereof. Example use cases include robot control systems, automated trading systems, predictive maintenance systems, recommendation engines, and more. Automatic feedback loops allow models to continuously optimize and refine their performance for these types of applications. Overall, the adaptation or learning process by which models and / or algorithms improve predictions over time provides machine learning models with an efficient and automatic way to learn and adapt on the fly based on their own output. This real-time adaptability makes closed-loop models very powerful for applications that require continuous optimization and improvement in dynamic environments. Continuous feedback loops allow models to learn by doing so in iterative, self-improving loops.

[0107] Those skilled in the art will understand that the algorithms and models presented in the examples herein can alternatively, additionally, or otherwise employ non-adaptive or static machine learning models for implementing parameterization for subsensory therapies according to the examples of this disclosure. However, for the sake of simplicity, closed-loop algorithms are used without limitation. For example, non-adaptive machine learning can refer to a model trained on a fixed dataset and deployed without any feedback loops. In a non-adaptive system, the model makes predictions on new input data but does not use those predictions to further update itself. Non-adaptive or static models can be trained, validated, and then statically deployed based on historical training data. Once deployed, they do not change or adapt. Learning is open, without any feedback flow. Furthermore, a neural stimulation model in which sensor feedback is not used to adjust therapy parameters can be referred to as an open-loop model.

[0108] Figure 2 Feature map 200 provides data for creating the mapping. For example, the ECAP data shown in feature map 200 depicts consistent neural activation, which is used to create (e.g., generate) Figure 3 and Figure 4 The preconditions or ideal conditions for the mapping between the described data. For example... Figure 2 As depicted, arrow 206 continues to... Figure 3 Example embodiment of the accelerometer data shown.

[0109] Figure 3 An example embodiment based on this disclosure is shown. Figure 2 The provided mapping and accelerometer graphs 310 / 320 / 330, which contain data received via arrow 306, are a series of accelerometer charts 300.

[0110] Figure 3Examples of three accelerometer plots are shown, including an accelerometer X plot 310, an accelerometer Y plot 320, and an accelerometer Z plot 330. According to an example embodiment of this disclosure, each plot provides data for creating a mapping based on accelerometer signals recorded simultaneously (or nearly simultaneously). For example, the mapping may include a regression model that maps directly from accelerometer signals (e.g., data signals, data inputs, etc.) to manipulated stimulus parameters. Manipulated stimulus parameters may include one or more parameters, such as amplitude, pulse width, frequency, etc.

[0111] Within the fields of machine learning and classical statistics, there are various mathematical, algorithmic, and computational techniques that can be used to generate mappings from accelerometer signals to stimulus parameters, and to recalibrate those mappings, for purposes such as regression, classification, prediction, pattern recognition, and similar tasks. For example, linear regression is a fundamental yet ubiquitous approach for modeling the relationship between a dependent variable and one or more explanatory variables in the form of a linear function. In other examples, support vector machines (SVMs) constitute a supervised learning method that constructs optimal hyperplanes for robust classification and regression tasks. Neural networks can encompass interconnected systems of nodes inspired by brain biology and capable of extracting complex patterns from data through sophisticated learning protocols. Autoregressive models employ regression equations where lagged values ​​of time series are used to predict future observations. The name autoregressive indicates that the prior values ​​of the variable itself are used as predictive covariates. In other examples, moving average models apply a mathematical average across a fixed window of previous time periods to smooth short-term fluctuations and reveal long-term trends. Based on the example embodiments presented herein, these (and others) paradigms provide tools for extracting insights from data spanning a wide range of application contexts to create direct or indirect mappings from accelerometer signals to raw, manipulated, or semi-manipulated stimulus parameters or other data.

[0112] In some examples, this confirmation can be made through subjective measurements (e.g., asking the patient whether the level of perception is too strong or too weak at any point, patient input, etc.). In other examples, the confirmation can be made based on objective measurements (e.g., based on physical recordings, patient status, recorded data, etc.). In additional examples, a combination of subjective and objective measurements can be used to confirm a consistent (or nearly consistent) level of perception. Once the level of perception is confirmed to be consistent, the system can generate (e.g., create) a mapping (e.g., a regression model) that uses one or more accelerometer signals as input and employs, for example, a linear model to predict the stimulus output in that situation.

[0113] like Figure 3 As depicted, arrow 336 continues to Figure 4An example embodiment receives accelerometer plot data at arrow 406 and creates a current amplitude plot 400.

[0114] Figure 4 A current amplitude graph 400 according to an example embodiment is shown, which defines a subsensory scaling factor that decays the predicted amplitude from the regression model (via data from arrow 406) to an appropriate subsensory dose to be applied to a patient's therapy.

[0115] Example embodiments of the system disclosed herein may adjust the stimulation amplitude and / or current 402, as expressed on a current amplitude graph 400 over time 404 (in seconds). In this case, the system may be configured to maintain the ECAP level at a consistent or near-constant level, providing a more consistent level of neural activation in the patient (e.g., a consistent level of therapy for the patient). In some examples, stimulation at a consistent level of perception is provided to the patient when the patient feels the stimulus (e.g., at or above the perceived level) to provide improved or desired therapy.

[0116] For example, current amplitude graph 400 can define a sub-sensory scaling factor that decays the predicted amplitude from the regression model to an appropriate sub-sensory dose. In some examples, data is captured continuously, semi-continuously, or in various ways from accelerometer signals captured by an accelerometer worn by the patient while running in the background. In additional example embodiments, such as in the IPG or ETS embodiments of the stimulation device, the accelerometer may be on a chip or other circuit system that can provide all or some of the measurements (e.g., three, five, etc.) for accelerometer sensing. One or more additional sensors (e.g., chips, circuit systems, etc.) are provided for ECAP sensing, which differs from accelerometer sensing. Once a consistent level of perception is confirmed (or approximately confirmed), the example embodiments are able to control the patient's level of perception or therapy (e.g., particularly in the sub-sensory example) and / or the level of neural activation in the patient via a prescribed therapeutic dose.

[0117] Hundreds of data points (which may range from tens to thousands depending on the feature map example) provide data points starting from position 208 within a time period of approximately 200 seconds (about 3.33 minutes) of a day. However, those skilled in the art will understand that data can be collected at any time period, whether continuous, semi-continuous, distributed, or spanning minutes, hours, days, months, etc. In the current amplitude feature map 200, after position 408, the patient positions movements in various postures within the time period, which are normalized according to a scaling factor. For example, from position 408, the patient moves to a seated position 410 (first time) for approximately 30 seconds until time t1 412. After time t1 412, the patient moves to a seated position (with neck tilted back) 414 for approximately 10 seconds until time t2 416. After time t2 416, the patient moves to a standing or rising position 418 for approximately 40 seconds until time t3 420. After time t3 420, the patient moved to a standing position 422 for approximately 30 seconds until time t4 424. After time t4 424, the patient moved to a reclining position (or was in a reclining position) 426 for approximately 40 seconds until time t5 428, at which point the patient moved back to a sitting position (for the second time) 430. Figure 2 , Figure 3 and Figure 4 The described and depicted motions are derived from the same dataset that records all three streams. Those skilled in the art will understand that additional datasets, data, and / or motions may be recorded or otherwise used to identify the same or different motions.

[0118] Figure 5 A series of accelerometer graphs 500 based on accelerometer data 510 / 520 / 530 according to an example embodiment of the present disclosure are shown. The accelerometer data is based on the patient’s motion and is correlated with and mapped to the stimulus amplitude based on a derived regression model.

[0119] The exemplary embodiments of this disclosure capture some or all of the ECAP factors (such as...). Figure 2 As shown), accelerometer signals recorded simultaneously on multiple axes (such as...) Figure 3 (as shown) and stimulation amplitudes based on closed-loop amplitude adjustment to maintain neural activation based on recorded ECAP features (e.g.) Figure 4The process (shown) is used to provide parameterization for subsensory therapy. ECAP features can be one or more of the following: range (e.g., maximum to minimum amplitude), area under the curve, curve length, N1 or trough time, P2 or peak time, conduction velocity, or others. Once ECAP features are recorded, the system adjusts stimulation parameters based on the recorded ECAP features to maintain consistent neural activation. For example, consistent neural activation may require maintaining one or more ECAP features (e.g., range, area under the curve, etc.) at consistent levels. In other examples, the system monitors and records consistent neural activation that follows features related to the distance between the stimulating electrode and the spinal cord, as well as features associated with a subsensory closed-loop algorithm generated to maintain the same level of neural activation.

[0120] In some configurations, features or combinations of features can be specifically selected to correlate with certain spinal cord physiological parameters, such as the distance between the stimulating electrode and the spinal cord. The closed-loop amplitude adjustment algorithm based on ECAP features can also have hierarchical logical embeddings based on multiple measured features to determine whether physiological parameters, such as the distance between the stimulating electrode and the spinal cord, may have changed. Based on some or all of this initially captured data, the system configures a model between feature map 200, a series of accelerometer maps 300, and current amplitude maps 400. Once the model is configured, the system scales down the configured model by a sub-sensory percentage or factor, and the algorithm executes at that level. In such an example embodiment, the algorithm adjusts only based on the accelerometer signal being recorded, and based on this, new recordings are predicted, and the stimulation level is scaled. For example, it can be applied to a sub-sensory procedure at a dose equal to 50% of the sensing threshold to apply the stimulation amplitude to the patient's therapy. For example, a stimulation current amplitude map 600 is generated to provide feedback and updated therapy to the patient, as well as feedback data provided via closed-loop control (in combination with...). Figure 6 (Detailed description and depiction).

