Electrode Selection for Electrical Stimulation Therapy

Abstract

In one example, a method includes selecting, by one or more processors and based on a sensed electrical signal representation for a particular patient and a plurality of sensed electrical signal representations for a plurality of other patients, an electrode combination from a plurality of combinations of one or more implantable electrodes for delivering electrical stimulation to the particular patient.

Description

[0001] This application is a divisional application of the invention patent application with the international application date of 2015-10-14, the international application number of PCT / US2015 / 055542, the application number entering the Chinese national phase of 201580059751.3, and the title of "Electrode Selection for Electrical Stimulation Therapy". Technical Field

[0002] The present disclosure relates to electrical stimulation therapy. Background Art

[0003] Implantable medical devices (e.g., electrical stimulators or therapeutic delivery devices) have been proposed for various therapeutic applications (e.g., deep brain stimulation (DBS), spinal cord stimulation (SCS), pelvic stimulation, gastric stimulation, peripheral nerve stimulation, functional electrical stimulation, or delivery of therapeutic agents, insulin, analgesics, or anti-inflammatory agents to target tissue sites in a patient). In some therapeutic systems, an implantable electrical stimulator delivers electrical therapy to a target tissue site in a patient with the aid of one or more electrodes, which may be deployed via medical leads and / or on a housing of the electrical stimulator, or both. In some therapeutic systems, therapy may be delivered via a specific combination of electrodes carried by the leads and / or housing of the electrical stimulator.

[0004] During a programming session (which may occur during implantation of a medical device), during a trial session, or during an in-clinic or remote follow-up session after implantation of the medical device in a patient, a clinician can generate one or more treatment programs (also referred to as treatment parameter sets) that have been found to provide effective treatment to a patient, wherein each treatment program can define the values ​​of the treatment parameter set. The medical device can deliver treatment to the patient according to the stored one or more treatment programs. In the case of electrical stimulation, the treatment parameters can define the characteristics of the electrical stimulation waveform to be delivered. For example, in an example where electrical stimulation is delivered in the form of electrical pulses, the treatment parameters can include electrode configuration (the electrode configuration includes electrode combination and electrode polarity), amplitude (the amplitude can be current amplitude or voltage amplitude), pulse width, and pulse rate. Summary of the invention

[0005] In general, the present disclosure relates to devices, systems, and methods for selecting electrode combinations for delivering electrical stimulation to a patient in one or more implantable electrodes. Throughout the present disclosure, the terms "electrode" and "contact" can be used in a substantially interchangeable manner. In some examples, the device can select an electrode (i.e., electrical contact) combination for delivering electrical stimulation to a specific patient based on an electrical signal representation for a specific patient and a plurality of electrical signal representations for a plurality of patients. In some examples, the device can use machine learning such as support vector machine (SVM) classification to select a contact combination. For example, the device can use an SVM trained based on one or more electrical signal representations in the plurality of electrical signal representations for the plurality of patients. In some examples, the device can use a form of machine learning (i.e., in addition to the SVM). For example, the device can use any kind of supervised or unsupervised form of machine learning to select a contact combination.

[0006] The contacts can be disposed on one or more implantable leads and / or housing of the implantable stimulator. The device using machine learning to select the contacts can be an implantable stimulator or an external programmer for the implantable stimulator. Alternatively, machine learning functionality can be provided in both the implantable stimulator and the external programmer for the implantable stimulator. In other examples, the implantable leads carrying the contacts can be percutaneously coupled to the external stimulator. In the described case, the external stimulator includes a machine learning functionality for selecting contacts for delivering stimulation to the patient.

[0007] As an example, a method for controlling the delivery of electrical stimulation therapy includes: selecting, by one or more processors and based on sensed electrical signal representations for the specific patient and multiple sensed electrical signal representations for multiple other patients, an electrode combination from multiple combinations of one or more implantable electrodes for delivering electrical stimulation therapy to the specific patient; and generating, by the one or more processors, information for controlling the delivery of the electrical stimulation based on the selected combination of the electrodes.

[0008] As another example, a device includes a memory configured to store a sensed electrical signal representation for a specific patient; and one or more processors configured to select an electrode combination from multiple combinations of one or more implantable electrodes for delivering electrical stimulation therapy to the specific patient based on the sensed electrical signal representation for the specific patient and multiple sensed electrical signal representations for multiple other patients; and generate information for controlling the delivery of the electrical stimulation based on the selected combination of the electrodes.

[0009] As another example, an implantable device includes: a device for selecting an electrode combination from multiple combinations of one or more implantable electrodes for delivering electrical stimulation to the specific patient based on a sensed electrical signal representation for the specific patient and a plurality of sensed electrical signal representations for a plurality of other patients; and a device for delivering the electrical stimulation to the specific patient via the selected electrode combination.

[0010] As another example, a programmer for an implantable device includes: a device for selecting an electrode combination from multiple combinations of one or more implantable electrodes for delivering electrical stimulation to the specific patient based on a sensed electrical signal representation for the specific patient and a plurality of sensed electrical signal representations for a plurality of other patients; and a device for programming the implantable medical device to deliver the electrical stimulation to the specific patient via the selected electrode combination.

[0011] As another example, a computer-readable storage medium stores multiple instructions that, when executed, cause one or more processors to select an electrode combination from multiple combinations of one or more implantable electrodes for delivering electrical stimulation to a specific patient based on a sensed electrical signal representation for the specific patient and a plurality of sensed electrical signal representations for multiple other patients; and generate information for controlling the delivery of the electrical stimulation based on the selected combination of the electrodes.

[0012] As another example, a system includes one or more processors configured to select an electrode combination from multiple combinations of one or more implantable electrodes for delivering electrical stimulation to a specific patient based on a sensed electrical signal representation for the specific patient and a plurality of sensed electrical signal representations for a plurality of other patients; and a medical device configured to deliver stimulation to the specific patient via the selected electrode combination.

[0013] The details of one or more examples are set forth in the accompanying drawings and the description below. Other features, objects, and advantages will be apparent from the description and drawings, and from the claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 is a conceptual diagram illustrating an example deep brain stimulation (DBS) system configured for delivering electrical stimulation therapy to a tissue site within a patient's brain.

[0015] Figure 2 is a functional block diagram showing the components of an example medical device.

[0016] Figure 3 is a functional block diagram showing the components of an example medical device programmer.

[0017] Figure 4 is a conceptual diagram showing an example of a method that can be implemented in accordance with one or more techniques of the present disclosure. Figure 2 and Figure 3 Further details of one example of a contact selection module provided in one or both of the devices shown in FIG.

[0018] Figure 5 is a flow diagram of an example technique for automatically selecting a contact combination for delivering electrical stimulation in accordance with one or more techniques of this disclosure.

[0019] Figure 6 is a flowchart of an example technique for training one or more machine learning models (e.g., one or more SVM classifiers) according to one or more techniques of this disclosure.

[0020] 7A to 7C Examples of data that can be used to create feature sets for classification in accordance with one or more techniques of this disclosure are presented.

[0021] Fig. 8A and Figure 8B Leave-One-out (LOO) classification error according to one or more techniques of the present disclosure is shown.

[0022] 9A to 9C is a scatter plot illustrating beta (β) band versus theta (θ) band power of local field potentials recorded on a bipolar pair in accordance with one or more techniques of the present disclosure.

[0023] Fig.9D yes 9A to 9C Legend for .

[0024] Fig. 10A and Fig. 10B is a diagram illustrating the output of a machine learning model for selecting a contact combination according to one or more techniques of the present disclosure. DETAILED DESCRIPTION

[0025] In general, the present disclosure relates to devices, systems, and methods for selecting a combination of one or more implantable electrodes for delivering electrical stimulation to a patient. For example, a device may select an electrode combination for a particular patient based on an electrical signal representation for the particular patient and a plurality of electrical signal representations for a plurality of patients. In some examples, a device may use machine learning such as a support vector machine (SVM) classifier (referred to as an SVM in the present disclosure) to select an electrode combination. For example, a device may use an SVM trained based on one or more of the plurality of electrical signal representations for the plurality of patients.

[0026] The contacts can be disposed on one or more implantable leads and / or housing of the implantable stimulator. The device using machine learning to select electrodes can be an implantable stimulator or an external programmer for an implantable stimulator. Alternatively, machine learning functionality can be provided in both the implantable stimulator and the external programmer for the implantable stimulator. In other examples, the implantable leads carrying the electrodes can be percutaneously coupled to the external stimulator. In the described case, the external stimulator includes a machine learning functionality for selecting electrodes for delivering stimulation to the patient.

[0027] Parkinson's disease (PD) is a progressive neurodegenerative disorder characterized by exhaustion of dopaminergic neurons in the basal ganglia-thalamus-cortical network. When reaching this stage, the manifestation of the disease may include one or more of the unique motor dysfunction, bradykinesia, rigidity and tremor of motor disability. In some examples, deep brain stimulation (DBS) treatment can be used to deliver electrical stimulation to treat the motor symptoms of drug-refractory PD patients. In some examples, DBS treatment may involve implanting one or more leads unilaterally or bilaterally into the brain to deliver electrical stimulation to the target structure in the basal ganglia. For both physicians (also referred to as clinicians) and patients, selecting effective stimulation parameters for DBS treatment may be time-consuming. In this way, it may be desirable to reduce the amount of time consumed by selecting stimulation parameters.

[0028] In some examples, the clinician can manually select stimulation parameters using two known items (stimulation input and behavioral output). For example, the clinician can manually select a set of stimulation parameters and cause the implantable stimulator to stimulate the patient according to the set of stimulation parameters. Based on the patient's response to the stimulation, the clinician can modify one or more of the stimulation parameters until a satisfactory stimulation response is achieved. A satisfactory response may include a reduction in PD symptoms such as motor dysfunction as described above and no side effects or at least tolerable side effects. As described above, manually selecting stimulation parameters may be very time-consuming. In this way, it may be desirable to reduce the amount of time consumed by selecting stimulation parameters.

[0029] According to one or more techniques of the present disclosure, a device can automatically select stimulation parameters for a patient based at least on the patient's sensed neural data. For example, a device can compare an electrical signal representation for a patient with one or more machine learning models to automatically select electrodes for delivering electrical stimulation to the patient. In some examples, one or more machine learning models can be trained based on one or more electrical signal representations of multiple electrical signal representations for multiple patients. In this way, the device can reduce programming time and can improve patient outcomes.

