Closed-loop neuromodulation device and control method

Abstract

An implantable device for interfacing with a nerve of a human or animal subject, the device comprising: a transducer comprising a sensing array having three or more electrodes arranged along an axis; an actuator configured to output a stimulus; a controller in electrical communication with the electrodes and the actuator, wherein the controller is configured to: receive a plurality of signals from the sensing array; process the signals using velocity selective recording (VSR) to determine a current electrophysiological representation of the subject, wherein VSR uses a plurality of measurement delays to provide a first plurality of velocity filtered signals representative of velocity data of a current signal travelling through the nerve; compare the current electrophysiological representation to a baseline electrophysiological representation, wherein the baseline electrophysiological representation comprises a second plurality of velocity filtered signals representative of velocity data of a baseline signal travelling through the nerve; based on the comparison, identify an electrophysiological biomarker; and control the actuator in the transducer to output said stimulus.

Description

[0001] CLOSED-LOOP NEUROMODULATION DEVICE AND CONTROL METHOD

[0002] FIELD OF THE INVENTION

[0003] The present invention relates to in the technical field of medical devices. More particularly, the present invention relates to the technical field of active implantable medical devices. Even more particularly, the present invention is in the technical field of active implantable medical devices for electrotherapy or neuromodulation.

[0004] BACKGROUND

[0005] The peripheral nervous system (PNS) is the body’s vast network of bioelectrical signalling that plays a fundamental role in regulating the function of cells and organs outside the brain and spinal cord. Increasingly, it is believed that by tapping into the signals that are translated via somatic and autonomic neural circuits, a plurality of conditions can be better understood and managed.

[0006] One condition of particular interest is epilepsy, which is the third most common neurological disorder in the world with over 50 million diagnosed. Moreover, about one third of these can be classified as refractory, or unresponsive to at least two or three types of anti-epileptic medication. This represents a large population unserved by pharmacological treatment and to which a surgical intervention would be considered to as the most appropriate treatment pathway.

[0007] Left untreated, epilepsy can have severe impacts on a patient’s health and quality of life, the average life expectancy for a refractory patient can be several years lower than the general population. Some patients will be suited to respective surgery, which is the removal of a locus or loci of seizure activity in the brain. However, this is a highly invasive and irreversible procedure. Other interventions include deep brain stimulation and RNS (responsive neurostimulation) but whilst reversible, these come with the risks of brain surgery (haemorrhage, infection) and also require an active implantable medical device to be left implanted in the patient. Several methods exist for neuromodulation of the nervous system, alongside pharmacology that affect the chemical transport of signals across the synapse, electrical stimulation is a well understood to elicit biopotential signals (action potentials) on the axon of the nerve cells. New methods exist including biological therapies and ultrasound stimulation that modulate neural signals in a variety of ways. Typically, conditions are treated in an ‘open loop’ manner, for example chronic pain that can be treated through a spinal cord stimulation with tonic waveform or drug pump for the slow release of pharmaceutical analgesics though an epidural drug pump. However, these open-loop treatments cannot provide any adaptive control such as for conditions that vary day-to-day or minute-to-minute.

[0008] Additionally, identification of seizures may be difficult, which discourages the use of any treatment methods that rely upon real-time measurements. Typically, two mechanisms are used for identifying seizure behaviour. The first is a magnet sensor that requires a user to sense the onset of their seizure (not all epilepsy patients have this sensation), and then swiping a wearable magnetic band over the implant in-time before the seizure. The other mechanism relies on detecting tachycardia (speeding up of heartrate) using an ECG signal - whilst this has been found to be present as a biomarker in 82% of patients in at least one seizure, there is significant variation seizure to seizure and even the opposite effect of bradycardia (slowing down of the heartrate) may be present making this an unreliable biomarker.

[0009] Other methods of detecting seizure activity without brain surgery include electroencephalography (EEG) or sub-scalp EEG, unfortunately both of these types of system require the use of externally worn components on the head that can be a large burden for the patient to use. These techniques identify ictal activity in the brain through the measurement of electrical fields outside the skull.