[0121] In additional example embodiments, the accelerometer may respond slowly to changes, and consequently, the patient may perceive a change in their level of perception as the subsensory therapy level increases (e.g., reaches or exceeds the perception level). In such an example, the system may provide messages (e.g., alerts, notifications, instructions, etc.) to inform the patient or other users that their perception level may be changing or feeling different. In such an example, the system may utilize (e.g., adopt, rely on, etc.) one or more ECAP signals to a greater extent in the model (compared to accelerometer recordings) to adjust the levels or parameters used for subsensory therapy and / or multisensory sensory abnormality therapy.

[0122] Figure 6An example stimulus amplitude graph 600 according to one embodiment is shown, which displays a stimulus amplitude curve based on a mapping used to parameterize a closed-loop algorithm for subsensory therapy.

[0123] According to some exemplary embodiments, the method includes: directly mapping from one or more sensors to one or more stimulus parameters. Example methods demonstrate consistent neural activation via closed-loop control at stimulus levels where an evoked response exists (potentially above the perceived level). Such methods can create a mapping (such as a regression model) directly from accelerometer signals to manipulated stimulus parameters (in this case, such as amplitude), but such parameters can include, for example, pulse width, frequency, or similar parameters. This mapping can be generated, for example, using methods such as linear regression, SVM, neural networks, autoregression, moving averages, etc. Such as combining... Figure 3 The graphs of a series of accelerometer charts 300 described and depicted show the accelerometer signals recorded simultaneously, which can be analyzed and projected to identify the subsensory scaling factor.

[0124] After the accelerometer signal is analyzed and projected to identify the subsensory scaling factor, a current amplitude plot can be run or generated to define the subsensory scaling factor, which will predict the amplitude decay from the regression model to an appropriate subsensory input (e.g., a therapeutic dose). For example, the current amplitude plot can show closed-loop amplitude adjustment to maintain neural activation in the patient during various activities (e.g., movement, orientation changes, etc.) occurring over a period of time. Amplitude adjustment can provide the stimulus amplitude to be provided to the user, which includes dose parameters such as current 602 set to a subsensory level within time period 604 based on accelerometer data recorded from the patient.

[0125] In some examples of creating mappings, the examples may switch (e.g., slide, move, rotate, etc.) data along a window as the system generates and / or uses the generated mapping. For example, on a sliding scale of time (e.g., approximately one second of data), at each time marker, the system may implement or execute an update to the signal level to form (e.g., calculate) an average that can be used as an input (e.g., as an input to a regression model). For example, this update may be based on the average (or other median, etc.) of each of the previous (e.g., the last) X seconds of data, which can be used as the time window employed by a parameterized closed-loop algorithm. For example, the step size of the algorithm may be determined based on the previous (e.g., the last) X seconds of data to determine how to change the stimulus output. For example, the X seconds of data could be half a second, 1 second, 5 seconds, 10 seconds, or other step sizes or average step sizes (or orders of magnitude) for the data used in the window. The step size or average step size can then be fine-tuned and / or adjusted around the patient's movement needs profile, pain therapy expectations, patient status, etc.

[0126] In the accompanying example systems, methods, machine-readable media, and the like, visualization of data related to both ECAP recordings and accelerometer readings is provided. In extreme (or rare) circumstances, a patient may be in a physical location that affects ECAP recordings and / or accelerometer readings. Examples of factors affecting ECAP recordings may include passing through airport or retail security scanners, being near medical imaging equipment, encountering fault conditions in the ECAP sensing circuitry that prevent proper ECAP recording, or similar situations. Examples of factors affecting accelerometer data may include whether the patient is underwater (e.g., below one atmosphere), outside Earth orbit, in a car that is accelerating or decelerating, in a flying aircraft, or in other locations that may affect the accelerometer device. In such example embodiments, a user interface may be provided to the patient to provide input data (e.g., text, multiple selections, etc.) indicating information that may cause the accelerometer device to provide different or anomalous data.

[0127] Figure 7 A flowchart is shown, illustrating an example of a process flow 700 (e.g., a processing method) for parameterizing a closed-loop algorithm for subsensory therapy according to an exemplary embodiment of the present disclosure. Embodiments of process flow 700 may be implemented by a system or device for analyzing, monitoring, and correlating health-related data for neurostimulation programming, such as evoked compound action potentials (ECAPs) and accelerometer data used for monitoring and recording closed-loop neurostimulation therapy and for training and predicting machine learning models. For example, process flow 700 may be embodied by electronic operations performed by one or more computing systems or devices (including computing systems or devices at a network-accessible remote service) specifically programmed to perform data monitoring, data analysis, and / or neurostimulation data processing operations, and to provide treatment and dosage recommendations and changes (e.g., modifications) to a patient system. In certain examples, the operation of process flow 700 may be implemented at a single entity or at multiple locations via the systems and data flows described throughout this disclosure. For example, process flow 700 may be performed by monitoring and processing system 130.

[0128] In box 702, routine 700 delivers neural stimulation. In box 704, routine 700 senses neural signals indicating neural responses, each a response to the delivery of neural stimulation. In box 706, routine 700 uses multiple stimulation parameters to control the delivery of neural stimulation. In box 708, routine 700 records evoked compound action potentials (ECAPs) received from sensing circuitry. In box 710, routine 700 records accelerometer data. In box 712, routine 700 generates a regression model for mapping the accelerometer data to optimal stimulation parameters to achieve consistent neural activation, enabling closed-loop control of subsensory therapy. In box 714, routine 700 applies the regression model and subsensory dose levels to multiple subsensory therapy procedures. In box 716, routine 700 deploys the regression model to the patient via stimulation output circuitry to deliver neural stimulation.

[0129] Example embodiments of this disclosure include the use of accelerometers in conjunction with recorded neural signals to enhance control algorithms for spinal cord stimulator (SCS) therapy or other stimulation therapies. Implantable devices (such as IPGs, sensors, or other devices) can be configured to sense neural activity, such as evoked compound action potentials (ECAPs), and to identify neural activation indices (NAIs), which may include scores or quantities to describe how close the system is to achieving optimal or near-optimal neural activation consistency for a given patient. Example embodiments allow for ECAP-based closed-loop algorithms for all therapeutic modalities, regardless of the presence of ECAP at therapeutic doses.

[0130] In a first example embodiment, for a clinical workflow, the mapping from the sensor directly to one or more stimulation parameters in a sensory abnormality procedure includes a constant neural activation map, which is used to fit an algorithm to identify the stimulation amplitude by applying a sub-sensory procedure dose equal to, for example, 50% of the perception threshold. For example, an example clinical workflow may include collecting data (e.g., ECAP data and accelerometer data) from closed-loop sensory abnormalities to confirm consistent perception. In such an example embodiment, the clinical workflow includes omitting actions that might cause changes in perception but for which accelerometer recordings show no significant changes (e.g., neck movements, coughing, sneezing, etc.). An example clinical workflow may include (if possible) a comprehensive set of movements such that the accelerometer mapping is well-defined for all patient postures (e.g., standing, sitting, leaning back, supine, left lateral decubitus, right lateral decubitus, forward tilt, etc.). The workflow continues to apply the mapping to all sub-sensory procedures for closed loops (e.g., without the need for additional closed-loop sensory abnormalities) as desired by the clinician programmer, and enables input of the desired dose for each procedure. An example clinical workflow also includes writing accelerometer-to-mapping and dose level recommendations (e.g., therapy changes) into IPG firmware or the like.

[0131] In example embodiments that monitor subsensory aberration patterns (e.g., rapid patterns) in patient therapy, different sensors may be present to measure what the patient feels and / or interprets (e.g., how the patient interprets the subsensory aberration pattern). For example, when a subsensory aberration pattern (e.g., a subsensory perception procedure) is applied to a patient, the patient primarily intends not to interpret the applied stimulus (e.g., the user typically does not feel the applied therapy or procedure). However, in some cases, the patient may feel (or interpret) the subsensory aberration pattern therapy by, for example, increasing the amplitude, rate, frequency, etc. of the subsensory aberration. In this case, in a clinical workflow example, the system applies a mapping to one or more subsensory perception procedures, and a user (such as a clinician or software model) can confirm whether the patient feels (e.g., interprets) the subsensory aberration pattern, therapy, or procedure. Further example embodiments of this disclosure may provide suggested adjustments, alterations, variations, etc., to be implemented on the mapping according to subsensory-based or perception-based programming.

[0132] In an additional example embodiment, similar to the clinical workflow example embodiment, the system includes a mapping from sensors directly to stimulus parameters that can be applied in a home setting (e.g., without requiring clinician or physician input or programming). For example, via a mobile application or remote control for receiving patient input, the application can collect data such as recorded ECAP factors and recorded accelerometer data from closed-loop sensory abnormality monitoring. The home workflow can confirm consistent or near-consistent perceptions from the patient and prompt the patient to move in various positions or postures, such as sitting upright, tilting, bending, lateral twisting, bathing movements, dressing movements, sleep movements, or similar actions. The system is configured to monitor and flag or omit actions that may lead to changes in perception but not significant changes in accelerometer motion (e.g., neck movements, coughing, etc.). The example home workflow also includes (where possible) one or more comprehensive sets of movements that provide accelerometer mappings for all postures (e.g., standing, sitting upright, leaning back, supine, left lateral decubitus, right lateral decubitus, forward tilt, etc.), some or all of which can be well-defined based on recorded patient data, exemplary data, hypothetical data, computer-rendered example data, or combinations thereof. The example workflow stores the collected data in, for example, ferroelectric RAM (FeRAM or FRAM), a random access memory structurally similar to DRAM, but using ferroelectric layers instead of dielectric layers to achieve non-volatility. The example home workflow can perform model training on the IPG (if computationally, resource-wise, and / or power-wise inexpensive), or, if more computational power is required, stream data from the IPG to mobile applications and / or computing systems for training and predictive modeling. The home workflow can apply mappings to all sub-sensory procedures within the closed loop desired by the programmer, with input for the desired dose applied at each patient level for each procedure. The home workflow can also write accelerometer mapping data and dose level data to IPG firmware, such as SRAM.