[0030] In order to implement automated programming algorithms based on sensory neural data, it may be useful to find the neurophysiological interrelationships of both behavior and stimulation. Medtronic Neuromodulation has developed an investigation device known as Activa PC+S. In addition to delivering DBS stimulation, this long-term implantable neurostimulator can sense and record the sub-microvolt local field potential (LFP) of the electrode pairs on the DBS leads of the patient. For research purposes, with the approval of the appropriate institutional review board and the informed consent of the parents, these devices are implanted in PD patients in the United States and in multiple centers in Europe. As part of the research protocol, LFP records are obtained from all possible electrode pairs during the programming session after the first implantation for the patient and during continuous follow-up. Because LFP is robust to long-term electrode-tissue impedance changes (unlike neuronal recordings) and provides sufficiently high resolution for sensing in deep brain structures (unlike cortical electroencephalography) (see G. Buzsáki, CA Anastassiou and C. Koch, "The Origin of Extracellular Fields and Currents—EEG, ECoG, LFP and Spikes", Nature Reviews Neuroscience, 2012, Vol. 13, No. 6, pp. 407-20), LFP is considered to be the optimal signal for measuring brain states in long-term settings.

[0031] We selected 15 patients with idiopathic Parkinson's disease who had DBS leads placed unilaterally or bilaterally in the subthalamic nucleus (STN) from the patients implanted with Activa PC+S to date. We did not preselect recordings based on the patient's disease subtype or LFP spectral pattern. Here, we used these recordings to preliminarily investigate whether the characteristics of the recordings are associated with the electrodes selected by the attending physician for DBS treatment of the patient. In some examples, the oscillatory content in the beta frequency band of the LFP (13 to 30 Hz) can be associated with the location and efficacy of deep brain stimulation. Thus, the beta activity in the LFP sensed by the Activa PC+S from various macroelectrodes can contain information that can help select electrodes to be used for effective stimulation. However, in some examples, it may be beneficial to start with a large set of spectral features in order to reduce feature selection bias.

[0032] Figure 1is a conceptual diagram illustrating an example therapeutic system 10 configured to deliver therapy to a patient 12 in order to manage a disorder of the patient 12. The patient 12 will typically be a human patient. However, in some cases, the therapeutic system 10 may be applied to other mammalian or non-mammalian non-human patients. Figure 1 In the example shown in , therapy system 10 includes a medical device programmer 14, an implantable medical device (IMD) 16, a lead extension 18, and one or more leads 20A and 20B (collectively referred to as "leads 20") having corresponding electrode sets 24, 26. IMD 16 includes a stimulation generator configured to generate and deliver electrical stimulation therapy to one or more regions of a brain 28 of patient 12 via one or more electrodes 24, 26 of leads 20A and 20B, respectively.

[0033] exist Figure 1 In the example shown in , because IMD 16 is configured to deliver electrical stimulation therapy directly to tissue within brain 28 (e.g., at a tissue site beneath the dura mater of brain 28 or one or more branches or nodes, or intersections of fiber trajectories), therapy system 10 may be referred to as a deep brain stimulation (DBS) system. In other examples, lead 20 may be positioned to deliver therapy to the surface of brain 28 (e.g., the cortical surface of brain 28). For example, in some examples, IMD 16 may provide cortical stimulation therapy to patient 12, for example, by delivering electrical stimulation to one or more tissue sites in the cortex of brain 28. As another example, IMD 16 may provide vagus nerve stimulation (VNS) therapy to patient 12 by delivering electrical stimulation to one or more vagus nerve tissue sites.

[0034] DBS can be used to treat or manage a variety of patient conditions in patient 12, such as, but not limited to, spastic disorders (e.g., epilepsy), pain, migraine, psychiatric disorders (e.g., major depressive disorder (MDD)), bipolar disorder, anxiety disorders, post-traumatic stress disorder, hypotension, obsessive-compulsive disorder (OCD), behavioral disorders, mood disorders, memory disorders, mental state disorders, movement disorders (e.g., essential tremor or Parkinson's disease), Huntington's disease, Alzheimer's disease, or other neurological or psychiatric diseases and impairments.

[0035] Treatment systems configured to treat other patient conditions via treatment delivered to the brain 28 or another suitable target treatment delivery site in the patient 12 may also be used in accordance with the techniques disclosed herein. For example, in other applications of the treatment system 10, the target treatment delivery site in the patient 12 may be a location proximate to the spinal cord or sacral nerves (e.g., S2, S3, or S4 sacral nerves) in the patient 12, or any other suitable nerve, organ, muscle, or muscle group in the patient 12, which location may be selected, for example, based on the patient's condition. For example, the treatment system 10 may be used to deliver electrical stimulation or therapeutic agents to tissue proximate to the pudendal nerve, perineal nerve, or other area of ​​the nervous system, in which case the lead 20 would be implanted or substantially fixed proximate to the corresponding nerve. As a further example, the electrical stimulation system may be positioned to deliver stimulation to help manage peripheral neuropathy or postoperative pain relief, ilioinguinal nerve stimulation, intercostal nerve stimulation, gastric stimulation for treating gastric dysmotility and obesity, urinary dysfunction, sexual dysfunction, muscle stimulation, and thereby relieve other peripheral and localized pain (e.g., leg pain or back pain).

[0036] exist Figure 1 In the example shown, IMD 16 can be implanted in a subcutaneous pocket in the pectoralis region of patient 12. In other examples, IMD 16 can be implanted in other areas of patient 12, such as in the abdomen or buttocks of patient 12 or in a subcutaneous pocket near the skull of patient 12. Implantable lead extension 18 is coupled to IMD 16 via connector block 30 (also referred to as a connector), which may include, for example, electrical contacts electrically coupled to corresponding electrodes on lead extension 18. The electrical contacts electrically couple electrodes 24, 26 carried by lead 20 to IMD 16. Lead extension 18 travels from the implantation site of IMD 16 in the chest of patient 12 along the neck of patient 12 and through the skull of patient 12 to enter brain 28. IMD 16 can be constructed of biocompatible materials that can resist corrosion and degradation from body fluids. IMD 16 can include an airtight housing 34 for substantially enclosing components (e.g., a processor, a therapy module, and a memory).

[0037] exist Figure 1 In the example shown in , leads 20 are implanted in the right and left hemispheres of brain 28, respectively, to deliver electrical stimulation to one or more regions of brain 28, which may be selected based on a number of factors, such as the type of patient condition for which therapy system 10 is implemented to manage. Other implantation sites for leads 20 and IMD 16 are contemplated. For example, IMD 16 may be implanted on or in skull 32, or leads 20 may be implanted at multiple target tissue sites in the same hemisphere, or IMD 16 may be coupled to a single lead implanted in one or both hemispheres of brain 28.

[0038] Lead 20 may be positioned for delivering electrical stimulation to one or more target tissue sites within brain 28 in order to manage patient symptoms associated with a disorder of patient 12. Lead 20 may be implanted so as to position electrodes 24, 26 at desired locations of brain 28 via any suitable technique, such as through corresponding burr holes in the skull of patient 12 or through common burr holes in skull 32. Lead 20 may be placed at any location within brain 28 so that electrodes 24, 26 are able to provide electrical stimulation to a target treatment delivery site within brain 28 during treatment. Different neurological or psychiatric disorders may be associated with activity in one or more regions of brain 28, which may differ from patient to patient. Thus, a target treatment delivery site for treatment by electrical stimulation delivered with lead 20 may be selected based on the patient's condition. For example, suitable target therapy delivery sites within the brain 28 for controlling movement disorders of the patient 12 may include one or more of the following: the pedunculopontine nucleus (PPN), thalamus, basal ganglia structures (e.g., globus pallidus, substantia nigra, or subthalamic nucleus), zona incerta, fiber tracts, lenticular fasciculus (and its branches), lenticular loops, or Field of Forel (thalamic fasciculus). The PPN may also be referred to as the pontine tegmental nucleus.

[0039] As another example, in the case of MDD, bipolar disorder, OCD, or other anxiety disorders, leads 20 may be implanted to deliver electrical stimulation to the anterior limb of the internal capsule of the brain 28, as well as only the ventral portion of the anterior limb of the internal capsule (also referred to as VC / VS), the subgenual portion of the cingulate cortex (which may be referred to as CG25), the anterior cingulate cortex Brodmann areas 32 and 24, various parts of the prefrontal cortex (including the dorsolateral and medial prefrontal cortices (PFC) (e.g., Brodmann area 9), the ventromedial prefrontal cortex (e.g., Brodmann area 10), the lateral and medial orbitofrontal cortex (e.g., Brodmann area 11)), the medial or nucleus accumbens, the thalamus, the intralaminar nucleus of the thalamus, the amygdala, the hippocampus, the lateral hypothalamus, the locus coeruleus, the dorsal raphe nucleus, the ventral tegmental area of ​​the midbrain, the substantia nigra, the subthalamic nucleus, the inferior thalamic peduncle, the dorsomedial nucleus of the thalamus, the habenula, the bed nucleus of the stria terminalis, or any combination thereof.

[0040] As another example, for example, in cases of convulsions or Alzheimer's disease, lead 20 can be implanted to deliver electrical stimulation to areas within the Circuit of Papez, such as, for example, one or more of the anterior thalamic nucleus, internal capsule, cingulate gyrus, fornix, mammillary bodies, mammillary thalamic tract (mammillary body thalamic fiber bundle), or hippocampus.

[0041] As another example, in the case of Parkinson's disease, leads 20 may be implanted to deliver electrical stimulation (either unilaterally or bilaterally) to an area within the subthalamic nucleus (STN). Target therapy delivery sites that are not located in the brain 28 of the patient 12 are also contemplated.

[0042] Although lead 20 is Figure 1 18, but in other examples, leads 20 may be coupled to IMD 16 via separate lead extensions or directly to IMD 16. Figure 1 System 10 is shown as including two leads 20A and 20B coupled to IMD 16 via lead extensions 18, but in some examples, system 10 may include one lead or more than two leads.

[0043] exist Figure 1 In the example shown in , the electrodes 24, 26 of the lead 20 are shown as ring electrodes. The ring electrodes can be relatively easy to program and may be able to deliver an electric field to any tissue adjacent to the lead 20. In other examples, the electrodes 24, 26 of the lead 20 can have different configurations. For example, one or more of the electrodes 24, 26 of the lead 20 can have a complex electrode array geometry that can generate a shaped electric field, including cross stimulation. Examples of complex electrode array geometries can include electrode arrays positioned at different axial positions along the length of the lead and at different angular positions around the periphery (e.g., circumference) of the lead. Complex electrode array geometries can include multiple electrodes (e.g., partial rings or segmented electrodes) around the circumference of each lead 20 in addition to or in place of ring electrodes. In this way, electrical stimulation can be directed from the lead 20 to a specific direction in order to enhance the therapeutic effect and reduce possible adverse side effects caused by stimulating a large tissue volume. As a further example, the electrode can be a patch electrode that can be carried on a flat strip lead or a cylindrical lead.