[0010] A further problem for long-term implants is that over time the electrical impedance increases between implanted electrodes and the neural tissue due to the buildup of fibrosis around the hard contacts. Moreover, there can be up to 30,000 individual axon fibres in a human nerve bundle, carrying signals from many different anatomies and systems so there can be a lot of background activity present, unrelated to the condition being monitored or treated. While clinical evidence indicates that closed-loop regimes often lead to better clinical outcomes in short-term studies, the signal-to-noise challenge in chronic implantable systems means that only a handful of systems have made it to the market.

[0011] Therefore, it is an object of the present invention to address the problems discussed above.

[0012] SUMMARY OF INVENTION

[0013] According to a first aspect of the present invention there is provided an implantable device for interfacing with a nerve of a human or animal subject, the device comprising: a transducer comprising a sensing array having three or more electrodes arranged along an axis; an actuator configured to apply a stimulus to the subject; a controller in electrical communication with the electrodes and the actuator, wherein the controller is configured to: receive a plurality of signals from the sensing array; process the signals using velocity selective recording (VSR) to determine a current electrophysiological representation of the subject; compare the current electrophysiological representation to a baseline electrophysiological representation; based on the comparison, identify an electrophysiological biomarker; and control the actuator to apply a stimulus to the subject based on the electrophysiological biomarker, wherein the current and the baseline electrophysiological representations each comprise velocity data of a signal travelling through the nerve.

[0014] The term “electrophysiological biomarker” preferably refers to an indication of a particular type of neural activity at a particular instant in time or during a given time period. The electrophysiological biomarker may be a seizure or pre-seizure biomarker, i.e. , an indication that a seizure (e.g., epileptic seizure) is occurring or is about to occur. The electrophysiological biomarker may identify ictal activity from the nerve. The electrophysiological biomarker may be a predetermined increase in the amplitude of nerve signals travelling between 5 and 15 m / s. Signals in this velocity range are commonly attributed to AG nerve fibres. These signals may be afferent or efferent signals. As used herein, the term “electrophysiological representation” refers to a representation of the neural activity of the subject at any given time or in any given period. In this way, a comparison of the current state with a baseline state can be used to indicate that a seizure is occurring or is imminent (i.e. , by identifying the electrophysiological biomarker). The term “current electrophysiological representation” may refer to a representation of the neural state of the subject in the present moment (e.g., as determined during ongoing measurements taken with the implantable device). The term “baseline electrophysiological representation” may refer to a representation of a known (healthy) neural state of the subject. This may be taken from clinical trial or pre-clinical data and loaded into the device at programming (e.g., a population baseline representation). Alternatively, this may be taken on implantation of the device, in a patient programming visit with their healthcare practitioner, or at any other time when the patient is in a known state (e.g., a personalised baseline). In this way, the current state can be compared to the baseline state to monitor for any abnormal neural activity; if abnormal activity is identified in the comparison, then this may be indicated with the electrophysiological biomarker. The controller may then adjust the stimulus from the actuator based on the identified electrophysiological biomarker.

[0015] Advantageously, by processing the signals using VSR, the electrophysiological representation may be provided in a manner that more easily and reliably enables the electrophysiological biomarker to be determined. In other words, variations in the electrophysiological representation determined by VSR processing are a better predictor of seizures than monitoring other measurements from a subject (such as their heartrate). For example, VSR processing is able to detect the speeds, magnitudes and directions at which signals propagate through the nerve, and therefore representations determined using VSR processing allow for changes these properties to provide the electrophysiological biomarker. Furthermore, VSR processing is able to determine an electrophysiological representation with a better signal to noise ratio, and can mitigate issues of stimulation artifacts typically present in eCAP recordings. Advantageously, this kind of electrophysiological representation can indicate the electrical activity in a particular type of nerve fibre (which typically have different propagation velocities); therefore, the electrophysiological biomarker can be sensitive to activity in individual types of fibre rather than a signal from the nerve bundle as a whole.