[0133] In an additional exemplary embodiment, the system is configured to enhance the ECAP control algorithm using accelerometer inputs. For example, the accelerometer mapping to stimulus control can be trained to have utility beyond sub-sensory perception. For example, for a sensory abnormality closed-loop (CL) application, the accelerometer mapping produces another possible control action in addition to the ECAP feedback adjustment rule. A weighted combination of algorithm outputs can deliver optimal (e.g., best, optimized, encouraging, desired, etc.) results:

[0134] CL-outputtotal= α*CL-outputECAP+ β*CL-outputaccelerometer

[0135] Here, α and β represent the relative weights given to the ECAP algorithm and the accelerometer calculation method, respectively (the sum of the two is 1). For combined therapies, such as CL applications, the available frequency of the ECAP signal may be lower than during the sensory abnormality CL period (due to other procedures running, possible arbitration considerations, etc.). Therefore, the accelerometer mapping and / or output can be given greater weight in the above equation to compensate for the less information from the ECAP signal. Additional utility includes anomaly detection: if the CL-output ECAP and CL-output accelerometer are substantially different over an extended period of time (e.g., several minutes or longer), anomaly detection is likely to determine that the IPG pocket or lead position / impedance has changed significantly. In such anomaly detection scenarios, the patient or user can be prompted to re-engineer the accelerometer mapping by following an example home (e.g., mobile) workflow. The anomaly detector can be configured to identify divergent outputs and take into account changes in values.

[0136] In another example embodiment, a performance metric for closed-loop algorithms is provided, including accelerometer data based on ECAP control. In such example embodiments, according to some example embodiments, the metric can be used to quantify therapy delivery and algorithm performance in closed-loop applications. The Neural Activation Index (NAI) can include a description of how close the system is to achieving optimal neural activation consistency. NAI = 1 means perfectly consistent neural activation (optimal). NAI = 0 means completely inconsistent neural activation (worst-case, least optimized).

[0137] For example, the display controlled by ECAP CL can be derived from various data, such as:

[0138] ParesthesiaNAI = Variance around static posture, variance around static stimulus amplitude / CL set point (paresthesia CL), or

[0139] Combo / Sub-P NAI = Paresthesia NAI * Correlation (Combined / Sub-P mapping, sensory abnormality mapping).

[0140] In additional example embodiments, the migration of the IPG pouch or lead may include remapping and / or reprogramming. Patients with limited mobility may have difficulty assisting in recording all necessary posture data (e.g., various postures required or preferred for programming examples of the algorithms used by the system). User input can be used to identify various postures that the patient cannot or has difficulty assuming (e.g., a difficulty scale of 0 to 10 for a given posture).

[0141] Although Figure 7The ECAP factors (e.g., features) and neural response parameters shown are discussed as examples, but other ECAP and neural response parameters may be suitable and used in this subject matter, as determined by those skilled in the art, such as dynamic curve length, time-domain parameters, spectral power parameters, or similar parameters.

[0142] Figure 8 Embodiments of an implantable device 801-N and a lead system 805, such as those that can be implemented in a neuromodulation system, according to some exemplary embodiments of this disclosure are illustrated by way of example. The implantable device 802 can represent, as in combination with... Figure 12 Examples of the programming device 1202 and / or stimulation device 1204 described and depicted.

[0143] The implantable device 802 may include a controller 806 for controlling various functions of the implantable device. The implantable device 802 may be configured to sense ECAP and automatically select and maintain ECAP sensing electrodes for sensing ECAP over an extended time period.

[0144] The implantable device 801-N may include a stimulation output circuit 807, and the controller 806 may include a stimulation control circuit 808 configured to control the stimulation output circuit 807. The stimulation output circuit 807 can generate and deliver a neuromodulation waveform. Such a waveform may include different waveform shapes. The waveform shape may include regular shapes (e.g., square waves, sine waves, triangle waves, sawtooth waves, etc.) or irregular shapes. The stimulation control circuit 808 can control which electrodes are used to deliver stimulation and can control the delivery of the neuromodulation waveform using multiple stimulation parameters specifying the pattern of the neuromodulation waveform. The lead system 805 may include one or more leads (each configured to be electrically connected to the stimulation device) and multiple electrodes distributed in the one or more leads.

[0145] In the example, the number of leads and / or the number of electrodes on each lead depends on, for example, the distribution of targets for neuromodulation and the need to control the distribution of the electric field at each target. In the example, the lead system includes, for example, 19 leads, each with 8 electrodes. Multiple electrodes may include electrode 801-1, electrode 801-2, electrode 801-3, ..., and electrode 801-N. An implantable device 802 (such as an IPG or ETS) can individually select the electrodes to be activated to provide an electrical interface between the stimulation output circuit 807 and the patient tissue. Neuromodulation waveforms can be delivered from the stimulation output circuit 807 through the set of activated electrodes selected from electrodes 801-1 to 801-N.

[0146] Implantable device 801-N may include ECAP sensing circuitry 809, and controller 806 may include ECAP sensing control circuitry 810 configured to control ECAP sensing circuitry 809. Implantable device 802 may be configured to individually select electrodes for sensing electrical activity in neural tissue. ECAP sensing circuitry may include amplifiers and filters for detecting ECAP in neural tissue. ECAP sensing control circuitry 810 may control the electrodes used for sensing ECAP and may also be configured to perform an evaluation of sensing electrodes in lead system 805, as described in more detail below.

[0147] The controller 806 may also include an ECAP analyzer 811 configured to evaluate the detected ECAP. By way of example, and not limitation, various features in the detected ECAP can be used to control therapy and / or monitor the efficacy of therapy. By way of example, and not limitation, the subject matter may use characteristics of the sensed ECAP, such as low amplitude or significant changes in the amplitude of the detected ECAP compared to a threshold or trend, to trigger evaluation of a sensing electrode. By way of example, and not limitation, the implantable device 802 may include a scheduler 812, which may or may not be part of the controller 806 to provide programmed scheduling for triggering evaluation of a sensing electrode. By way of example, and not limitation, the implantable device 802 may include additional sensing circuitry 813 that interfaces with other sensors 814. By way of example, these sensors can be used to control therapy and / or monitor the efficacy of therapy.

[0148] These sensors can be used to trigger the evaluation of sensing electrodes. For example, the sensors can be used to detect significant changes in patient activity, movement, or posture, and the evaluation of sensing electrodes can be triggered at least in part based on the detected activity, movement, or posture. For sensors used to monitor the efficacy of a therapy, the system can be configured to trigger the evaluation of sensing electrodes when the efficacy of the monitored therapy is worse than expected or shows a declining trend. The implantable device may include a power source 815, such as a rechargeable battery or a passive energy source configured to receive power from an external device, and may also include features for connecting to external devices (such as...). Figure 12 Telemetry 816 communicates with the stimulation device 1204 (described and depicted).

[0149] Figure 9Block diagram 900 is shown by way of example, not limitation, and includes a patient system 902 for monitoring parametric techniques for subsensory therapy and multisensor sensory abnormality therapy, and is configured to identify patient problems from health-related information that may adversely affect the patient's treatment plan. The monitor can collect various types of vital data 950 related to chronic pain, which may include, for example, objective data 904 and / or subjective data 952.

[0150] Objective data are data that can be observed using human senses and can be obtained from measurement or direct observation. Objective data can be measured by sensors and can be provided via user input when the user has access to objectively determined information. Examples of objective data may include physiological parameter data 953, therapy data 954, device data 955, environmental data 956, and / or additional objective data 972 (e.g., health-related data). By way of example, and not limitation, physiological parameter data 953 may include data such as: heart rate, blood pressure, respiratory rate, activity, posture, electromyography (EMG), neural responses such as evoked compound action potentials (ECAP), glucose measurements, oxygen levels, body temperature, oxygen saturation, and gait. By way of example, and not limitation, therapy data 954 may include: neuromodulation procedures, therapy on / off schedules, dosage, neuromodulation parameters (such as waveform, frequency, amplitude, pulse width, and period), therapy usage, and therapy type. By way of example, and not limitation, device data 955 may include: battery information (voltage, state of charge, charging history (if rechargeable)), impedance data, faults, device model, lead type, MRI status, Bluetooth connection logs, and connection history with the clinician programmer (CP). By way of example, and not limitation, environmental data 956 may include: temperature, air quality, air pressure, location, altitude, sunny / cloudy, precipitation, etc. By way of example, supplementary data 958 may include: healthcare-related data (e.g., menstrual cycle, surgery, treatment process, acute illness, other non-painful chronic conditions, etc.), lifestyle-related data (e.g., interpersonal relationship problems, financial problems, family stressors, work stressors, etc.), and similar data.

[0151] Subjective data can include information received from the patient. For example, a patient's quantification of pain is subjective data. Subjective data typically involves data input by the user. Examples of subjective data 952 include questions 957 with free-text answers, multiple-choice questions 958, question trees 959, and / or additional subjective data 970. Other data can be stored and / or transmitted, including detected events 960, contextual data 963 (e.g., context 961) for other collected data and / or events, and clocks 962 (e.g., time) (such as timestamps that can be used to provide association with retrieved data). Events 960, context 961, and time can be detected by the system or provided via user input.