[0044] As in Figure 1 As shown in the example of , electrode group 24 of lead 20A may include electrodes 24A, 24B, 24C, and 24D, and electrode group 26 of lead 20B may include electrodes 26A, 26B, 26C, and 26D. In some examples, each of electrodes 24 and 26 may be configured to deliver electrical stimulation independently.

[0045] In some examples, housing 34 of IMD 16 may include one or more stimulation electrodes and / or sensing electrodes. For example, housing 34 may include a conductive material that is exposed to the tissue of patient 12 when IMD 16 is implanted in patient 12 or electrodes that may be attached to housing 34. In other examples, lead 20 may have a plurality of electrodes other than the electrodes shown in FIG. Figure 1 For example, lead 20 may be a flat bar lead, a ball lead, a bendable lead, or any other type of shape that is effective in treating patient 12.

[0046] IMD 16 can deliver electrical stimulation therapy to the brain 28 of patient 12 according to one or more stimulation therapy programs (also referred to herein as "stimulation parameter value sets"). A stimulation therapy program can define one or more electrical stimulation parameter values ​​for a therapy generated by a stimulation generator of IMD 16 and delivered from IMD 16 via one or more electrodes 24, 26 to a target therapy delivery site in patient 12. Electrical stimulation parameters can define aspects of the electrical stimulation therapy and can include, for example, voltage or current amplitude of the electrical stimulation signal, charge level of the electrical stimulation, frequency of the electrical stimulation signal, waveform shape, on / off cycle state (e.g., if the cycle is "off", the stimulation is always on, and if the cycle is "on", the stimulation cycles on or off), and in the case of electrical stimulation pulses, pulse rate, pulse width, and other appropriate parameters (e.g., duration or duty cycle). In addition, if different electrodes can be used to deliver stimulation, the treatment parameters of the therapy program can be further characterized by an electrode combination that can define the selected electrodes 24, 26 and their corresponding polarities. In some examples, a continuous waveform can be used to deliver stimulation and the stimulation parameters are related to defining this waveform.

[0047] In addition to being configured to deliver therapy to manage a disorder of patient 12, therapy system 10 may be configured to sense bioelectric brain signals or another physiological parameter of patient 12. For example, IMD 16 may include a sensing module configured to sense bioelectric brain signals within one or more regions of brain 28 via a subset of electrodes 24, 26, another electrode group, or both. Thus, in some examples, electrodes 24, 26 may be used to deliver electrical stimulation from a therapy module to a target site within brain 28 and to sense brain signals within brain 28. However, IMD 16 may also use a separate sensing electrode group to sense bioelectric brain signals. In some examples, the sensing module of IMD 16 may sense bioelectric brain signals via one or more electrodes 24, 26 that are also used to deliver electrical stimulation to brain 28. In other examples, one or more of electrodes 24, 26 may be used to sense bioelectric brain signals, while one or more different electrodes 24, 26 may be used to deliver electrical stimulation.

[0048] The external medical device programmer 14 is configured to wirelessly communicate with the IMD 16 as needed to provide or retrieve treatment information. The programmer 14 is an external computing device that a user (e.g., a clinician and / or patient 12) can use to communicate with the IMD 16. For example, the programmer 14 can be a clinician programmer used by the clinician to communicate with the IMD 16 and program one or more treatment programs of the IMD 16. In addition or alternatively, the programmer 14 can be a patient programmer that allows the patient 12 to select a program and / or view and modify treatment parameter values. The clinician programmer can include more programming features than the patient programmer. In other words, more complex or sensitive tasks can be allowed only by the clinician programmer to prevent untrained patients from making undesirable changes to the IMD 16.

[0049] Programmer 14 may be a handheld computing device having a display viewable by a user and an interface (i.e., a user input mechanism) for providing input to programmer 14. For example, programmer 14 may include a small display screen (e.g., a liquid crystal display (LCD) or a light emitting diode (LED) display) for presenting information to the user. In addition, programmer 14 may include a touch screen display, a keyboard, buttons, a peripheral pointing device, voice activation, or another input mechanism that allows a user to navigate through the user interface of programmer 14 and provide input. If programmer 14 includes buttons and a keyboard, the buttons may be dedicated to performing a certain function (e.g., a power button), the buttons and keyboard may be soft keys that change in function depending on the portion of the user interface currently being viewed by the user, or any combination thereof.

[0050] In other examples, programmer 14 may be a larger workstation or a separate application within another multifunction device rather than a dedicated computing device. For example, the multifunction device may be a laptop computer, a tablet computer, a workstation, one or more servers, a cellular telephone, a personal digital assistant, or another computing device that can run an application that enables the computing device to operate as secure medical device programmer 14. A wireless adapter coupled to the computing device may enable secure communication between the computing device and IMD 16.

[0051] When programmer 14 is configured for use by a clinician, programmer 14 may be used to transmit programming information to IMD 16. The programming information may include, for example, hardware information (e.g., type of lead 20, arrangement of electrodes 24, 26 on lead 20, location of lead 20 within brain 28), one or more therapy programs defining therapy parameter values, therapy windows for one or more electrodes 24, 26, and any other information useful for programming into IMD 16. Programmer 14 may also be able to perform functional testing (e.g., measuring impedance of electrodes 24, 26 of lead 20).

[0052] The clinician can also generate treatment programs with the aid of programmer 14 and store them in IMD 16. Programmer 14 can assist the clinician in creating / identifying treatment programs by providing a system for identifying potentially beneficial treatment parameter values. For example, during a programming session, programmer 14 can automatically select an electrode combination for delivering treatment to a patient. In some examples, at least some of the treatment programs can have the same electrode combination (but different values ​​for at least one other treatment parameter), and these treatment programs can be organized into subsets, each with the same electrode combination. The processor of programmer 14 can select the most effective treatment program for each subset and display a list of selected treatment programs. The clinician can select from the list a treatment program for providing treatment to patient 12 to address the symptoms associated with the patient's condition.

[0053] Programmer 14 may also be configured for use by patient 12. When configured as a patient programmer, programmer 14 may have limited functionality (compared to a clinician programmer) in order to prevent patient 12 from altering critical functions of IMD 16 or applications that may be harmful to patient 12.

[0054] Regardless of whether programmer 14 is configured for use by a clinician or a patient, programmer 14 is configured to communicate with IMD 16 and optionally another computing device via wireless communications. Programmer 14 may, for example, communicate with IMD 16 via wireless communications using radio frequency (RF) and / or inductive telemetry techniques known in the art, which may include techniques for proximal, mid-range, or longer-range communications. Programmer 14 may also communicate with another programmer or computing device via a wired or wireless connection using any of a variety of local wireless communication techniques (e.g., RF communications set up according to 802.11 or Bluetooth specifications, infrared (IR) communications set up according to IRDA specifications, or other standard or proprietary telemetry protocols). Programmer 14 may also communicate with other programming devices or computing devices via the exchange of removable media (e.g., magnetic or optical disks, memory cards, or memory sticks). In addition, programmer 14 may communicate with IMD 16 and another programmer via remote telemetry techniques known in the art, such as via a personal area network (PAN), local area network (LAN), wide area network (WAN), public switched telecommunications network (PSTN), or cellular telephone network.

[0055] Therapeutic system 10 can be implemented to provide long-term stimulation therapy to patient 12 over the course of months or years. However, system 10 can also be used experimentally to evaluate treatment prior to full implantation. If implemented temporarily, some components of system 10 may not be implanted in patient 12. For example, patient 12 may be fitted with an external medical device (e.g., a trial stimulator) instead of IMD 16. The external medical device can be coupled to a percutaneous lead or to an implantable lead via a percutaneous extension. If the trial stimulator indicates that DBS system 10 provides effective treatment to patient 12, the clinician can implant a long-term stimulator in patient 12 for relatively long-term treatment.

[0056] Although DBS can successfully reduce the symptoms of some neurological diseases, the stimulation may also cause undesirable side effects (also referred to herein as adverse reactions). Side effects may include incontinence, tingling, imbalance, paralysis, slurred speech, amnesia, loss of self-control, and many other neurological problems. Side effects can range from mild to severe. DBS can cause one or more adverse reactions by inadvertently providing electrical stimulation pulses to anatomical regions near the target anatomical region. These anatomical regions may be referred to as regions associated with adverse stimulation effects. For this reason, the clinician can program the IMD 16 with a treatment program (or multiple treatment programs) that defines stimulation parameter values ​​that balance effective treatment and minimize side effects.

[0057] With the aid of programmer 14 or another computing device, a clinician can select values ​​for treatment parameters, including electrode combinations, for treatment system 10. By selecting specific electrodes 24, 26 for delivering electrical stimulation therapy to patient 12, a clinician can tailor electrical stimulation therapy to one or more specific tissue regions (e.g., specific anatomical structures) within the brain 28 and avoid other tissue regions within the brain 28. In addition, by selecting values ​​for other stimulation parameter values ​​(e.g., amplitude, pulse width, and pulse rate) that define the electrical stimulation signal, a clinician can generate an effective therapy for patient 12 that is delivered via a selected subset of electrodes. Parameter values ​​may vary between patients due to physiological differences, differences in disease conditions, and inaccuracies in lead placement.

[0058] During the programming session, the clinician may determine one or more treatment programs that may provide effective treatment to patient 12. Patient 12 may provide feedback to the clinician regarding the efficacy of the particular program being evaluated, which may include information regarding adverse effects of delivering treatment according to the particular program. In some examples, patient feedback may be used to determine a clinical rating scale score. Once the clinician has identified one or more programs that may be beneficial to patient 12, patient 12 may continue the evaluation process and determine which program best alleviates patient 12's condition or otherwise provides effective treatment to patient 12. Programmer 14 may assist the clinician in creating / identifying treatment programs by providing a method system for identifying potentially beneficial treatment parameter values.

[0059] In accordance with one or more techniques of the present disclosure and as discussed in further detail below, in some examples, a device (e.g., IMD 16, programmer 14, and / or another computing device) may be configured to automatically select an electrode combination from a plurality of electrode combinations for delivering electrical stimulation to a patient. For example, IMD 16 may determine an electrical signal representation for a particular patient based on electrical signals measured between one or more electrode combinations from the plurality of electrode combinations. As an example, IMD 16 may sense electrical signals (e.g., sub-microvolt local field potentials (LFPs)) from electrode pair 24 and / or electrode pair 26.