[0016] Therefore, the device is more effective at detecting the electrophysiological biomarker and therefore is able to provide stimulation more reliably when required. Likewise, the device is less likely to apply stimulation when it is not required, which may have negative side effects. The stimulation may be applied to prevent, delay or inhibit the seizure. Controlling the actuator to apply a stimulus may include controlling the actuator to emit a stimulus, controlling the actuator to change a stimulus, and / or controlling the actuator to stop a stimulus being applied. As used herein, the term “output a stimulus” may refer to applying a stimulus to the subject. The term "output a stimulus" may refer to any of the emission of a stimulus, change of a stimulus and / or stopping of a stimulus.

[0017] Additionally, since the device both receives the signals and also controls the stimulation applied to the subject, the device uses “closed loop” feedback. This is much more effective than typical “open loop” devices (i.e., which provide predetermined stimulation patterns over time or require external input to adjust the stimulation patterns), since the stimulation is specifically adapted to the subject and to their current neural activity. The stimulation applied by the at least one electrode may be based on the presence, loss amplitude (magnitude) or composition of the electrophysiological biomarker. For example, a larger magnitude biomarker (e.g., indicating a more severe seizure) may result in a more aggressive stimulation to the nerve.

[0018] The velocity data of a signal travelling through a nerve may include a plurality of velocity filtered signals corresponding to a plurality of measurement delays. In other words, VSR uses a plurality of measurement delays to provide a plurality of velocity filtered signals representative of velocity data of a signal travelling through the nerve. The VSR may be used to provide a first plurality of velocity filtered signals representative of velocity data of a current signal travelling through the nerve (which may be referred to as a current electrophysiological representation). The VSR may be used to provide a second plurality of signals representative of velocity data of a baseline signal travelling through the nerve (which may be referred to as a baseline electrophysiological representation). Preferably, the current and the baseline electrophysiological representations each comprise velocity data of a signal travelling through the nerve.

[0019] The velocity filtered signals are preferably phase aligned. As used herein, the term “velocity filtered signal” preferably refers to a signal that is processed using VSR to select a particular velocity. A particular velocity may be selected using VSR by using a particular measurement delay. Therefore, by using a plurality of velocity filtered signals to provide both the baseline and current electrophysiological representations, a greater amount of data can be used by the controller for a more nuanced comparison. This may allow for an electrophysiological biomarker to be more accurately and consistently identified.

[0020] The current and the baseline electrophysiological representations may each comprise a distribution obtained using the plurality of velocity filtered signals. The current electrophysiological representation may comprise a (e.g., first) distribution obtained using the first plurality of velocity filtered signals. The baseline electrophysiological representation may comprise a (e.g., second) distribution obtained using the second plurality of velocity filtered signals. As used herein, the term “distribution” preferably indicates a relationship linking two properties. Preferably, one of the properties (e.g., a first property) comprises the velocity of the velocity filtered signals. It will be appreciated that the velocity of the signals may be represented in other ways, such as using the measurement delays corresponding to each velocity. For example, the distribution may comprise a distribution of signal strength (e.g., amplitude) against the velocity of the signal. Alternatively or additionally, the distribution may include a distribution of signal phase against the velocity of the signal, or any other suitable property, such as frequency (e.g., obtained using a Fourier transform, such as a DFT). The distribution may be represented as a spectrum, plot or graph, though it will be appreciated that this is not strictly necessary for the controller to compare the current and baseline representations. Where the distribution is represented as a spectrum, spectral analysis may be performed, using techniques such as applying 2D filters, performing principal component analysis, and / or deconvolution of individual signals from the spectrum.

[0021] In this way, the comparison can be made by the controller using the distribution. Since the distribution represents a more nuanced relationship between measured properties (e.g., how the signal strength changes at different velocities), the controller may be able to more accurately and precisely identify an electrophysiological biomarker. Furthermore, the actuator may be able to provide a more accurate and appropriate stimulus using the distribution.

[0022] The comparison between the current and baseline electrophysiological representations may include identifying a difference (or change) in the distribution at one or more velocities, and identifying the electrophysiological biomarker based on this difference. For example, if a signal strength at a certain signal velocity (e.g., from 5 to 15 m / s) increases substantially relative to the baseline, this may be indicative of a particular neurological event, such as an (e.g., epileptic) seizure. Alternatively or additionally, the controller may identify the biomarker based on a change in peak amplitude, peak velocity or distribution shape.