[0152] The collected data can be processed, as typically illustrated at data processing 963. Data processing 963 can occur in a medical device or patient device (such as a mobile phone, tablet, or remote control), or in a remote data receiving system. Data processing 963 can include one or more models 964. Model 964 can be used to determine how to use patient data to identify problems (e.g., “common problems”) for which automated triage alerts 969 can benefit users (e.g., patients, clinicians, device representatives, patient caregivers, etc.). Model 964 can be used to determine how to use patient data to classify subsensory and multisensory sensory aberration therapies (e.g., disease progression, updated patient goals, changes in status, etc.) or patient interactions, determine the type of intervention and contact, identify problems for which automated patient assistance is beneficial (e.g., “common problems”), determine the content used for patient assistance, determine whether automated triage alerts are beneficial, determine whether treatment plan updates are beneficial, etc. Model 964 can use the acquired data to deliver therapy. Machine learning 965 (or other artificial intelligence) can be implemented on the collected data to develop or refine model 964. Data processing 963 can include data imputation, such as that which can be used to prevent missing data from introducing bias into models or machine learning.

[0153] In a broad sense, machine learning can involve using computer algorithms to automatically learn patterns and relationships in data, potentially without explicit programming. Machine learning algorithms can be divided into three main categories: supervised learning, unsupervised learning, and reinforcement learning.

[0154] For example, supervised learning involves training a model using labeled data to predict outputs for new, unseen inputs. Examples of supervised learning algorithms include linear regression, decision trees, and neural networks. Unsupervised learning involves training a model on unlabeled data to discover hidden patterns and relationships within the data. Examples of unsupervised learning algorithms include clustering, principal component analysis, and generative models such as autoencoders. Reinforcement learning involves training a model to make decisions in a dynamic environment by receiving feedback in the form of rewards or penalties. Examples of reinforcement learning algorithms include Q-learning and policy gradient methods. Based on some examples, examples of specific machine learning algorithms that can be deployed include logistic regression, a supervised learning algorithm for binary classification tasks. Logistic regression models the probability of a binary response variable against one or more predictor variables. Another example type of machine learning algorithm is Naïve Bayes, another supervised learning algorithm for classification tasks. Naïve Bayes is based on Bayes' theorem and assumes that the predictor variables are independent of each other. Random forests are another supervised learning algorithm for classification, regression, and other tasks. Random forests build an ensemble of decision trees and combine their outputs to make predictions.

[0155] Further examples include neural networks, which consist of interconnected layers of nodes (or neurons) that process information and make predictions based on input data. Matrix factorization is another machine learning algorithm used in recommender systems and other tasks. Matrix factorization breaks a matrix down into two or more matrices to reveal hidden patterns or relationships in the data. Support Vector Machines (SVMs) are a supervised learning algorithm used for classification, regression, and other tasks. SVMs search for hyperplanes that separate different classes in the data. Other types of machine learning algorithms include decision trees, k-nearest neighbors, clustering algorithms, and deep learning algorithms such as convolutional neural networks (CNNs), recurrent neural networks (RNNs), and transformer models. The choice of algorithm depends on the nature of the data, the complexity of the problem, and the performance requirements of the application. The performance of machine learning models is typically evaluated on a separate test dataset that has not been used during training to ensure that the model can generalize to new, unseen data.

[0156] While this article discusses several specific examples of machine learning algorithms, the principles discussed can also be applied to other machine learning algorithms. Deep learning algorithms (such as convolutional neural networks, recurrent neural networks, and transformers) as well as more traditional machine learning algorithms (such as decision trees, random forests, and gradient boosting) can be used in a wide variety of machine learning applications.

[0157] Two typical types of problems in machine learning are classification problems and regression problems. Classification problems, also known as categorization problems, aim to classify items into one of several class values ​​(e.g., is this object an apple or an orange?). Regression algorithms aim to quantify some item (e.g., by providing a real numerical value).

[0158] Moving toward the training phase in machine learning algorithms and models, generating a trained machine learning program can include multiple phases forming part of a machine learning pipeline, including, for example, the following phases: data collection and preprocessing, feature engineering, model selection and training, model evaluation, prediction, validation, refinement, retraining, deployment, or combinations thereof. For example, data collection and preprocessing can include phases for acquiring and cleaning data to ensure it is suitable for use in a machine learning model. This phase may also include removing duplicates, handling missing values, and converting the data to a suitable format. Feature engineering can include phases for selecting and transforming training data to create features useful for predicting a target variable. Feature engineering can include (1) receiving features (e.g., structured or labeled data in supervised learning) and / or (2) identifying features in the training data (e.g., unstructured or unlabeled data for supervised learning). Model selection and training can include phases for selecting an appropriate machine learning algorithm and training it on preprocessed data. This phase may also involve decomposing the data into training and test sets, evaluating the model using cross-validation, and fine-tuning hyperparameters to improve performance.

[0159] In additional examples, model evaluation may include a phase for evaluating the performance of a trained model (e.g., a trained machine learning program) on a separate test dataset. This phase can help determine whether the model is overfitting or underfitting and whether it is suitable for deployment. Prediction may include a phase for generating predictions on new, unseen data using the trained model (e.g., a trained machine learning program). Validation, refinement, or retraining may include a phase for updating the model based on feedback generated from the prediction phase, such as new data or user feedback. Deployment may include a phase for integrating the trained model (e.g., a trained machine learning program) into a wider range of systems or applications, such as web services, mobile applications, or IoT devices. This phase may include setting up the API, building the user interface, and ensuring the model is scalable and capable of handling large amounts of data.

[0160] Example phases of machine learning may also include a training phase (e.g., a portion of model selection and training) and a prediction phase (a portion of prediction). Prior to the training phase, feature engineering is used to identify features. This can include identifying informative, discriminative, and independent features used to effectively operate a trained machine learning program in pattern recognition, classification, and regression. In some examples, training data includes pre-identified features and labeled data known about one or more outcomes. Each feature can be a variable or attribute, such as an individually measurable property of a process, item, system, or phenomenon represented by a dataset (e.g., training data). Features can also be of different types, such as numerical features, strings, and graphs, and can include one or more of content, concepts, attributes, historical data, and / or user data, to name just a few.

[0161] During the training phase, the machine learning pipeline uses training data to find correlations between features that influence predictions or predict / infer data. Using the training data and labeled features, the trained machine learning program is trained during the training phase of the machine learning program training process. The machine learning program is trained to evaluate the values ​​of features as they relate to the training data. The result of the training is the trained machine learning program (e.g., a trained or learned model). Furthermore, the training phase can involve machine learning where the training data is structured (e.g., labeled during preprocessing operations). The trained machine learning program implements a neural network capable of performing operations such as classification and clustering. In other examples, the training phase can involve deep learning where the training data is unstructured, and the trained machine learning program implements a deep neural network capable of performing both feature extraction and classification / clustering operations.

[0162] In some examples, neural networks can be generated during the training phase and implemented within a trained machine learning program. A neural network comprises a hierarchical (e.g., layered) organization of neurons, where each layer consists of multiple neurons or nodes. Neurons in the input layer receive input data, while neurons in the output layer produce the network's final output. Between the input and output layers, there may be one or more hidden layers, each consisting of multiple neurons. Each neuron in the neural network operationally computes a function, such as an activation function, which takes as input a weighted sum of the outputs of neurons in the previous layer and a bias term. The output of this function is then passed as input to neurons in the next layer. If the output of the activation function exceeds a certain threshold, the output is passed from that neuron (e.g., a transmitting neuron) to connected neurons (e.g., receiving neurons) in subsequent layers. Connections between neurons have associated weights that define the influence of the input from the transmitting neuron to the receiving neuron. During the training phase, these weights are adjusted by a learning algorithm to optimize the network's performance. Different types of neural networks may use different activation functions and learning algorithms, thus affecting their performance on different tasks. The hierarchical organization of neurons, along with the use of activation functions and weights, enables neural networks to model complex relationships between inputs and outputs and to generalize to new inputs not seen during training.

[0163] In some examples, a neural network can also be one of several different types of neural networks, such as a single-layer feedforward network, a multilayer perceptron (MLP), an artificial neural network (ANN), a recurrent neural network (RNN), a long short-term memory network (LSTM), a bidirectional neural network, a symmetric connection neural network, a deep belief network (DBN), a convolutional neural network (CNN), a generative adversarial network (GAN), an autoencoder neural network (AE), a restricted Boltzmann machine (RBM), a Hopfield network, a self-organizing map (SOM), a radial basis function network (RBFN), a spike neural network (SNN), a liquid state machine (LSM), an echo state network (ESN), a neural Turing machine (NTM), or a transformer network, to name just a few.

[0164] In addition to the training phase, the validation phase can be performed on a separate dataset called the validation dataset. The validation dataset is used to fine-tune the model's hyperparameters, such as the learning rate and regularization parameters. The hyperparameters are tuned to improve the model's performance on the validation dataset. Once the model is fully trained and validated, in the testing phase, the model can be tested on a new dataset. The testing dataset is used to evaluate the model's performance and to ensure that the model has not overfitted the training data.

[0165] During the prediction phase, the trained machine learning program uses features used to analyze the query data to generate inferences, results, or predictions as examples of prediction / inference data. For instance, during the prediction phase, the trained machine learning program generates output. The query data is provided as input to the trained machine learning program, and the trained machine learning program generates prediction / inference data as output in response to the receipt of the query data.

[0166] Figure 10 A block diagram 1000 of an embodiment of system 1001 (e.g., a computing system) is shown by way of example. System 1001 implements a neurostimulation programming circuit system 1206 such that an implantable electroneurostimulation device is programmed to achieve therapeutic goals in human subjects based on a trained closed-loop programming model as discussed herein.