[0060] Based on the electrical signal representation for a particular patient, the device may select an electrode combination from a plurality of electrode combinations for delivering electrical stimulation to the particular patient. For example, IMD 16 may use a machine learning model to select electrode combination 24, electrode combination 26, and / or electrodes such as so-called outer shell or metal shell electrodes of IMD 16. In some examples, IMD 16 may use a machine learning model based on Figure 5For example, the IMD 16 may use a machine learning model trained based on multiple electrical signal representations for multiple patients. In some examples, each corresponding electrical signal representation in the multiple electrical signal representations is associated with a corresponding electrode combination selected from the multiple electrodes for each corresponding patient in the multiple patients. In some examples, the multiple patients may not include the specific patient, and instead may be patients other than the specific patient. In some examples, the multiple patients may include the specific patient. In this way, the technology of the present disclosure can automate electrode selection.

[0061] IMD 16 may be configured to deliver electrical stimulation to a specific patient via a selected electrode combination. As an example of IMD 16 automatically selecting an electrode combination, IMD 16 may be self-configured to deliver electrical stimulation to a specific patient via a selected electrode combination. As another example of programmer 14 selecting an electrode combination, programmer 14 may configure IMD 16 to deliver electrical stimulation to a specific patient via a selected electrode combination. As yet another example of another computing device selecting an electrode combination, another computing device may output an indication of the selected electrode combination to a clinician, who may input the selected electrode combination into programmer 14, which may configure IMD 16 to deliver electrical stimulation to a specific patient via a selected electrode combination. As yet another example, a computing device may communicate the indication to programmer 14, which may configure IMD 16 to deliver electrical stimulation to a specific patient via a selected electrode combination.

[0062] Figure 2 is a functional block diagram illustrating components of an example IMD 16. Figure 2 In the example shown in , IMD 16 includes processor 60, memory 62, stimulation generator 64, sensing module 66, switch module 68, telemetry module 70, and power supply 72. Memory 62, as well as other memories described herein, may include any volatile or nonvolatile media, such as random access memory (RAM), read-only memory (ROM), nonvolatile RAM (NVRAM), electrically erasable programmable ROM (EEPROM), flash memory, etc. Memory 62 may store computer-readable instructions that, when executed by processor 60, cause IMD 16 to perform various functions described herein.

[0063] exist Figure 2In the example shown in , the memory 62 may store treatment programs 74, operating instructions 76, and an electrode selection module 78, for example, in a separate memory within the memory 62 or in a separate area within the memory 62. Each stored treatment program 74 defines a particular treatment program in terms of corresponding values ​​of electrical stimulation parameters (e.g., electrode combinations, current or voltage amplitudes), and if the stimulation generator 64 generates and delivers stimulation pulses, the treatment program may define values ​​for the pulse width and pulse rate of the stimulation signal. Each stored treatment program 74 may also be referred to as a set of stimulation parameter values. The operating instructions 76 direct the general operation of the IMD 16 under the control of the processor 60, and may include instructions for monitoring brain signals within one or more brain regions and delivering electrical stimulation therapy to the patient 12 via the electrodes 24, 26. As discussed in further detail below and in accordance with one or more techniques of the present disclosure, in some examples, the memory 62 may store an electrode selection module 78, which may include instructions executable by the processor 60 to select one or more electrodes for delivering electrical stimulation. For example, the electrode selection module 78 may be executed by the processor 60 to select electrodes 24 and / or electrodes 26 for use in accordance with the present disclosure. Figure 4 One or more electrodes of the technology to deliver electrical stimulation.

[0064] Stimulation generator 64 generates stimulation signals for delivery to patient 12 via the selected combination of electrodes 24, 26 under the control of processor 60. In some examples, stimulation generator 64 generates stimulation signals and delivers the stimulation signals to brain 28 (via the selected combination of electrodes 24, 26) based on one or more stored treatment programs 74. Figure 1 ) of one or more target areas of the brain 28. The target tissue sites within the brain 28 for stimulation signals or other types of treatments and stimulation parameter values ​​may depend on the patient condition for which the treatment system 10 is implemented to manage. Although stimulation pulses are described, the stimulation signals may be of any form, such as a continuous time signal (e.g., a sine wave), etc.

[0065] The processors described in the present disclosure, including the processor 60, may include one or more digital signal processors (DSPs), general purpose microprocessors, application specific integrated circuits (ASICs), field programmable logic arrays (FPGAs), or other equivalent integrated or discrete logic circuits, or combinations thereof. The functionality attributed to the processors described herein may be provided by hardware devices and may be embodied as software, firmware, hardware, or any combination thereof. The processor 60 is configured to control the stimulation generator 64 according to a treatment program 74 stored by the memory 62 to apply specific stimulation parameter values ​​specified by one or more programs, such as amplitude, pulse width, and pulse rate.

[0066] exist Figure 2In the example shown in , the electrode group 24 of the lead 20A includes electrodes 24A to 24D, and the electrode group 26 of the lead 20B includes electrodes 26A to 26D. The processor 60 can control the switch module 68 to apply the stimulation signal generated by the stimulation generator 64 to the selected electrode combination 24 and / or electrode combination 26. Specifically, the switch module 68 can couple the stimulation signal to the selected conductor within the lead 20, and the conductor then transmits the stimulation signal across the selected electrode 24 and / or electrode 26. The switch module 68 can be a switch array, a switch matrix, a multiplexer, or any other type of switching module configured to selectively couple the stimulation energy to the selected electrode 24 and / or electrode 26 and to selectively sense the bioelectric brain signal using the selected electrode 24 and / or electrode 26. Therefore, the stimulation generator 64 is coupled to the electrode 24 and / or electrode 26 via the switch module 68 and the conductor within the lead 20. However, in some examples, the IMD 16 does not include the switch module 68. For example, in some examples, IMD 16 may include a separate voltage source or current source coupled to each electrode (ie, a separate voltage source and / or current source for each of electrodes 24 and / or electrodes 26).

[0067] As discussed above, the processor 60 can control the switch module 68 to apply the stimulation signal generated by the stimulation generator 64 to the selected electrode combination 24 and / or electrode combination 26. In some examples, the selected electrode combination 24 and / or electrode combination 26 can be monopolar. For example, the selected monopolar combination can include one of the contact combinations of electrode 24 or electrode 26, an electrode on the housing (i.e., housing or metal shell) of IMD16, where one is an anode and the other is a cathode. In some examples, the selected electrode combination 24 and / or electrode combination 26 can be bipolar. As an example, the selected bipolar combination can include two contacts from electrode 24, where one is an anode and the other is a cathode. As another example, the selected bipolar combination can include two electrodes from contact 26, where one is an anode and the other is a cathode. As another example, the selected bipolar combination can include an electrode from electrode 24 and an electrode from electrode 26, where one is an anode and the other is a cathode. In some examples, the selected electrode combination 24 and / or electrode combination 26 can be multipolar. As an example, the selected multipolar combination may include multiple anodes and / or multiple cathodes selected from electrode 24. As another example, the selected multipolar combination may include multiple anodes and / or cathodes selected from electrode 26. As an example, the selected multipolar combination may include multiple anodes and / or multiple cathodes selected from electrode 24 and electrode 26.

[0068] The stimulation generator 64 can be a single-channel or multi-channel stimulation generator. Specifically, the stimulation generator 64 may be capable of delivering a single stimulation pulse, multiple stimulation pulses, or a continuous signal via a single electrode combination at a given time, or delivering multiple stimulation pulses via multiple electrode combinations at a given time. However, in some examples, the stimulation generator 64 and the switch module 68 can be configured to deliver multiple channels on a time-interleaved basis. For example, the switch module 68 can be used to time-divide the output of the stimulation generator 64 between different electrode combinations at different times in order to deliver multiple programs or multiple channels of stimulation energy to the patient 12.

[0069] Sensing module 66 is configured under control of processor 60 to sense bioelectrical brain signals of patient 12 via a selected subset of electrodes 24 and / or electrodes 26 or using one or more electrodes 24 and / or electrodes 26 and at least a portion of conductive housing 34 of IMD 16, electrodes on the housing of IMD 16, or another reference. Processor 60 may control switch module 68 to electrically connect sensing module 66 to selected electrodes 24 and / or electrodes 26. In this way, sensing module 66 may selectively sense bioelectrical brain signals using different electrode combinations 24 and / or electrode combinations 26 (and / or references other than electrodes in electrodes 24 and / or electrodes 26).

[0070] Despite Figure 2 Sensing module 66 is incorporated into common housing 34 with stimulation generator 64 and processor 60, but in other examples, sensing module 66 is in a housing separate from housing 34 of IMD 16 and communicates with processor 60 via wired or wireless communication techniques.

[0071] The telemetry module 70 is configured to support wireless communication between the IMD 16 and an external programmer 14 or another computing device under the control of the processor 60. The processor 60 of the IMD 16 can receive the values ​​of various stimulation parameters (e.g., amplitude and electrode combination) from the programmer 14 as program updates via the telemetry module 70. Updates to the treatment program can be stored in the treatment program 74 portion of the memory 62. The telemetry module 70 in the IMD 16 and the telemetry modules in other devices and systems described herein (e.g., programmer 14) can communicate via RF communication technology. In addition, the telemetry module 70 can communicate with the external medical device programmer 14 via the proximal induced interaction of the IMD 16 using the programmer 14. Therefore, the telemetry module 70 can send information to the external programmer 14 on a continuous basis or at periodic intervals based on a request from the IMD 16 or the programmer 14.

[0072] Power supply 72 delivers operating power to the components of IMD 16. Power supply 72 may include a small rechargeable or non-rechargeable battery and power generation circuitry for generating operating power. Recharging may be accomplished through proximal induced interaction between an external charger and a conductive charging coil within IMD 16. In some examples, the power requirement may be small enough to allow IMD 16 to take advantage of patient motion and implement a kinetic energy extraction device to trickle charge the rechargeable battery. In other examples, conventional batteries may be used for limited periods of time.

[0073] Figure 3 is a functional block diagram showing an example medical device programmer 14 ( Figure 1 ) components. Programmer 14 includes a processor 80, a memory 82, a telemetry module 84, a user interface 86, and a power supply 88. Processor 80 controls user interface 86 and telemetry module 84, and stores and retrieves information and instructions to and from memory 82. Programmer 14 can be configured to be used as a clinician programmer or a patient programmer. Processor 80 may include any combination of one or more processors, including one or more microprocessors, DSPs, ASICs, FPGAs, or other equivalent integrated or discrete logic circuits. Therefore, processor 80 may include any appropriate structure (whether in hardware, software, firmware, or any combination thereof) to perform the functions attributed to processor 80 herein.