[0023] Alternatively, the plurality of velocity filtered signals may be analysed without using respective distributions. For example, one velocity filtered signal may be compared with another.

[0024] The current and the baseline electrophysiological representations may each comprise both amplitude and velocity data of a signal travelling through a nerve, and preferably a distribution of the signal velocities travelling through the nerve.

[0025] By using the distribution of signal velocities as the electrophysiological representation (which may be referred to herein as a velocity distribution or velocity spectrum), the determination of the presence of the electrophysiological biomarker may be even more reliable since changes in multiple types of nerve fibre can be assessed at the same time. For example, by determining a distribution of the signal velocities through the nerve, it is possible to distinguish and quantify neural activity in different types of nerves, such as the myelinated (fast fibre), and unmyelinated (slow fibre) nerves. In particular, using the velocity distribution may allow the magnitudes, the speeds, and directions of signals through these nerves to be compared. It has been found that changes in the velocity distribution provide a particularly reliable indication of seizures.

[0026] The velocity distribution may be obtained using a sum of squares, a wavelet transform and / or a discrete Fourier transform (DFT). Preferably, the velocity distribution indicates the direction of the signals, thereby allowing the device to distinguish afferent and efferent signals.

[0027] The biomarker may be determined from the electrophysiological representation using convolution, filtering and / or template matching.

[0028] Preferably, the current electrophysiological representation is used by the controller to adjust one or more parameters of the stimulus applied to the subject.

[0029] In this way, the stimulation applied in response to the detected difference does not need to be prescribed but can be tailored to the subject and the measured signals. For example, a larger difference between current and baseline representation may result in a larger subsequent stimulation being applied to the nerve. The stimulation may be an electrical stimulation, ultrasound stimulation, or a pharmaceutical stimulation (e.g., a drug). Parameters of the stimulation may include the type, location, amplitude, frequency, pulse width, and / or phase of the stimulation signal. Adjustment of these parameters may be referred to herein as titration of the parameters. The stimulation may additionally be controlled by other sources such as external instructions or a preprogrammed sequence.

[0030] The plurality of signals may be received from the sensing array at a biomarker measurement interval.

[0031] The biomarker measurement interval may be predetermined or may be adjustable. Advantageously, receiving the signals at a measurement interval reduces power consumption, since the device does not need to record continuously. The measurement interval may be between 1 second and 10 minutes, preferably between 5 seconds and 5 minutes, more preferably between 10 seconds and 1 minute. The measurement interval may be about 30 seconds.

[0032] The controller may be further configured to control the actuator to emit a background stimulation to a nerve.

[0033] The background stimulation may be referred to herein as a “tonic stimulation”. In other words, the actuator may not only emit a signal in response to identification of a biomarker, but may stimulate the subject at other times. This may allow for ongoing therapy to be applied to the subject. The stimulation may be provided at a stimulation interval. The stimulation interval may be a regular interval (e.g., metronomic) or may vary over time. The stimulation interval may be predetermined or may be adjustable. The stimulation interval may be between 1 second and 6 hours, preferably between 1 minute and 1 hour, more preferably between 2 minutes and 30 minutes. The stimulation interval may be about 5 minutes.

[0034] Optionally, after controlling the actuator to emit the stimulus, the controller may be further configured receive a plurality of signals from the sensing array to measure nerve activation in response to the emitted stimulus.

[0035] Where the stimulus is an electrical signal, this may be referred to as recording the evoked compound action potential (eCAP). Alternatively, the sensing array may measure the nerve activation in response to other types of stimulus (e.g., ultrasound). In this way, if a biomarker is identified and stimulation is applied by emitting electrical signals, the device can monitor the subject to assess the effect of the stimulation, such as to check whether the stimulation is at (or within) a predetermined threshold or range.

[0036] If the at least one electrode is controlled to emit a tonic (background) stimulation, then the device can monitor the subject to assess the effect of the tonic stimulation, such as to check whether the stimulation is at (or within) a predetermined threshold or range.