[0167] System 1001 can be operated by clinicians, patients, caregivers, medical facilities, research institutions, medical device manufacturers, or distributors, and is embodied in multiple different computing platforms. System 1001 can be a remote control device, a patient programmer device, a program modeling system, or other external device, including a supervised device for directly implementing programming commands and modifications using a neurostimulation device. In some examples, system 1001 can be a networked device connected to a computing system via a network (or a combination of networks) that operates a user interface computing system using communication interface 1008. This network can include local, short-range, or long-range networks, such as Bluetooth, cellular, IEEE 802.11 (Wi-Fi), or other wired or wireless networks.

[0168] System 1001 includes a processor 1002 and a memory 1004, which may optionally be included as part of a neurostimulation programming circuit system 1006. Processor 1002 may be any single processor or a group of processors working in concert. Memory 1004 may be any type of memory, including volatile or non-volatile memory. Memory 1004 may include instructions that, when executed by processor 1002, cause processor 1002 to implement the characteristics of the stimulation programming circuit system 1006 (e.g., a neurostimulation circuit system). Therefore, electronic operations in system 1200 can be performed by either processor 1002 or stimulation programming circuit system 1006.

[0169] The processor 1002 or circuit system 1006 can directly or indirectly perform neurostimulation operations, including the use of programming the neurostimulation device based on a trained programming model. The processor 1002 or circuit system 1006 can also provide data and commands to assist in the processing and implementation of programming using the communication interface 1008 or the stimulation device interface 1010 (e.g., a neurostimulation interface). It will be understood that the processor 1002 or circuit system 1006 can also implement other aspects of the device data processing or device programming functions described above.

[0170] Figure 11 A block diagram 1100 of an embodiment of system 1101 (e.g., a computing system) is shown by way of example. System 1101 is used to perform analysis of patient feedback data (e.g., patient input data collected as training data in user stimulation device interface 1110) in conjunction with the data processing operations discussed above.

[0171] System 1101 can be integrated or coupled to computing devices, remote control devices, patient programmer devices, clinician programmer devices, program modeling systems, or other external devices deployed for neurostimulation therapy. In some examples, system 1101 can be a networked device (server) connected via a network (or a combination of networks) that communicates with one or more devices (clients) using a communication interface 1108 (e.g., communication hardware implementing software network interfaces and services). The network can include local, short-range, or remote networks such as Bluetooth, cellular, IEEE 802.11 (Wi-Fi), or other wired or wireless networks.

[0172] System 1101 includes a processor 1102 and a memory 1104, which may optionally be included as part of a stimulus programming circuitry system 1106 (e.g., a user input / output data processing circuitry system). The processor 1102 may be any single processor or a group of processors working in concert. The memory 1104 may be any type of memory, including volatile or non-volatile memory. The memory 1104 may include instructions that, when executed by the processor 1102, cause the processor 1102 to perform data processing or enable other features of the user input / output data processing stimulus programming circuitry system 1106. Therefore, electronic operations in system 1101 can be performed by either the processor 1102 or the circuitry system 1106.

[0173] For example, processor 1102 or circuit system 1106 may implement any of the features of method 800 to acquire and process patient feedback data, generate user interface displays, and provide training data for training the programming selection model. It will be understood that processor 1102 or circuit system 1106 may also implement the logic and processing aspects described above for use in various forms of open-loop, closed-loop, and partially closed-loop device programming or related device actions.

[0174] Figure 12 A block diagram 1200 illustrating an embodiment of a neurostimulation system is shown. Block diagram 1200 includes an electrode 1206, a stimulation device 1204, and a programming device 1202. The electrode 1206 is configured to be placed on or near one or more neural targets in a patient. The stimulation device 1204 is configured to be electrically connected to the electrode 1206 and to deliver neurostimulation energy (such as in the form of electrical pulses) to one or more neural targets via the electrode 1206. The delivery of neurostimulation is controlled using a number of stimulation parameters, such as specifying the pattern of the electrical pulses and the selection of the electrode through which each electrical pulse is delivered.

[0175] In various embodiments, at least some of the multiple stimulation parameters are selected or programmed by a clinical user (such as a physician or other caregiver treating a patient using system 1300); however, some parameters may also be provided in conjunction with closed-loop programming logic and adjustments. Programming device 1202 provides the user with access to implement, change, or modify the programmable parameters. In various embodiments, programming device 1220 is configured to be communicatively coupled to stimulation device 1204 via a wired or wireless link.

[0176] In various embodiments, the programming device 1202 includes a user interface 1210 (e.g., a user interface embodied through graphics, text, voice, or a hardware-based user interface) that allows a user to set and / or adjust values ​​of user-programmable parameters by creating, editing, loading, and removing programs that include combinations of parameters such as patterns and waveforms. These adjustments may also include individually changing and editing values ​​for user-programmable parameters or sets of user-programmable parameters (including values ​​set in response to indications of therapeutic effect). Such waveforms may include, for example, waveforms of patterns of neural stimulation pulses delivered to a patient, as well as individual waveforms that serve as building blocks of patterns of neural stimulation pulses. Examples of such individual waveforms include pulses, groups of pulses, and groups of pulse groups. The program and the corresponding set of parameters may also define electrode selection specific to each individually defined waveform.

[0177] This method also provides examples of an evaluation system 1212, such as a data analysis system, for adapting, modifying, initiating, stopping, monitoring, or identifying neurostimulation therapy performed using the stimulation device 1204. The evaluation system 1212 initiates actions related to neurostimulation therapy based on text analysis performed on input 1220 (e.g., text). This input text 1220 may be collected directly from the patient and analyzed by the evaluation system 1212 to subsequently cause programming effects in the programming device 1202 and the stimulation device 1204, as well as neurostimulation therapy provided by the electrodes 1206.

[0178] Users (e.g., patients, clinicians, device representatives, software, etc.) can provide parameter inputs to the evaluation system 1212. These parameter inputs are used to select, load, modify, implement, measure, analyze, monitor, and / or evaluate one or more parameters of a defined procedure for neurostimulation therapy implemented by the stimulation device 1204, or the operation of the stimulation device 1204. The evaluation can be based on a combination of natural language processing, sentiment analysis, rules, and other identified operational or therapeutic goals. Various logics or algorithms can then determine appropriate actions to take based on the patient's state, including but not limited to: generating improved procedures or parameter changes or recommendations for therapeutic goals (such as pain relief, increased mobility, reduced sleep disruption, etc.); diagnostic or remedial actions against the stimulation device 1304; data logging or alerts to the patient or associated clinicians; and similar actions.

[0179] Examples of parameters that can be implemented by the selected neural stimulation procedure include, but are not limited to, the following: amplitude, pulse width, frequency, duration, total charge injected per unit time, cycle (e.g., on / off time), pulse shape, number of phases, phase sequence, inter-phase time, charge balance, ramp, and spatial variance (e.g., changes in electrode configuration over time). Figure 14 As detailed, the controller (e.g.) Figure 14 The controller 1430 can use a storage device (e.g., Figure 14 The program or settings in the external storage device 1416, or via the program corresponding to the selected program. Figure 14 The external communication device 1418 transmits settings to implement procedures and parameter settings to influence specific neural stimulation waveforms, patterns, or energy outputs. Implementations of such procedures or settings can also define therapeutic intensities and treatment types corresponding to specific pulse groups or groups of pulse groups based on specific procedures or settings. The evaluation system 1212 and the evaluation of inputs 1220 (e.g., text, signals, vision, etc.) provide a mechanism to determine the effectiveness of such procedures or settings, identify problems and provide remedies for ineffective procedures or settings, provide suggestions or recommendations for new or updated procedures and / or settings, or even automatically change procedures or settings.

[0180] The evaluation system 1212, the stimulation device 1204 (e.g., an implantable medical device, an attachable device, etc.), or portions of the programming device 1202 may be implemented using hardware, software, or any combination of hardware and software. Portions of the stimulation device 1204 or the programming device 1202 may be implemented using dedicated circuitry that can be constructed or configured to perform one or more specific functions; or they may be implemented using general-purpose circuitry that can be programmed or otherwise configured to perform one or more specific functions. Such general-purpose circuitry may include a microprocessor or a portion thereof, a microcontroller or a portion thereof, or a programmable logic circuit or a portion thereof. The neuromodulation system shown in block diagram 1200 may also include subcutaneous medical devices (e.g., subcutaneous implantable cardioverter-defibrillators (S-ICDs), subcutaneous diagnostic devices, wearable medical devices (e.g., patch-based sensing devices), or other external medical devices.

[0181] Figure 13 A block diagram 1300 illustrates an example of a patient system 1310 according to some example embodiments and devices that may comprise the patient system. The patient system 1310 may include components that act on a patient (such as delivering a therapy to the patient), components for sensing the patient's condition, state, or environment, and components that enable the patient to interface with the patient system.

[0182] For example, the illustrated patient system 1310 may include a sensor 1301 for sensing patient parameters. Examples of sensors may include, but are not limited to, sensors for neural activity, muscle movement, patient activity, patient posture, patient position, respiration, heart rate, blood pressure, body temperature, analyte sensors, and similar sensors. The illustrated patient system 1310 may include a therapy delivery device (such as a neuromodulator 1302) configured to deliver neuromodulation therapies (such as DBS, SCS, PNS, FES, transcutaneous electrical nerve stimulation (TENS), or other therapies). Those skilled in the art will understand upon reading and understanding this disclosure how this subject matter can be applied to other therapies, such as, but not limited to, cardiac rhythm therapy or drug pump therapy.