[0074] A user (e.g., a clinician or patient 12) can interact with programmer 14 through user interface 86. User interface 86 includes a display (not shown), such as an LCD or LED display or other type of screen, using which processor 80 can present information related to treatment (e.g., electrodes and associated treatment windows). In addition, user interface 86 may include an input mechanism to receive input from the user. The input mechanism may include, for example, any one or more of the following: buttons, a keyboard (e.g., an alphanumeric keyboard), a peripheral pointing device, a touch screen, or another input mechanism that allows a user to navigate through a user interface presented by processor 80 of programmer 14 and provide input. In other examples, user interface 86 also includes audio circuitry for providing audible notifications, instructions or other sounds to patient 12, receiving voice commands from patient 12, or both.

[0075] The memory 82 may also include instructions for operating the user interface 86 and the telemetry module 84 and for managing the power supply 88. Figure 3 In the example shown in , the memory 82 also stores the electrode selection module 78 .

[0076] As discussed in further detail below and in accordance with one or more techniques of the present disclosure, in some examples, memory 62 may store an electrode selection module 78 that may include instructions executable by processor 80 to select one or more electrodes for delivering electrical stimulation. For example, electrode selection module 78 may be executable by processor 60 to select electrodes for use in delivering electrical stimulation based on the electrode selection module. Figure 4 One or more electrodes of the technology to deliver electrical stimulation.

[0077] In some examples, patient 12, a clinician, or another user may otherwise interact with user interface 86 of programmer 14 to manually select a therapy program, generate a new therapy program, modify a therapy program, transfer a new program to IMD 16, or any combination thereof.

[0078] Memory 82 may include any volatile or nonvolatile memory, such as RAM, ROM, EEPROM, or flash memory. Memory 82 may also include a removable memory portion, which may be used to provide memory updates or increases in memory capacity. Removable memory may also allow sensitive patient data to be removed before programmer 14 is used by a different patient.

[0079] Wireless telemetry in programmer 14 may be accomplished through RF communication or proximal induction interaction of external programmer 14 with IMD 16. Such wireless communication is possible through the use of telemetry module 84. Thus, telemetry module 84 may be similar to the telemetry module included within IMD 16. In other examples, programmer 14 may be capable of infrared communication or direct communication via a wired connection. In this way, other external devices may be able to communicate with programmer 14 without the need to establish a secure wireless connection.

[0080] Power supply 88 is configured to deliver operating power to the components of programmer 14. Power supply 88 may include a battery and a power generation circuit for generating operating power. In some examples, the battery may be rechargeable to allow for extended operation. Recharging may be accomplished by electrically coupling power supply 88 to a cradle or plug connected to an alternating current (AC) outlet. In addition, recharging may be accomplished by proximal induced interaction between an external charger and an inductive charging coil within programmer 14. In other examples, conventional batteries (e.g., nickel-cadmium batteries or lithium-ion batteries) may be used. In addition, programmer 14 may be directly coupled to an AC outlet for operation.

[0081] Although various information is illustrated and described as being stored in memory 82 of programmer 14, it will be understood that some or all of such information may alternatively or additionally be stored within memory 62 of IMD 16. Furthermore, as discussed below, at least some of the functionality attributed to processor 80 of programmer 14 may alternatively or additionally be attributed to processor 60 of the IMD (and vice versa).

[0082] Figure 4 is a conceptual diagram showing further details of an example of an electrode selection module 78 according to one or more techniques of the present disclosure. As discussed above, the electrode selection module 78 may be included in either the IMD 16 or the programmer 14, or any other computing device, and may include instructions for execution by one or more processors to perform electrode selection as described in the present disclosure. When the functionality of the electrode selection module 78 is described below, the description refers to the execution of the instructions associated with the electrode selection module.

[0083] According to one or more techniques of the present disclosure, electrode selection module 78 can receive an electrical signal representation for a particular patient. Figure 4 , contact selection module 78 may receive local field potential (LFP) data 92. In some examples, LFP data 92 may be generated by measuring LFP across various electrode combinations, such as electrodes 24 and / or electrodes 26 of IMD 16.

[0084] like Figure 4 As shown in , the electrode selection module 78 may include multiple machine learning models 94A to 94N (collectively referred to as "machine learning models 94"). In some examples, each model in the model 94 can be a support vector machine (SVM) classifier trained based on one or more electrical signal representations of multiple electrical signal representations for multiple patients. Each machine learning model in the machine learning model 94 can be associated with a specific electrode combination in the multiple electrodes. For example, the machine learning model 94A can be associated with the electrode 24A (i.e., when the electrode 24A is selected for monopolar stimulation, the machine learning model 94A can be trained based on the sensed electrical signal representation); the machine learning model 94B can be associated with the electrode 24B (i.e., when the electrode 24B is selected for monopolar stimulation, the machine learning model 94B can be trained based on the sensed electrical signal representation); and the machine learning model 94N can be associated with both the electrode 24A and the electrode 24C (i.e., when the electrode 24A and the electrode 24C are selected for bipolar stimulation, the machine learning model 94N can be trained based on the sensed electrical signal representation).

[0085] Each of the machine learning models 94 can be configured to determine a corresponding score (respectively scores 96A to 96N, collectively referred to as "scores 96") indicating the degree to which the electrical signal pattern for a particular patient is similar to electrical signal patterns previously obtained from multiple patients that share the fact that the particular electrode selection used to deliver electrical stimulation produced effective treatment for the patient. For example, machine learning model 94A can determine score 96A, which indicates the degree to which the patient's electrical signal pattern is similar to the pattern from a patient who received effective electrical stimulation treatment via electrode 24A; machine learning model 94B can determine score 96B, which indicates the degree to which the patient's electrical signal pattern is similar to the pattern from a patient who received effective electrical stimulation treatment via electrode 24B; machine learning model 94N can determine score 96N, which indicates the degree to which the patient's electrical signal pattern is similar to the pattern from a patient who received effective electrical stimulation treatment via a combination of electrodes 24A and electrodes 24C.

[0086] like Figure 4 As shown in , the electrode selection module 78 may include a selector 98, which may be configured to select an electrode combination for delivering electrical stimulation to a patient. For example, the selector 98 may select an electrode combination based on score 96. In some examples, the higher the score determined by the machine learning model, the more likely it is that the electrode combination associated with the machine learning model will be beneficial for delivering therapy to a particular patient. As an example, the selector 98 may select a contact or electrode combination associated with the machine learning model that determines the highest score in score 96 in machine learning model 94. For example, where score 96A is 0.9, score 96B is 0.43, and score 96N is 0.56, the selector may select the electrode combination associated with machine learning model 94A (i.e., contact 24A) to deliver electrical stimulation to a particular patient. As shown in FIG. Figure 4 As shown, selector 98 can output the selected electrode combination as selected electrode 100.

[0087] As discussed above, IMD 16 can be configured to deliver electrical stimulation to a specific patient via a selected electrode combination. Electrode selection module 78 can generate information for controlling the delivery of electrical stimulation via a selected electrode combination. As an example in which electrode selection module 78 is included in IMD 16, electrode selection module 78 can cause IMD 16 to self-configure for delivering electrical stimulation to a specific patient via selected electrodes 100. As another example in which electrode selection module 78 is included in programmer 14, electrode selection module 78 can cause programmer 14 to configure IMD 16 to deliver electrical stimulation to a specific patient via a selected electrode combination. As yet another example in which electrode selection module 78 is included in another computing device (i.e., not IMD 16 or programmer 14), electrode selection module 78 can cause another computing device to output an indication of the selected electrode combination to a clinician, who can input the selected contact combination into programmer 14, who can configure IMD 16 to deliver electrical stimulation to a specific patient via a selected electrode combination. As yet another example, electrode selection module 78 may cause computing device to communicate the indication to programmer 14, which may configure IMD 16 to deliver electrical stimulation to the particular patient via the selected electrode combination.

[0088] Figure 5 is a flow chart of an example technique for automatically selecting an electrode combination for delivering electrical stimulation in accordance with one or more techniques of this disclosure. Figure 5 The techniques are primarily described as being performed by processor 60 of IMD 16, but in other examples, another processor (alone or in conjunction with processor 60) may perform Figure 5 For example, the processor 80 of the programmer 14 or a processor of another computing device (alone or in conjunction with the processor 80) may execute Figure 5 any part of the technology.

[0089] like Figure 5As shown in , a device such as processor 60 of IMD 16 can determine an electrical signal representation for a particular patient (502). As discussed above, a therapeutic system such as therapeutic system 10 can be configured to sense a patient's bioelectrical brain signals. For example, processor 60 can measure the local field potential (LFP) of a particular patient's brain at both ends of different electrode combinations 24 and 26 using one or more of electrodes 24, 26 of leads 20A and 20B, respectively. U.S. application serial number 14 / 195,489 filed by Kaemmerer on March 3, 2014, "THERAPEUTIC WINDOWDETERMINATION" (the entire contents of which are incorporated herein by reference) describes how electrodes can be used to both sense LFPs and deliver electrical stimulation.

[0090] The processor 60 may then select an electrode combination for delivering electrical stimulation therapy to a particular patient based on the electrical signal representation for the particular patient and the multiple electrical signal representations for multiple patients (504). U.S. Patent No. 8,428,733, "STIMULATION ELECTRODE SELECTION," issued to Carlson et al. on April 23, 2013 (the entire contents of which are incorporated herein by reference) describes how a stimulation electrode combination for delivering stimulation to a patient may be selected based on bioelectric signals sensed within the patient's brain. In some examples, the processor 60 may use one or more machine learning models to select an electrode combination. For example, the processor 60 may utilize multiple support vector machines (SVMs) (each of the multiple support vector machines being associated with a specific electrode or electrode combination among the multiple electrodes) to select an electrode or electrode combination for a particular patient. In some examples, each corresponding electrical signal representation in the multiple electrical signal representations is associated with a corresponding electrode or electrode combination selected from the multiple electrodes for each corresponding patient in the multiple patients, and each SVM in the multiple SVMs can be trained based on one or more electrical signal representations in the multiple electrical signal representations.