[0037] In either case, if the stimulation is not within the predetermined threshold or range, the stimulation applied by the actuator may be adjusted. In this way, closed loop feedback is provided to repeatedly check and update the stimulus so that the device operates more optimally.

[0038] Preferably, the actuator is provided by at least one electrode in the transducer, and the at least one electrode is controlled to emit an electrical signal. Electrical signals are a particularly effective form of stimulus, and can be applied quickly and precisely.

[0039] Preferably, the at least one electrode is provided in a centre of the sensing array, and received signals from the remaining electrodes either side of the at least one electrode are processed by the controller to determine an electrophysiological representation for both afferent and efferent signals.

[0040] In other words, since the stimulation signal is emitted from the centre of the sensing array, the sensing array may be used to measure an electrophysiological representation for signals moving both directions through the nerve away from the at least one electrode. Preferably, the at least one electrode in the centre of the sensing array is a single electrode with an (approximately) equal number of electrodes on either side along the longitudinal axis. In this way, the recruitment of efferent and afferent nerve fibres can be assessed individually in response to a predetermined signal from the at least one electrode. The set of signals from the remaining electrodes on either side of the at least one electrode may be processed separately to each other using VSR. The controller may be configured to maximise a stimulation effect travelling towards a target organ (such as the brain) and minimise the stimulation effect travelling away from the target organ. Preferably, for the treatment of conditions such as epilepsy, the controller is further configured to adjust the emitted electrical signal to maximise afferent stimulation and / or minimise efferent stimulation. Advantageously, this may increase the seizure modulation efficacy whilst also reducing off-target stimulation side effects. Alternatively, the controller may adjust the emitted electrical signal to minimise afferent stimulation and / or maximise efferent stimulation.

[0041] Alternatively, or in addition to the at least one electrode, the actuator may be provided by at least one of: a light source, a drug pump and / or an ultrasound transducer. In this way, therapeutic treatment can be provided to the subject by the device, such as by releasing a pharmacological or bio-pharmacological substance into the subject, and / or by using ultrasound or light to stimulate the subject’s tissue.

[0042] The controller may be further configured to update the baseline electrophysiological representation based on the current electrophysiological representation.

[0043] For example, if normal neural activity of the subject changes over time, the implantable device may update accordingly so that the electrophysiological biomarker can still be obtained reliably. Furthermore, over time the impedance between the electrodes and the nerve may change; particularly where the device is implanted for a long period of time, updating the baseline electrophysiological representation allows the device to remain more effective for longer.

[0044] The implantable device may comprise a data storage unit configured to record data corresponding to identification of the electrophysiological biomarker.

[0045] The data may include the magnitude of the electrophysiological biomarker and / or data from the electrodes. The data may include metadata such as the date and time when each electrophysiological biomarker is identified, and / or other information such as location or other health properties of the patient (e.g., from other sensors such as heartrate sensors). For example, where the biomarker indicates a seizure, the data storage unit records the times at which the seizure occurs, which may allow their frequency and magnitude to be tracked over time. Advantageously, this may allow the efficacy of different treatments and / or stimulation regimes to be tested. The data storage unit may comprise a non- volatile memory. The biomarker measurement interval and / or the stimulation interval may be adapted based on data stored in the data storage unit.

[0046] The implantable device may comprise a wireless interface configured to transmit data to and / or from the device.

[0047] The wireless interface may comprise a communications antenna, which may transmit data via any suitable means, such as electromagnetic (e.g., radio), acoustic (e.g., ultrasound) or photonic. Advantageously, this allows the device to be controlled, and data (such as the data recorded in the data storage unit) to be transmitted from the device, without the need for mechanical connections or removal of the device from the subject. In this way, data from the device can be reviewed by the subject and / or a healthcare practitioner and monitored over time. The wireless interface may connect to an external device such as a mobile phone, tablet, or a custom product. The external device may connect to other healthcare records, such as on a local data store or on the cloud.

[0048] The implantable device may comprise a rechargeable battery and an interface for connecting a battery-charging device.