[0183] The illustrated patient system 1310 may also include patient devices 1303 for the patient to interface with the patient system. These devices may include sensors (such as accelerometers, temperature sensors, heart rate sensors, etc.) and / or may be used to receive patient feedback (such as responses to questions or free-text responses). These devices may include interfaces for interacting with therapy delivery. Examples of patient devices 1303 include, but are not limited to, patient remote controls 1304, wearable devices (such as watches 1305), or mobile phones or tablets 1306, or other personal devices or additional inputs 1307.

[0184] Figure 14 A block diagram 1400 of a programming system 1402 according to an example embodiment is shown. The programming system 1402 is used as part of an implantable neurostimulation system, such as an external system 1714, wherein the programming system 1402 is configured to send and receive device data (e.g., commands, parameters, program selections, information). Figure 14 An embodiment of a data analysis computing system 1450, which is communication-coupled to a programming system 1402, is also shown, wherein the data analysis computing system 1450 is used to perform data analysis on free-form text and device data (or other types of data) in conjunction with neurostimulation therapy performed by an implantable neurostimulation system.

[0185] Programming system 1402 indicates Figure 12 An embodiment of the programming device 1202 includes an external telemetry circuit 1440, an external storage device 1416, a programming control circuit 1420, a user interface (UI) device 1410, a controller 1430, and an external communication device 1418 to enable programming of a connected neurostimulation device. Operation of the neurostimulation parameter selection circuit 1422 enables the selection, modification, and implementation of a specific set of parameters or settings for neurostimulation programming. The specific set of parameters or settings selected, modified, or implemented can be based on parameters, such as reference... Figures 1 to 7 What is described and depicted.

[0186] External telemetry circuitry 1440 provides the closed-loop programming system 1402 with wireless communication to and from another controllable device (such as an implantable stimulator), including transmitting one or more stimulation parameters (e.g., selected, identified, or modified stimulation parameters of the selected program) to the implantable stimulator via programming data 1560. In one embodiment, external telemetry circuitry 1440 also transmits power to the stimulator via inductive coupling or similar means, such as combining... Figure 15 The implantable stimulator 1521 described and depicted.

[0187] External communication device 1518 may provide a mechanism for communicating with a programming information source (such as a data service, program modeling system) via an external communication link (not shown) to receive program information, settings and values, models, function controls, or the like. In a particular example, external communication device 1518 communicates with data analysis system 1550 (e.g., a computing system) to obtain commands or instructions from data analysis system 1450 in conjunction with parameters or settings selected, modified, or implemented based on free-form text analysis. External communication device 1418 may communicate using any number of wired or wireless communication mechanisms described in this document, including but not limited to IEEE 802.11 (Wi-Fi), Bluetooth, infrared, and similar standardized and proprietary wireless communication implementations. Although external telemetry circuitry 1440 and external communication device 1418 are depicted as separate components within programming system 1402, the functionality of these two components may be integrated into a single communication chipset, circuit system, or device.

[0188] External storage device 1416 stores multiple existing neural stimulation waveforms, including definable waveforms used as part of a pattern of neural stimulation pulses, settings and setting values, other parts of the program, and associated therapeutic efficacy indicators. In various embodiments, each waveform in multiple individually definable fields includes one or more pulses in a neural stimulation pulse, and may include one or more other waveforms in multiple individually definable fields. Examples of such waveforms include pulses, pulse blocks, pulse trains, and pulse train groups, as well as programs. Existing waveforms stored in external storage device 1416 can be defined at least in part by one or more parameters, including but not limited to: amplitude, pulse width, frequency, duration, electrode configuration, total charge injected per unit time, cycle (e.g., on / off time), waveform shape, spatial location of the waveform shape, pulse shape, number of phases, phase sequence, inter-phase time, charge balance, and ramp.

[0189] External storage device 1416 can also store multiple individually definable fields, which can be implemented as part of a program. Each waveform in the multiple individually definable fields is associated with one or more of the multiple individually definable fields. Each of the multiple individually definable fields is defined by one or more electrodes (through which the nerve stimulation pulse is delivered) and the current distribution of the pulse on the one or more electrodes. Various settings in the program can be associated with the control of these waveforms and definable fields.

[0190] Programmable control circuitry 1420 represents an embodiment of control circuitry and can transform or generate specific stimulation parameters or variations to be transmitted to an implantable stimulator based on the results of neurostimulation parameter selection circuitry 1422. This pattern can be defined using one or more waveforms selected from a plurality of individually definable waveforms (e.g., program-defined) stored in external storage device 1416. In various embodiments, programmable control circuitry 1420 checks the values ​​of multiple stimulation parameters against safety rules to constrain these values ​​within the constraints of the safety rules. In one embodiment, the safety rules are heuristic rules.

[0191] User interface device 1410 represents an embodiment of a user interface device and allows a user (e.g., a patient, representative, clinician, etc.) to provide input related to therapeutic goals, such as switching programs or changing the operation of those programs. User interface device 1410 includes a display screen 1412, a user input device 1414, and may be implemented or coupled to parameter selection circuitry 1422 or data provided from data analysis and computing system 1450. Display screen 1412 may include any type of interactive or non-interactive screen, and user input device 1414 may include any type of user input device supporting the various functions discussed herein, such as a touchscreen, keyboard, keypad, touchpad, trackball, joystick, mouse, physical human-computer interaction (e.g., finger, voice, sound, etc.), or the like. User interface device 1410 may also allow the user to perform other functions (e.g., select, modify, enable, disable, activate, schedule, or otherwise define programs, assemblies, provide feedback or input values, or perform other monitoring and programming tasks) when appropriate based on user interface input. Although not shown, the user interface device 1410 can also generate visualizations of such features implemented or programmed by the device, and receive and execute commands for implementing or resuming procedures and neurostimulator operating values ​​(including implementation states for such operating values). These commands and visualizations can be executed in a viewing and guidance mode, a state mode, or a real-time programming mode.

[0192] The controller 1430 may be a microprocessor that communicates with external telemetry circuitry 1440, external communication device 1418, external storage device 1416, programming control circuitry 1420, parameter selection circuitry, and user interface device 1410 via a bidirectional data bus or the like. The controller 1430 may be implemented using a state machine-type design through other types of logic circuit systems, such as discrete components or programmable logic arrays. As used in this disclosure, the term "circuit system" should be understood to refer to discrete logic circuit systems, firmware, microprocessor programming, or combinations thereof.

[0193] The data analysis system 1450 is configured to operate the therapeutic action circuitry system 1460, which can generate or initiate certain actions based on device data (received and processed by the device data processing circuitry 1452) and input text (received and processed by the processing circuitry 1454). The therapeutic action circuitry system 1460 can identify one or more actions related to neurostimulation therapy and provide output to the patient or clinician using either the patient output circuitry system 1462 or the clinician (or other non-patient person, entity, procedure, or combination thereof) output circuitry system 1464. Such output and the actions provided by that output are based on the evaluation and detection of specific patient data, triage data, parameter data, device status from text, and associated device data, or combinations thereof.

[0194] The data analysis system 1450 is also described as including a storage device 1456 for storing or saving data related to device data, text input, patient and clinician outputs, and related settings, logic, or algorithms. For simplicity, other hardware features of the data analysis computing system 1450 are not depicted, but are inferred from the functional capabilities and operation shown in the following figures.

[0195] As will be understood, patients experiencing chronic pain are often willing to provide detailed information about their current medical status within varying input options, such as free-form text responses to questions, questionnaires, question trees, related applications, device data, etc. For example, free-form text in the form of narratives, descriptive statements, or interjections is readily generated by patients and can provide much detail about their actions, physiological state, physical and psychological state, as well as past events, reflecting both objective and subjective outcomes of neurostimulation therapy. However, free-form text can be time-consuming or difficult to interpret for physicians and clinicians, especially when patient feedback may be contradictory (e.g., "I feel fine this morning, but I can't do anything") or incomplete without additional context (e.g., "I can't get out of bed"). Capturing patient feedback using this system can provide many new data points for treatment outcomes, allow for patient triage using alert notifications, and provide a basis for determining whether or why a specific neurostimulation therapy (as well as treatment procedures, programmed values, and programmed effects) is effective.

[0196] Figure 15Block diagram 1500 is illustrated by way of example, showing an embodiment of data interaction between monitoring and processing systems implemented as a programming data service 1570 for the operation of stimulation device 1521 using closed-loop programming, open-loop programming, partially closed-loop programming, or a combination thereof. At a higher level, programming data service 1570 uses a trained model (one of programming models 1502) to generate patient-customized programming settings and parameters (e.g., in one or more procedures 1505). Programming data service 1570 includes computing hardware 1503 to control model training 1501 and other data processing operations, such as generating or controlling diagnostic actions, alerts, programming recommendations, or programming actions. Programming settings and parameters can be implemented automatically or manually (e.g., using the programming techniques mentioned above) on stimulation device 1521 or via patient computing device 1520 or patient programming device 1530.

[0197] Programming data service 1570 communicates via network 1510 with one or both of patient computing device 1520 and programming device 1530 to obtain training data for use by model training logic 1501. The training data may be stored in a database 1504 for patients or patient groups or another large-scale data store (e.g., a data lake). Programming data service 1570 may also include a data analysis or processing engine (not shown) that parses and determines the specific patient's state from various inputs and associated program usage (e.g., to determine which programs and programming settings are beneficial or detrimental to the patient). In some examples, the state of treatment may be based on the correlation between the historical use of a neural stimulation program or parameter set and the patient's current state (e.g., identifying whether pain conditions worsen or improve after the start of a specific program).