[0091] In some examples, each of the multiple SVMs can be trained based on signals in the theta band and / or signals in the beta band of the multiple electrical signal representations. In some examples, each of the multiple SVMs can be trained based on electrical signal representations corresponding to electrical signals between less than all possible electrode combinations in the multiple electrical signal representations. For example, in the presence of twelve electrode combinations, each of the multiple SVMs can be trained based on electrical signal representations corresponding to electrical signals between six electrode combinations in the multiple electrical signal representations.

[0092] In some examples, the processor 60 can be configured to use the multiple SVMs by determining a corresponding score for each corresponding SVM in the multiple SVMs, the score indicating the degree to which the patient's electrical signal pattern is similar to the pattern of a patient who received effective electrical stimulation therapy from the corresponding electrode combination associated with the corresponding SVM. In some examples, the greater the degree to which the electrical signal representation for a particular patient is similar to the pattern of a patient who received effective electrical stimulation therapy using the corresponding electrode combination, the greater the likelihood that delivering therapy via the corresponding electrode combination will benefit the particular patient. In this way, the processor 60 can be configured to select an electrode combination for delivering stimulation to a particular patient based on the score.

[0093] In some examples, processor 60 may control the stimulator to deliver electrical stimulation to the specific patient via the selected electrode combination (506). In some examples, processor 60 may control the stimulator to avoid delivering electrical stimulation to the specific patient via the selected electrode combination until approval is received from the clinician and / or the specific patient. In some examples, processor 60 may begin delivering electrical stimulation to the specific patient via the selected electrode combination without receiving approval from the clinician and / or the specific patient.

[0094] In some examples, processor 60 may periodically execute Figure 5 For example, processor 60 may periodically (e.g., hourly, daily, weekly, monthly, etc.) determine an updated electrical signal representation for a particular patient, select an updated electrode combination based on the updated electrical signal representation, and deliver electrical stimulation to the particular patient via the updated electrode combination. In this way, the techniques of the present disclosure may enable processor 60 to automatically adapt to changes in the brain of a particular patient over time.

[0095] For example, at a first time, the processor 60 may determine an electrical signal representation for a particular patient, select an electrode combination based on the electrical signal representation, and deliver electrical stimulation to the particular patient via the electrode combination. For example, at a later second time, the processor 60 may determine an updated electrical signal representation for a particular patient, select an updated electrode combination based on the updated electrical signal representation, and deliver electrical stimulation to the particular patient via the updated electrode combination. In some examples where the brain of the particular patient has not changed between the first time and the second time, the updated electrode combination may be the same as the electrode combination selected based on the electrical signal representation determined at the first time. In some examples where the brain of the particular patient has changed between the first time and the second time, the updated electrode combination may be different from the electrode combination selected based on the electrical signal representation determined at the first time.

[0096] Figure 6is a flowchart of an example technique for training one or more machine learning models in accordance with one or more techniques of this disclosure. Figure 6 The techniques are primarily described as being performed by processor 80 of programmer 14, but in other examples, another processor (alone or in conjunction with processor 80) may perform Figure 6 For example, processor 60 of IMD 16 or a processor of another computing device (alone or in conjunction with processor 80) may execute Figure 6 any part of the technology.

[0097] In some examples, the device can perform data collection and management to develop and train one or more machine learning models. Figure 6 As shown in , a device such as the processor 80 of the programmer 14 can determine multiple electrical signal representations, each of which is associated with a corresponding electrode combination (602). In the case where a clinician selects a specific electrode combination during the same session as the session in which the electrical signal representation is sensed (or temporally close to (e.g., within 1 hour, 1 day, etc.) the time when the electrical signal representation is sensed), the electrical signal representation can be associated with a specific electrode combination. As discussed above, a therapeutic system such as the therapeutic system 10 can be configured to sense a patient's bioelectric brain signal or another physiological parameter. An example of a therapeutic system that can be configured to sense bioelectric brain signals is the Activa PC+S produced by Medtronic Inc. In some examples, a therapeutic system (e.g., Activa PC+S) configured to sense bioelectric brain signals can be implanted in multiple patients (e.g., patients with congenital Parkinson's disease). In each of the multiple patients, the therapeutic system can be attached to a DBS lead (e.g., model 3389 produced by Medtronic Inc.) positioned in the patient's STN.

[0098] An implantable therapeutic system can be used to generate the multiple electrical signal representations. For example, during device implantation and follow-up sessions (e.g., extended to six months or more), records from DBS leads can be obtained / saved. In some examples, the records can be obtained when the patient is at rest and in a drug-free state. In some examples, the set of records obtained on a single occasion can be referred to as a "montage" or electrical signal representation. An example montage can include 30 to 60 second bipolar records between multiple electrode combinations. For example, in the case where the electrode includes four electrodes (E0, E1, E2, E3), an example montage can include 30 to 60 bipolar records between six combinations of four electrodes (e.g., E0-E1, E0-E2, E0-E3, E1-E2, E1-E3, E2-E3), the records are sampled at about 422Hz and hardware filtered between about 0.5Hz and about 100Hz.

[0099] In some embodiments, more than one electrode can be implanted in a single patient. For example, a first electrode can be implanted in the left hemisphere of the patient's brain and a second electrode can be implanted in the right hemisphere of the patient's brain. In such examples, the electrodes can be numbered sequentially. For example, the electrodes (C) of the first electrode can be referred to as C0, C1, C2, and C3, and the contacts of the second electrode can be referred to as C4, C5, C6, and C7. In other examples, the electrodes of the second electrode can be referred to as C8, C9, C10, and C11.

[0100] In some examples, at the end of each clinical visit, a clinician (e.g., a neurologist) can program the DBS treatment of a patient by selecting, among other parameters, the stimulation electrode(s) (e.g., C0, C1, C2, C3, or some combination thereof, or two or more electrodes therein, along with corresponding polarities) for providing stimulation. In this way, each montage (e.g., electrical signal representation) in the montage can be associated with a corresponding electrode (e.g., electrode selected by the clinician) combination. Table (1) below shows which electrodes were selected by the clinician during the clinical visit when montages for fifteen patients were recorded. Because Table (1) includes more than fifteen data points, some patients had montages and electrode selections recorded on multiple occasions. As shown in Table (1), most of the time, the clinician selected electrode C1 to deliver monopolar (also known as unipolar) stimulation (i.e., stimulation delivered by, for example, a single anode on the stimulator housing and a single cathode on the lead, or vice versa). However, the data set shown in Table (1) represents the selection of electrodes C2, C3 and bipolar stimulation using both C1 and C2. As discussed above, in some examples, the housing of the implanted device (e.g., housing 34 of IMD 16) can include one or more stimulation electrodes that can function as an anode or cathode for delivering stimulation.

[0101] Table (1) Counts of patients and montage records

[0102]

[0103] In some examples, spectral analysis can be performed to identify features that can be used to train one or more machine learning models. For example, for each of six records in a single montage, a power spectrum can be calculated using a Hamming window of length 1024 with 50% overlap (given a frequency resolution of 0.41 Hz). The resulting power spectral density (PSD) can be used to calculate the average power in multiple frequency bands (e.g., 3-5 Hz, 5-10 Hz, and every 10 Hz between 10 Hz and 90 Hz). In some examples, frequencies below 3 Hz can be eliminated due to artifacts of baseline drift and hardware noise centered at 2.5 Hz. The average power in multiple frequency bands (e.g., the 11 frequency bands described above) of one or more of the records in the montage can be used to create a feature set for subsequent analysis. For example, where a montage includes six records, one or more of the six records can be used to create a feature set.

[0104] 7A to 7C Some examples of data that may be used to create feature sets for classification in accordance with one or more techniques of this disclosure are presented. Fig. 7A It is an instruction Figure 7B and Figure 7C The data in the figure represent which electrodes. Fig. 7A Different wire types corresponding to electrode combinations are shown with Figure 7B and Figure 7C For example, if Fig. 7A As shown, the thin solid lines correspond to the E0-E1 electrode combination, and the thick solid lines correspond to the E0-E3 electrode combination. Figure 7B and Figure 7C Each of the graphs in the diagram includes a horizontal axis representing frequency in Hertz (Hz) and a vertical axis representing millivolts squared per Hertz. Power in units; and a plurality of plots showing the power from having ( Figure 7B ) and does not have ( Figure 7C ) An example montage of relatively high amounts of beta band activity (compared to activity in other frequency bands) of power versus frequency for bipolar electrode pairs from two different patients, both of whom had electrode El as their clinician-selected treatment. For example, Fig. 7A and Figure 7B The thin solid line plots in show the relationship between power and frequency for electrodes E0 and El, respectively, for each of the two patients. The 3-5 Hz theta frequency band and the 10-20 Hz beta frequency band are highlighted.

[0105] In some examples, many machine learning algorithms / models can be used to enable the device to select the electrode combination of the user patient. Some example machine learning algorithms / models that can be used include but are not limited to linear discriminant analysis, k nearest neighbor classification, classification tree, and support vector machine (SVM). Classifier performance can be evaluated based on the number of errors generated in the leave-one-out (LOO) cross-validation scheme. In some examples, the SVM method using the radial basis kernel may produce the least classification error. In this way, the SVM method can be used as a classifier. Before being used to train the classifier, the features (power and bipolar pairs in the frequency band) can be standardized in the training observation. SVM can only perform binary classification. However, there are more than one potential electrode combination to be selected. In this way, a separate SVM classifier can be trained by comparing the observations in the group (e.g., those montages associated with a specific electrode combination) with the observations not in the group (e.g., those montages not associated with a specific electrode combination). In order to predict a new observation group, the score representing the possibility in the group [0,1] is calculated for each possible group in the possible group, and the observation is classified as in the group with the maximum score.

[0106] In some examples, using a large number of features (relative to the number of observations) as input to the classification problem may produce non-generalizable results. In this way, it may be desirable to identify the minimum number of features that still achieve good classification performance. In some examples, in order to objectively select the best features, each feature can be used individually for training the classifier, and the LOO performance can be compared. In some examples, the features that cause the worst performance can be eliminated continuously until the minimum feature subset that can fully classify the data set is left. In some examples, the number of bipolar recording pairs can be reduced from six to three by ignoring the pairs whose elimination does not reduce the classifier performance.

[0107] In some examples, visual inspection of the power spectra of all montages may not reveal an obvious way to classify them into groups corresponding to clinically selected stimulation electrodes. In one example, training an SVM to classify montages using 10 frequency bands (3-5 Hz, 5-10 Hz, 10-20 Hz, ..., 80-90 Hz) and six recording channels (electrode pairs) yielded a leave-one-out cross-validation error of 6 / 83 = 7% using a total of 60 features.