[0049] The interface may comprise a physical port to facilitate wired charging of a battery of the device. Preferably, the charging interface comprises an inductive charging coil to allow the device to be charged without the need for mechanical connections. Optionally, the inductive charging coil may also provide an antenna of the wireless interface. Alternatively or additionally, the device may have a non- rechargeable battery.

[0050] The transducer may be provided by a cuff configured to substantially surround the nerve, during use. For example, the cuff may be “o-shaped” to entirely surround the nerve, or “c-shaped” or “u-shaped” to partially surround the nerve. In this way, the electrodes can be located on the cuff to surround the nerve (at least partially) thereby allowing them to interface with the nerve more effectively, which produces data with an improved signal-to-noise ratio. Alternatively, the transducer may be aligned adjacent to the nerve (i.e. , without surrounding it partially or substantially). The transducer may comprise an electrode matrix having one or more additional electrodes arranged off the axis of the three or more electrodes. In other words, the sensing array is not a one-dimensional array of electrodes but has further electrodes adjacent to (or offset from) those along the axis. In this way, any of the electrodes in the matrix may be selected in order to account for misalignment of the axis of the transducer with an axis of the subject’s nerve. This reduces the surgical precision required to effectively implant the device. The matrix may be a two-dimensional matrix provided on a flat transducer. Alternatively, the matrix may be provided on a cuff that at least partially surrounds the nerve; in this way, different electrodes in the matrix may be used to selectively record signals in particular parts of the nerve bundle (e.g., on different sides of the nerve bundle).

[0051] The controller may be housed in an electronics package surrounded by a protective housing. Advantageously, the protective housing prevents damage to the electrical components (e.g. , the controller) and / or to the subject. For example, the protective housing may comprise a titanium, ceramic, or steel package. The protective housing may be hermetically sealed.

[0052] The transducer may be connected to the electronics package by a flexible lead. Advantageously, the electronics package may be implanted in a different part of the subject than the transducer, such as a part of the subject where there is more place and / or where it is easier to implant the device. This offers more flexibility for which nerves may be stimulated, since the bulkier electronics can be located away from the nerve. Furthermore, this allows the electronics package itself to be larger such as to include a larger battery. The flexible lead may be detachable from the electronics package.

[0053] Alternatively, the transducer may be provided as part of the electronics package. In other words, the implantable device is a single monolithic package containing both the processor and the transducer. In this implementation, the implantable device is implanted adjacent to the nerve of the subject. Advantageously, this decreases the number of surgical steps required in order to implant the device. The transducer may be configured to provide the actuator, and preferably the actuator is provided by at least one electrode in the sensing array. Alternatively, the actuator may be a separate component to the transducer, and may apply the stimulation elsewhere in the subject (such as in a separate nerve). The stimulation to the nerve may be applied using different electrode(s) than those in the sensing array, such as electrodes in a separate stimulating array. By using the same electrode(s) for both measurement and stimulation, the transducer can be made more compact. However, the electrical connections may need to be more complicated in order to allow for both measurement and stimulation using common electrodes. The device may comprise a multiplex component so that the controller can receive signals from all of the electrodes in the sensing array substantially simultaneously. Where a plurality of electrodes are used to stimulate the nerve (i.e. , in a stimulating array), the device may include a demultiplex component to allow all of the electrodes in the stimulating array to be operated substantially simultaneously.

[0054] According to another aspect of the present invention, there is provided a method of controlling the implantable device as described above and herein, the method comprising: receiving a plurality of signals from the sensing array; processing the signals using velocity selective recording (VSR) to determine a current electrophysiological representation of the subject; comparing the current electrophysiological representation to a baseline electrophysiological representation; based on the comparison, identifying an electrophysiological biomarker; and determining a stimulus to be applied by the actuator based on the electrophysiological biomarker, wherein the electrophysiological representation comprises an amplitude and velocity of at least one signal travelling through the nerve.

[0055] The features discussed below may apply to the implantable device described above and herein, and / or the method described above and herein.