[0198] The programming data service 1570 can also analyze various forms of patient input and patient data related to the use of neurostimulation programs or neurostimulation programming parameters. For example, the programming data service 1570 can receive information from program usage, questionnaire selections, text input from human patients, and similar sources via the patient computing device 1520 and the patient programming device 1530. In addition to providing recommended programs, the programming data service can also provide therapy content, triage alerts, and usage recommendations to the patient computing device 1520 and the patient programming device 1530.

[0199] Patients can provide training data (e.g., input data) via patient computing device 1520 or patient programming device 1530. Additional details regarding how input data is collected are discussed with reference to the data processing logic and user interface described in the relevant patent applications. In the example, patient computing device 1520 is a computing device (e.g., a personal computer, tablet, smartphone) or other form of user interaction device that receives and provides interaction with the patient using a graphical user interface 1523, employing programming input logic 1524 and programming output logic 1522. For example, programming input logic 1524 can receive input from the patient via questionnaires, surveys, messages, or other inputs. This input can provide text related to pain or general health, which can be used to identify the patient's mental state, physical state, physiological state, somatic state, or combination of states, the outcome of neurostimulation therapy, or related symptoms. As used herein, the terms “neurostimulator,” “stimulator,” “neurostimulation,” and “stimulation” generally refer to the delivery of electrical energy that affects the neuronal activity of neural tissue, which can be excitatory or inhibitory; for example, by initiating action potentials, inhibiting or blocking the propagation of action potentials, influencing changes in the release or uptake of neurotransmitters / neuromodulators, and inducing changes in the neural plasticity or neurogenesis of tissues. It will be understood that other clinical effects and physiological mechanisms can also be provided by using such stimulation techniques.

[0200] The patient programming device 1530 is depicted as including a user interface 1531 and program implementation logic 1532. Specifically, the program implementation logic 1532 can provide the patient with the ability to implement or switch to a specific program generated by the programming data service 1570. Other forms of programming may also include receiving instructions, recommendations, or feedback (including clinician recommendations, behavioral modifications, etc., selected automatically based on the detected symptoms)

[0201] The programming data service 1570 can also utilize sensor data 1540 from one or more patient sensors 1550 (e.g., wearable devices, sleep trackers, motion trackers, implantable devices, etc.) in one or more internal or external devices. Sensor data 1540 can be used to determine customized and current parameterization techniques for the outcomes of subsensory therapy and multisensor sensory abnormality therapy or neurostimulation therapy. In various examples, the stimulation device 1521 also includes sensors that contribute to the sensor data 1540 to be evaluated by the programming data service 1570.

[0202] In the example, patient sensor 1550 is a physical, physiological, biopsychosocial, or similar sensor that collects data related to physical, biopsychosocial (e.g., stress and / or emotional biomarkers), or physiological factors relevant to the patient's state. Examples of such sensors may include a sleep sensor for sensing a patient's sleep state (e.g., for detecting sleep deprivation); a respiratory sensor for measuring a patient's respiratory rate or respiratory capacity; a motion sensor for identifying the amount or type of exercise; a heart rate sensor for sensing a patient's heart rate; a blood pressure sensor for sensing a patient's blood pressure; an electric skin activity (EDA) sensor for sensing a patient's EDA (e.g., skin conductance response); a facial recognition sensor for sensing a patient's facial expressions; an acoustic sensor (e.g., a microphone) for sensing a patient's voice; and / or an electrochemical sensor for sensing stress biomarkers from the patient's bodily fluids (e.g., enzymes and / or ions, such as lactate or cortisol from saliva or sweat). Other types or forms of sensor devices may also be utilized.

[0203] Figure 16 An embodiment of a data processing flow 1600 affecting neurostimulation therapy for patients is illustrated by way of example, including a neurostimulation control system 1610 based on the functions of collected training data 1612, data processing 1614, and device data processing 1616. Additional details regarding the data flow between the monitoring and processing system 130 and the example user interface 1601 are provided herein. For simplicity, other user interfaces and actions are not depicted.

[0204] In this example, input data 1604 (e.g., parameter inputs, questionnaire responses, question trees, free-form text, patient feedback, interaction results, etc.) is obtained by monitoring and processing system 130. Figure 16 The evaluation of device data 1630 is also described, such as sensor data 1632, therapy status data 1634, and other therapeutic aspects that can be obtained or derived from the neurostimulation device or related neurostimulation programming. Also in this example, output data 1602 (e.g., content) is obtained at the user interface from the monitoring and processing system 130, such as in the form of recommendations, alerts, triage messages, intercalation suggestions, treatment updates, patient information, and the like. The monitoring and processing system 130 may provide clinician recommendations, clinician alerts, or other relevant actions separately from the patient, to the clinician, physician, or device representative.

[0205] The remainder of the data processing flow 1600 illustrates how the data processing results from the monitoring and processing system 130 can be used to implement programming, such as in closed-loop (or partially closed-loop, open-loop, or other) systems. The programming system 1640 can use parameters or programming information 1642 (e.g., programmed neurostimulation programming information) provided from the monitoring and processing system 130 as input to the program implementation logic 1650. The program implementation logic 1650 can be implemented by a parameter adjustment algorithm 1654, which influences neurostimulation program selection 1652 or neurostimulation program modification 1656. For example, some parameter changes can be implemented through simple modifications to program operation; other parameter changes may require the deployment of new programs. The result of parameter or program changes or selections leads to the definition or adjustment of various stimulation parameters at the neurostimulation device 1621, thereby causing different or new stimulation therapeutic effects 1660.

[0206] By way of example, stimulation parameter data 1670 includes operating parameters of the neurostimulation device, which are generated, identified, and / or evaluated by this system and technology, and may include amplitude, frequency, duration, pulse width, pulse type, neurostimulation pulse pattern, waveform in the pulse pattern, and similar settings regarding the intensity, type, and location of the neurostimulator output on one or more corresponding leads. The neurostimulator can use a current source or a voltage source to provide the neurostimulator output, and any number of control techniques can be applied to modify the electrical simulation applied to anatomical sites or systems related to pain or analgesia effects.

[0207] In various embodiments, the neurostimulator program can be defined or updated to indicate parameters defining the spatial, temporal, and informational characteristics for the delivery of modulated energy, including the definition or parameters of the pulses of modulated energy, the waveform of the pulses, pulse blocks (each pulse block comprising a pulse train), pulse trains (each pulse train comprising a sequence of pulse blocks), pulse train groups (each pulse train group comprising a sequence of pulse trains), and programs for such definitions or parameters (each program comprising one or more pulse train groups scheduled for delivery). The characteristics of the waveform defined in the program may include, but are not limited to, the following: amplitude, pulse width, frequency, total charge injected per unit time, cycle time (e.g., on / off time), pulse shape, number of phases, phase sequence, inter-phase time, charge balance, ramp, and spatial variance (e.g., changes in electrode configuration over time). It will be understood that, based on the many characteristics of the waveform itself, the program can have many potentially usable combinations of parameter settings.

[0208] Figure 17 A block diagram 1700 of the neural modulation system 1715 is shown by way of example rather than limitation, for example, similar combinations. Figure 12 and Figure 13The systems depicted and described can be implemented as spinal cord stimulation (SCS) systems or deep brain stimulation (DBS) systems, and example embodiments of this disclosure may include additional forms of stimulation in a patient. The illustrated neuromodulation system 1715 includes an external system 1714 that may include at least one programming device. The illustrated external system 1714 may include a programmer 1711 configured for use by a clinician, patient, device representative, other caregiver, or a combination thereof to communicate with and program the neuromodulator; and a remote control (not shown) configured for use by a patient to communicate with and program the neuromodulator.

[0209] For example, the remote control device may allow the patient to turn the therapy on and off and / or allow the patient to adjust multiple control parameters as patient-programmable parameters. The external system 1714 may also be operatively coupled to one or more patient wearable devices 1713 (e.g., watches, rings, brain-sensing monitors, necklaces, heart rate monitors, Holter monitors, etc.), patient computing devices 1716 (e.g., mobile phones, computers, tablets, etc.), and / or patient artificial intelligence (AI) devices 1717 (e.g., Amazon® Alexa, Google® Assistant).

[0210] Figure 17 A medical device is illustrated as an ambulatory medical device 1715. Examples of ambulatory devices include wearable or implantable neuromodulators. An external system 1714 may include a network of computers, including a computer remotely positioned relative to the ambulatory medical device 1715, capable of communicating with a programmer 1711 and / or a remote controller via one or more communication networks. The remotely positioned computer 1712 and the ambulatory medical device 1715 may be configured to communicate with each other via another external device, such as the programmer 1711 or a remote controller. The remote control device and / or programmer 1711 may allow a user (e.g., a patient and / or clinician or device representative) to answer questions as part of a data collection process. The ambulatory medical device may include internal or external sensors that can be used to collect data that may form at least a portion of overall pain data, healthcare-related data, or other patient data to be transmitted to a data receiving system. Parameter data for programmed therapy may form at least a portion of the healthcare-related data to be transmitted to the data receiving system.

[0211] Figure 18This is a block diagram 1800 illustrating a machine in the exemplary form of a computer system 1801 according to an example embodiment, within which a set or sequence of instructions can be executed to cause the machine to perform any of the methods discussed herein. In alternative embodiments, the machine operates as a stand-alone device or can be connected (networked) to other machines.