[0108] Fig. 8A and Figure 8B Leave-One-out (LOO) classification error according to one or more techniques of the present disclosure is shown. Fig. 8A Included is a horizontal axis representing frequency bands, a vertical axis representing the number of LOO errors, and a plot showing the LOO errors using average power in corresponding frequency bands (on the x-axis) in all six dipole pairs. Figure 8B 802 to illustrate the LOO error using the average power in two frequency bands among all six dipole pairs. Figure 8B In , because the bands do not pair with themselves, the values ​​on the diagonal do not apply; instead, Fig. 8A The LOO errors obtained for individual frequency bands not paired with different frequency bands are shown in .

[0109] To reduce the number of features, the individual frequency bands can be compared by using all pairs of records (e.g. Fig. 8AThe power in this band in the six features shown (a total of six features) is used to train a classifier to help determine the value (i.e., significance) of this band for classification purposes. In some examples, the smallest LOO error (18 / 83) can be produced by using power in the 3-5 Hz theta band, while using only power in the 10-20 Hz beta band can produce 23 / 83 LOO errors. A classifier using one of the high gamma bands above 60 Hz may perform poorly (e.g., more than 28 / 83 errors). Next, all recording pairs (e.g., a total of 12 features, Figure 8B ) to train the classifier. Although the 10-20 Hz beta band may not perform well alone, the combination of 3-5 Hz theta and 10-20 Hz beta can produce the least LOO error (5 / 83, Figure 8B In fact, a classifier trained on this subset of frequencies may perform better than a classifier using all 60 features.

[0110] If there is redundant information in the recording channel, some recording channels can be omitted and the size of the feature set can be further reduced. Following this strategy, 3-5Hz Theta and 10-20Hz Beta features from three bipolar pairs in these six bipolar pairs can be used to train the classifier. Table (2) below shows the example LOO errors of different frequency bands and different bipolar pairs. As shown in Table (2) and when using 3-5Hz Theta band and 10-20Hz Beta band, 8 / 83 LOO errors can be generated using three pairs (E0-E1, E0-E2, E0-E3) with reference to E0; 15 / 83 LOO errors can be generated using three adjacent pairs (E0-E1, E1-E2, E2-E3); Minimum error (7 / 83) can be generated with reference to three pairs (E0-E3, E1-E3, E2-E3) with reference to E3 and only six features can be used to produce the best performance.

[0111] Table 2 - Error rates using feature subsets

[0112]

[0113] Thus, a set of six features of three bipolar pairs including reference E3 and the two frequency bands 3-5 Hz theta and 10-20 Hz beta may contain sufficient information to identify the stimulation electrode group to which a given observation belongs, with an accuracy exceeding 91% in some cases.

[0114] 9A to 9C is a scatter plot showing beta (10-20 Hz) power versus theta (3-5 Hz) power recorded on bipolar pairs E0-E3, E1-E3, and E2-E3 for all montages in accordance with one or more techniques of the present disclosure. Fig.9D According to one or more techniques of the present disclosure, 9A to 9C As shown in the figure. Fig.9D As shown, 9A to 9C The solid black circles in correspond to the recorded powers when E1 and E2 were manually selected (e.g., by a clinician) for delivering electrical stimulation; 9A to 9C The circles with vertical lines in them correspond to the recorded powers when E3 was manually selected for delivering electrical stimulation; 9A to 9C The dotted circles correspond to the recorded powers when E2 was manually selected for delivering electrical stimulation; and 9A to 9C The open circles in correspond to the recorded powers when E1 was manually selected for delivering electrical stimulation.

[0115] However, if 9A to 9C As shown, examining the data in groups with respect to these features may not reveal a way to classify the observations by visual inspection alone. Thus, in some examples, it is desirable to utilize a computational approach to, for example, assist a practitioner in selecting stimulation electrodes for a patient. In accordance with one or more techniques of the present disclosure, such a computation may consist of a set of SVMs, each of which outputs a likelihood score that a particular electrode configuration selection is similar to electrode configuration selections that have been made for other cases previously (i.e., based on a training set).

[0116] Now return to reference Figure 6 And according to one or more techniques of the present disclosure, a device such as processor 80 of programmer 14 can train one or more machine learning models based at least on the plurality of electrical signal representations (604). For example, each SVM in a plurality of SVMs (each SVM associated with a particular electrode combination) can be trained to distinguish between montages (i.e., electrical signal representations) associated with a particular electrode combination (i.e., a montage if the clinician selects the particular electrode combination) and all other montages (if the clinician does not select the particular electrode combination associated with this SVM).

[0117] The SVM may be considered to be associated with a particular electrode combination when the SVM is configured to determine a score indicating the extent to which the pattern of sensed electrical signals from the patient is similar to patterns of sensed electrical signals previously obtained from multiple patients for whom delivery of therapy via the particular electrode combination would be beneficial. Figure 4A particular SVM, such as machine learning model 94A, can be associated with a particular electrode combination, such as electrode 24A. In this way, a particular SVM can be trained using electrical signal representations sensed during the same session as (or sensed in time close to) the session when the clinician selected the particular electrode combination to deliver therapy.

[0118] Fig. 10A and Fig. 10B is a diagram illustrating the output of a machine learning model for selecting an electrode or electrode combination according to one or more techniques of the present disclosure. Fig. 10A Scores generated by a machine learning model (e.g., SVM) using six features (three bipolar records, two frequency bands) are shown that are related to the fit of the observation to the group of the machine learning model. Fig. 10B Shows how each observation was classified into the group for which the machine learning model output had the highest score. Each dot represents the clinician-selected electrode for each observation. The black X shows an observation that was misclassified, and the vertical position shows the incorrectly predicted group.

[0119] exist Fig. 10A and Fig. 10B In the example, the classifier consists of four SVMs. Fig. 10A The output of each SVM (a score indicating the degree of similarity to its corresponding classification, Y-axis) is shown for each observation in our dataset (X-axis). The classifier output can be assigned to the group with the highest score as a Winner-Take-All. For example, because the C1 score is close to 1 and the other group scores are close to 0 ( Fig. 10B ), so observation 1 is correctly classified as group C1. However, observation 12 (vertical line) is misclassified as C1 because it has the highest score (0.226) in this group, but it is actually in group C2, which the classifier gave a score of 0.183.

[0120] In some examples, because beta oscillations (recorded from microelectrodes or macroelectrodes) may be associated with the location of active stimulation electrodes independently selected for optimal clinical outcomes, the power in the beta band can be used to provide some of the information used to identify the electrodes selected by the clinician for DBS therapy delivery. However, in some examples, beta power alone may not be sufficient to identify the therapeutic stimulation electrodes. Thus, in some examples, the theta band (3-5 Hz) can also be used to improve classifier performance. One possible explanation for this observation is that patients were not "screened" for significant beta peaks. The values ​​of theta band are consistent with the report of Zaidel et al. (Zaidel A., Spivak A., Bergman H., and Z. Israel, "Subthalamic Span of β Oscillations Predicts Deep Brain Stimulation Efficacy for Patients with Parkinson's Disease", Brain, p. awq144, 2010) (low frequency (3-7 Hz) power also increased in the dorsolateral oscillatory region of STN), and thus in their study, were related to the location of the selected active electrode. If the stimulation electrode location selected for optimal clinical effect is associated with the dorsolateral region of STN, further research may be necessary to understand.

[0121] According to one or more techniques of the present disclosure, it is possible to develop algorithms that use LFP recordings from DBS leads like the Activa PC+S system to guide the identification of electrodes closest to the "physiological sweet spot" for effective DBS stimulation in the STN. Combined with post-implantation brain images of the patient's lead position in the STN for use as part of visually guided DBS programming, or used in real time during lead implantation or closed-loop therapy delivery, this information from brain sensing holds promise for making DBS therapy programming more efficient and effective.

[0122] In light of the foregoing, the present disclosure contemplates a device comprising a memory and one or more processors, the memory being configured to store a sensed electrical signal representation for a specific patient. The one or more processors may be configured to select an electrode combination from a plurality of combinations of one or more implantable electrodes for delivering electrical stimulation therapy to the specific patient based on the sensed electrical signal representation for the specific patient and a plurality of sensed electrical signal representations for a plurality of other patients, and to generate information for controlling the delivery of the electrical stimulation based on the selected combination of the electrodes.

[0123] In an embodiment of the device as described in the preceding paragraph, each corresponding sensed electrical signal representation in the plurality of sensed electrical signal representations may be associated with a corresponding electrode combination of the plurality of electrodes selected for delivering electrical stimulation to each corresponding patient in the plurality of patients. Alternatively or additionally, in order to select the electrode combination, the one or more processors of the device may be configured to use one or more machine learning models trained based on one or more sensed electrical signal representations in the plurality of sensed electrical signal representations to select the electrode combination for delivering electrical stimulation to the particular patient.

[0124] In an embodiment in which the one or more processors of the device are configured to use one or more machine learning models trained based on one or more of the plurality of sensed electrical signal representations to select the electrode combination for delivering electrical stimulation to the specific patient, the device may be further optionally configured so that the one or more machine learning models include a plurality of support vector machines (SVMs), each of which is associated with a specific electrode combination of the plurality of electrodes. According to this embodiment, in order to select the electrode combination, the one or more processors are configured to determine a corresponding score for each corresponding SVM of the plurality of SVMs, the score indicating the degree to which the sensed electrical signal representation for the specific patient is similar to the sensed electrical signal representation from a plurality of patients, and the corresponding electrode combination associated with the corresponding SVM has been therapeutically effective for delivering electrical stimulation to the plurality of patients. Additionally, the one or more processors are configured to select the electrode combination for delivering stimulation to the specific patient based on the score. In this embodiment, each corresponding sensed electrical signal representation of the plurality of sensed electrical signal representations may be optionally associated with a corresponding electrode combination of the plurality of electrodes selected for delivering electrical stimulation to each corresponding patient of the plurality of patients.

[0125] In any of the embodiments described in the preceding paragraphs, in one embodiment, the device may be further optionally configured such that the machine learning model is trained based on the signals in the theta frequency band and the signals in the beta frequency band represented by the multiple electrical signals, and wherein, in order to select the electrode combination, the one or more processors are configured to select the electrode combination for delivering electrical stimulation to the specific patient based on the signals in the theta frequency band and the signals in the beta frequency band represented by the electrical signals for the specific patient. Alternatively, in a different embodiment, the device described in the preceding paragraphs may be configured such that the machine learning model is trained based on the signals in the theta frequency band represented by the multiple electrical signals or the signals in the beta frequency band represented by the multiple electrical signals, and wherein, in order to select the electrode combination, the one or more processors are configured to select the electrode combination for delivering electrical stimulation to the specific patient based on the signals in the theta frequency band represented by the electrical signals for the specific patient and the signals in the beta frequency band represented by the electrical signals for the specific patient.

[0126] In any of the foregoing embodiments, the one or more processors may be included in a device implanted in the specific patient configured to deliver electrical stimulation to the specific patient using the one or more implantable electrodes, and the one or more processors may be further configured to determine the electrical signal representation for the specific patient based on electrical signals measured between one or more contact combinations in the plurality of electrode combinations. In an alternative embodiment, the device is configured to program a device implanted in the specific patient, and wherein, in order to generate information for controlling the delivery of the electrical stimulation based on the selected combination of electrodes, the one or more electrodes are configured to program the device implanted in the specific patient to deliver the electrical stimulation to the specific patient via the selected electrode combination.

[0127] In any of the foregoing embodiments, in order to determine the electrical signal representation for the specific patient, the one or more processors may optionally be configured to periodically determine an updated electrical signal representation for the specific patient, and wherein the one or more processors are further configured to automatically select an updated electrode combination from the multiple electrode combinations for delivering electrical stimulation based on the updated electrical signal representation for the specific patient and to deliver electrical stimulation to the specific patient via the updated electrode combination.

[0128] In another example, a computer-readable storage medium storing instructions is disclosed that, when executed, causes one or more processors of a device to select an electrode combination from multiple combinations of one or more implantable electrodes for delivering electrical stimulation to a specific patient based on a sensed electrical signal representation for the specific patient and a plurality of sensed electrical signal representations for a plurality of other patients, and to generate information for controlling the delivery of the electrical stimulation based on the selected combination of the electrodes.

[0129] Optionally, the device as described in the previous paragraph is a programmer for a device implanted in the specific patient, and the instructions may cause the one or more processors to generate information for controlling the delivery of the electrical stimulation based on the selected combination of electrodes. The instructions may cause the one or more processors to program the device implanted in the specific patient to deliver the electrical stimulation to the specific patient via the selected combination of electrodes. Alternatively, the device as described in the previous paragraph may be implanted in the specific patient, and the instructions may cause the one or more processors to generate information for controlling the delivery of the electrical stimulation based on the selected combination of electrodes.

[0130] According to another embodiment, a system including one or more processors and a medical device is disclosed, wherein the one or more processors are configured to select an electrode combination from multiple combinations of one or more implantable electrodes for delivering electrical stimulation to the specific patient based on a sensed electrical signal representation for the specific patient and a plurality of sensed electrical signal representations for a plurality of other patients, and the medical device is configured to deliver stimulation to the specific patient via the selected electrode combination.

[0131] Although the techniques described above are primarily described as being performed by processor 60 of IMD 16 or processor 80 of programmer 14, in other examples, one or more other processors may perform any portion of the techniques described herein alone or in addition to processor 60 or processor 80. Thus, references to a "processor" may refer to "one or more processors." Likewise, "one or more processors" may refer to a single processor or multiple processors in different examples.

[0132] The techniques described in this disclosure, including those attributed to IMD 16, programmer 14, or various component parts, may be implemented, at least in part, in hardware, software, firmware, or any combination thereof. For example, various aspects of the techniques may be implemented in one or more processors, including one or more microprocessors, DSPs, ASICs, FPGAs, or any other equivalent integrated or discrete logic circuits, as well as any combination of such components, embodied in a programmer such as a clinician or patient programmer, a medical device, or other device.

[0133] In one or more examples, the functions described in the present disclosure may be implemented in hardware, software, firmware, or any combination thereof. If implemented in software, the functions may be stored as one or more instructions or codes on a computer-readable medium and executed by a hardware-based processing unit. Computer-readable media may include computer-readable storage media forming a tangible, non-volatile medium. Instructions may be executed by one or more processors, such as one or more DSPs, ASICs, FPGAs, general-purpose microprocessors, or other equivalent integrated or discrete logic circuits. Therefore, as used herein, the term "processor" may refer to one or more of the above structures or any other structures suitable for implementing the techniques described herein.

[0134] In addition, in some aspects, the functions described herein can be provided in dedicated hardware and / or software modules. Depicting different features as modules or units is intended to highlight different functional aspects and does not necessarily imply that such modules or units must be implemented by separate hardware or software components. On the contrary, the functions associated with one or more modules or units can be performed by separate hardware or software components, or integrated in public or separate hardware or software components. In addition, the technology can be fully implemented in one or more circuits or logic elements. The technology of the present disclosure can be implemented in a wide range of equipment or devices, including an IMD, an external programmer, a combination of an IMD and an external programmer, an integrated circuit (IC) or an IC set, and / or a discrete circuit resident in an IMD and / or an external programmer.

[0135] A number of different examples of the disclosure have been described. These and other embodiments are within the scope of the following claims.

Claims

1. A medical system, comprising a programmer device, the programmer device comprising: a memory configured to store patient-specific sensed electrical signals; as well as One or more processors configured to: selecting an electrode combination from a plurality of combinations of implantable electrodes for delivering electrical stimulation therapy to the particular patient based on a comparison between the sensed electrical signal for the particular patient and sensed electrical signals for a plurality of other patients; as well as An implantable medical device implanted in the specific patient is programmed to deliver the electrical stimulation therapy to the specific patient via a selected electrode combination, wherein, in order to program the implantable medical device, the one or more processors are configured to transmit programming information indicating the selected electrode combination to the implantable medical device implanted in the specific patient.

2. The medical system of claim 1, wherein: The one or more processors are configured to select the electrode combination based on a comparison between the sensed electrical signal representation for the particular patient and a plurality of sensed electrical signal representations for a plurality of other patients.

3. The medical system of claim 1, wherein: Each corresponding one of the sensed electrical signals is associated with a corresponding one of the plurality of electrode combinations selected for delivering electrical stimulation to each corresponding one of the plurality of patients.

4. The medical system of claim 3, wherein: To select the electrode combination, the one or more processors are configured to: The electrode combination for delivering electrical stimulation to the particular patient is selected using one or more machine learning models trained based on one or more of the sensed electrical signals.

5. The medical system of claim 4, wherein: The one or more machine learning models include a plurality of support vector machines (SVMs), each of the plurality of support vector machines being associated with a particular electrode combination of the plurality of electrode combinations, and wherein, to select the electrode combination, the one or more processors are configured to: determining, for each corresponding SVM of the plurality of SVMs, a corresponding score indicating the extent to which the sensed electrical signal for the particular patient is similar to the sensed electrical signals for a plurality of patients for which the corresponding electrode combination associated with the corresponding SVM has been therapeutically effective for delivering electrical stimulation; and The electrode combination for delivering electrical stimulation to the particular patient is selected based on the score.

6. The medical system of claim 1, further comprising the implantable medical device, wherein: The implantable medical device is configured to deliver electrical stimulation via an active electrode combination, and wherein, in response to being programmed to deliver the electrical stimulation therapy to the particular patient via the selected electrode combination, the implantable medical device is configured to set the selected electrode combination as the active electrode combination.

7. A computer readable storage medium storing instructions which, when executed, cause one or more processors of a programmer device to: selecting an electrode combination from a plurality of combinations of implantable electrodes for delivering electrical stimulation therapy to a particular patient having a movement disorder based on a comparison between the sensed electrical signals for the particular patient and the sensed electrical signals for a plurality of other patients; and programming a device implanted in the particular patient to deliver the electrical stimulation therapy to the particular patient via a selected combination of electrodes, wherein: The instructions that cause the one or more processors to program the device include instructions that cause the one or more processors to transmit programming information indicative of the selected electrode combination to the device implanted in the particular patient.

8. The computer-readable storage medium of claim 7, wherein: The instructions causing the one or more processors to select include instructions causing the one or more processors to select the electrode combination based on a comparison between a sensed electrical signal representation for a particular patient and a plurality of sensed electrical signal representations for a plurality of other patients.

9. The computer-readable storage medium of claim 7, wherein: Each corresponding one of the sensed electrical signals is associated with a corresponding one of the plurality of electrode combinations selected for delivering electrical stimulation to each corresponding one of the plurality of other patients.

10. The computer-readable storage medium of claim 9, wherein: The instructions causing the one or more processors to select the electrode combination include instructions causing the one or more processors to: The electrode combination for delivering electrical stimulation to the particular patient is selected using one or more machine learning models trained based on one or more of the sensed electrical signals.

11. The computer-readable storage medium of claim 10, wherein: The one or more machine learning models include a plurality of support vector machines (SVMs), each of the plurality of support vector machines being associated with a particular electrode combination of the plurality of combinations of the one or more electrodes, and wherein the instructions causing the one or more processors to select the electrode combination include instructions causing the one or more processors to: determining, for each corresponding SVM of the plurality of SVMs, a corresponding score indicating the extent to which the sensed electrical signal for the particular patient is similar to the sensed electrical signals for a plurality of patients for which the corresponding electrode combination associated with the corresponding SVM has been therapeutically effective for delivering electrical stimulation; and The electrode combination for delivering electrical stimulation to the particular patient is selected based on the score.

12. A medical system comprising: An implantable medical device configured to deliver electrical stimulation via an active electrode combination; as well as A programmer device configured to: outputting information indicating an electrode combination of a plurality of combinations of implantable electrodes for delivering electrical stimulation to a particular patient with a movement disorder, the electrode combination selected based on a comparison between a sensed electrical signal for the particular patient and a plurality of sensed electrical signals for a plurality of other patients; as well as The implantable medical device is programmed to set the active electrode combination to a selected electrode combination so that the implantable medical device delivers electrical stimulation to the specific patient via the selected electrode combination to treat the movement disorder of the specific patient, wherein, in order to program the implantable medical device, the programmer device is configured to transmit programming information indicating the selected electrode combination to the implantable medical device.

13. The medical system of claim 12, wherein: The sensed electrical signal for the specific patient includes a sensed electrical signal representation for the specific patient, wherein the multiple sensed electrical signals for the multiple other patients include multiple sensed electrical signal representations for the multiple other patients, and wherein the electrode combination is selected based on a comparison between the sensed electrical signal representation for the specific patient and the multiple sensed electrical signal representations for the multiple other patients.

14. The medical system of claim 12, wherein: The implantable medical device is further configured to: sensing an electrical signal for the specific patient based on electrical signals measured across one or more electrode combinations in the plurality of electrode combinations; and The electrical signal for the specific patient is transmitted to the programmer device.

Images