[0212] In a networked deployment, the machine can operate as a server machine or a client machine in a server-client network environment, or it can act as a peer machine in a peer-to-peer (or distributed) network environment. The machine can be a personal computer (PC), tablet PC, hybrid tablet, personal digital assistant (PDA), mobile phone, implantable pulse generator (IPG), external experimental stimulator (ETS), external remote controller (RC), user programmer (UP), or any machine capable of executing instructions (sequentially or otherwise) specifying actions to be taken by the machine. Furthermore, although only a single machine is shown, the term "machine" should also be understood to include any collection of such machines that individually or jointly execute one or more sets of instructions to perform any one or more of the methods discussed herein. Similarly, the term "processor-based system" should be considered to include any collection of one or more such machines that are controlled or operated by a processor (e.g., a computer) to individually or jointly execute instructions to perform any one or more of the methods discussed herein.

[0213] Example computer system 1801 includes at least one processor 1802 (e.g., a central processing unit (CPU), a graphics processing unit (GPU), or both, a processor core, a computer node, etc.), main memory 1804, and static memory 1806, which communicate with each other via link 1808 (e.g., a bus, an interconnect). Computer system 1801 may also include a video display unit 1810, an alphanumeric input device 1812 (e.g., a keyboard), and a user interface (UI) navigation device 1814 (e.g., a mouse). In one embodiment, the video display unit 1810, the input device 1812, and the user interface (UI) navigation device 1814 are integrated into a touchscreen display. Computer system 1801 may additionally include a storage device 1816 (e.g., a drive unit), a signal generation device 1818 (e.g., a speaker), an output controller 1828, a network interface device 1820, and one or more sensors 1821, such as a global positioning system (GPS) sensor, a compass, an accelerometer, or other sensors. It will be understood that other forms of machines or apparatus (such as IPG, RC, CP devices and the like) capable of implementing the methods discussed in this disclosure may not be combined with or utilize Figure 18 Each component described (such as the GPU, video display unit, keyboard, etc.).

[0214] Storage device 1850 includes mass storage 1816 and machine-readable medium 1822 on which one or more sets of data structures or instructions 1824 (e.g., software) embodying or utilized by any or more of the methods or functions described herein are stored. Instructions 1824 may also reside wholly or at least partially in main memory 1804, static memory 1806, and / or processor 1802 during execution by computer system 1801, wherein main memory 1804, static memory 1806, and processor 1802 also constitute machine-readable medium.

[0215] Although machine-readable medium 1822 is shown as a single medium in the example embodiment, the term "machine-readable medium" can include a single medium or multiple media (e.g., a centralized or distributed database, and / or associated caches and servers) storing one or more instructions 1824. The term "machine-readable medium" should also be understood to include any tangible (e.g., non-transitory) medium capable of storing, encoding, or carrying instructions for execution by a machine and causing the machine to perform any one or more methods of this disclosure, or capable of storing, encoding, or carrying data structures utilized by or associated with such instructions. Therefore, the term "machine-readable medium" should be understood to include, but is not limited to, solid-state memory, as well as optical and magnetic media. Specific examples of such machine-readable media include non-transitory memory, including, by way of example, semiconductor memory devices (e.g., electrically programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM)) and flash memory devices; disks, such as internal hard disks and removable disks; magneto-optical disks; and CD-ROM and DVD-ROM disks.

[0216] Instruction 1824 can also be transmitted or received over communication network 1826 via network interface device 1820 using a transmission medium, utilizing any of a variety of well-known transport protocols (such as HTTP). Examples of communication networks may include local area networks (LANs), wide area networks (WANs), the Internet, mobile phone networks, ordinary old-style telephone (POTS) networks, and wireless data networks (such as Wi-Fi, 3G and 4G LTE / LTE-A, or 5G networks). The term "transmission medium" should be understood to include any intangible medium capable of storing, encoding, or carrying instructions for execution by the machine, and includes digital or analog communication signals or other intangible media to facilitate communication of such software.

[0217] As used herein, the terms “machine storage medium,” “device storage medium,” “machine-readable medium,” and “computer storage medium” refer to the same thing and are used interchangeably in this disclosure. The term refers to one or more storage devices and / or media (e.g., centralized or distributed databases, and / or associated caches and servers) that store executable instructions and / or data. The term should therefore be understood to include, but is not limited to, solid-state memory, as well as optical and magnetic media, including memory internal or external to a processor. Specific examples of machine storage media, computer storage media, and / or device storage media include non-volatile memory, including, as examples, semiconductor memory devices such as erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), field-programmable gate arrays (FPGAs), and flash memory devices; disks, such as internal hard disks and removable disks; magneto-optical disks; and CD-ROM and DVD-ROM disks. The terms “machine storage medium,” “computer storage medium,” and “device storage medium” expressly exclude carrier waves, modulated data signals, and other such media, at least some of which are covered by the term “signal medium” discussed below.

[0218] The terms "transmission medium" and "signal medium" refer to the same thing and are used interchangeably in this disclosure. The terms "transmission medium" and "signal medium" should be understood to include any intangible medium capable of storing, encoding, or carrying instructions for execution by the machine depicted in block figure 1800, and include digital or analog communication signals or other intangible media to facilitate communication of such software. Therefore, the terms "transmission medium" and "signal medium" should be understood to include any form of modulated data signal, carrier wave, etc. The term "modulated data signal" means a signal whose one or more characteristics are set or changed in such a way that information is encoded in that signal.

[0219] The terms “machine-readable medium,” “computer-readable medium,” and “device-readable medium” refer to the same thing and are used interchangeably in this disclosure. The term is defined to include both machine storage media and transmission media. Therefore, the term includes both storage devices / media and carrier / modulated data signals.

[0220] The detailed description above is intended to be illustrative and not restrictive. Therefore, the scope of this disclosure should be determined by reference to the appended claims and the full scope of their equivalents. For example, the examples (or one or more aspects thereof) described above may be used in combination with each other. Other embodiments may also be used, such as those used by those skilled in the art in reviewing the above description. The scope of the invention should be determined by reference to the appended claims and the full scope of their equivalents.

Claims

1. A system for parameterizing a closed-loop algorithm for providing subsensory therapy, the system comprising: A stimulation output circuit configured to deliver neural stimulation; A sensing circuit configured to sense neural signals indicating neural responses, each of which is a response to the delivery of the neural stimulus; A stimulation control circuit, coupled to the stimulation output circuit and the sensing circuit, is configured to use a plurality of stimulation parameters to control the delivery of the neural stimulation; as well as One or more memories, the one or more memories storing instructions that, when executed by one or more processors, cause the one or more processors to perform the following operations: Record the characteristics of the evoked composite action potential (ECAP) received from the sensing circuit; Stimulation parameters are adjusted, at least in part, based on the recorded ECAP features, in order to maintain consistent neural activation; Record accelerometer data; A regression model is generated to map the accelerometer data to optimal stimulation parameters to achieve consistent neural activation, the mapping enabling closed-loop control of the subsensory therapy. The regression model and subsensory dose levels were applied to multiple subsensory therapy procedures; as well as The regression model is deployed to the patient via the stimulation output circuit to deliver the neural stimulation.

2. The system according to claim 1, wherein, The instruction performs the following operations: Identify the target range for the ECAP feature, the target range including an upper limit and a lower limit of the ECAP range; and The consistent neural activation is determined by maintaining the values ​​of the ECAP feature within the target range.

3. The system according to any one of claims 1 to 2, wherein, The regression model is selected from at least one of the following: linear regression model, support vector machine model, neural network model, autoregressive model, or moving average model.

4. The system according to any one of claims 1 to 3, wherein, The instruction performs the following operations: The patient is prompted to perform one or more physical movements; and The user interface receives patient input that takes into account environmental factors that affect sensor readings.

5. The system according to claim 4, wherein, The one or more body movements include one or more of a plurality of postural changes performed by the patient, including at least one of lying down, standing up, bending over, twisting to the side, performing daily living activities, and leaning back.

6. The system according to any one of claims 1 to 5, wherein, The instruction performs the following operations: One or more settings are identified in which changes are proposed for delivering the neural stimulation to the patient, the neural stimulation being defined at least in part based on the accelerometer data.

7. The system according to any one of claims 1 to 6, wherein, The instruction performs the following operations: Provide a monitoring platform; as well as Receive information about multiple accelerometer data points, which represent overall patient health based on a combination of neural stimulation parameters, including: Electrical waveform; and The selection of electrodes through which the electrical waveform is delivered.

8. The system according to claim 7, wherein, The instruction performs the following operations: Generate suggestions for one or more additional combinations of the neural stimulation parameters; and The recommendations are provided to the monitoring platform.

9. The system according to any one of claims 1 to 8, wherein, The multiple accelerometer data provide supplementary data to the ECAP feature to optimize stimuli in the sub-sensory domain.

10. The system according to any one of claims 1 to 9, wherein, The system is also configured to analyze data used for neural stimulation programming.

11. The system according to any one of claims 1 to 10, wherein, The instruction performs the following operations: The closed-loop algorithm is initialized at least in part based on historical patient data; and The closed-loop algorithm is adjusted using patient-specific data.

12. The system according to any one of claims 1 to 11, wherein, The instruction performs the following operations: The mapping is recalibrated to take into account the changes identified in the implantable pulse generator (IPG), which is configured to deliver electrical stimulation via leads and electrodes.

13. The system according to any one of claims 1 to 12, wherein, The regression model is also configured to generate a customized mapping for each patient by collecting ECAP and accelerometer data in the clinical setting.

14. The system according to claim 13, wherein, The regression model collects ECAP factors during the mapping period, which are selected from at least one of the following: amplitude, detection threshold, perception level, linearity with stimulus intensity, stimulus parameters, electrode location, or neural activation level.

15. The system according to any one of claims 1 to 14, further comprising: An anomaly detector is configured to identify divergent outputs between the ECAP algorithm and the acceleration calculation method.

Images