[0056] Preferably, processing the signals using VSR comprises: delaying each signal from each electrode or each pair of (e.g., adjacent) electrodes by a time period that is dependent upon: the distance of the electrode along the longitudinal axis of the transducer, and a target signal measurement velocity; summing the delayed signals together to produce a phase aligned signal for the target signal measurement velocity. The steps described above may be referred to as apply a “delay-and-add” algorithm. In other words, a delay of variable length is applied to each signal corresponding to the target signal measurement velocity and the resulting signals are summed together (e.g., for each timestep).

[0057] Preferably, the target measurement velocity is varied to generate phase aligned signals for a plurality of target signal measurement velocities, which can be processed to produce a velocity spectrum. In otherwords, phased aligned signals are produced over a range of target measurement velocities whereby to provide a velocity spectrum. As discussed above, monitoring for changes in the velocity spectrum may be particularly reliable for identifying electrophysiological biomarkers. Preferably, the signals are digitised (e.g., with an ADC) by the controller. Preferably, the signals from the electrodes are amplified before being digitised. The amplification may be achieved using an operational amplifier and / or a transistor. By amplifying the signal, it is possible to obtain a high input impedance to enable measurement of signals in the 1 uV to 10 mV range. Preferably, the signals are filtered before being digitised. The filter may be a bandpass filter. Preferably, the VSR processing is performed after the signals from the electrodes are digitised. Alternatively, at least part of the VSR processing may be performed on the analog signals, such as applying the signal delay.

[0058] The VSR processing may be based on bipolar data obtained from pairs of electrodes in the sensing array. Advantageously, bipolar measurements reduce the impact of signals that are common to all the electrodes (known as “common mode” signals); this removes fluctuations from the signals which do not contribute towards biomarker identification, and also allows the voltage measurement to be taken over a smaller range. Alternatively, the VSR processing may be based on monopolar data recorded from each electrode individually. The monopolar data may be measured relative to a reference potential elsewhere in the subject. Alternatively, tripolar recording, or other recording protocols may be used. The sensing array preferably includes at least four electrodes arranged along the axis. The sensing array preferably includes less than 30 electrodes arranged along the axis. The spacing between the electrodes may be between 1 and 30 mm. The controller may be configured to receive signals from the electrodes at a sample rate of at least 2kHz. The sampling frequency is preferably at least twice the target signal measurement velocity divided by the inter-electrode spacing. The controller may be configured to receive signals from the electrodes having peakpeak values of between 1 uV and 100mV.

[0059] The identified electrophysiological biomarker may correspond to the onset or occurrence of a (e.g., predetermined) neurological event. The term “neurological event” preferably refers to any abnormal neural state of the central or peripheral nervous system, when assessed relative to normal (e.g., baseline) neurological function.

[0060] The neurological event may be an epileptic seizure. The method may include operating the actuator to apply the stimulus to the subject. The electrophysiological biomarker may be identified based on an increase in nerve signals travelling from 5 to 15 m / s, more preferably from 7 to 10 m / s.

[0061] According to a further aspect of the present invention there is provided a method of applying therapy to a nerve of a subject, comprising: implanting a device into the subject, the device comprising a transducer and a controller; aligning a longitudinal axis of the transducer with the nerve, the transducer comprising a sensing array having three or more electrodes arranged parallel to the longitudinal axis; receiving, at the controller, a plurality of signals from the sensing array; processing the signals using velocity selective recording (VSR) to determine a current electrophysiological state of the subject; comparing the current electrophysiological state to a baseline electrophysiological state; based on the comparison, identifying an electrophysiological biomarker; and controlling at least one electrode in the transducer to stimulate the nerve.

[0062] The nerve is preferably the vagus nerve (or vagus nerve bundle). In this way, the device may control therapies of the autonomic nervous system, such as cardiac, respiratory, rheumatologic, gastro-intestinal, endocrinologic, renal, inflammatory, or neurological conditions in a closed loop manner with feedback from the transducer. More preferably, the device is used for monitoring and treatment of seizures via vagus nerve stimulation for patients with epilepsy.

[0063] It will be understood by a skilled person that any device or apparatus feature described herein may be provided as a method feature, and vice versa. It will also be understood that particular combinations of the various features described and defined in any aspects described herein can be implemented and / or supplied and / or used independently.

[0064] Moreover, it will be understood that the present invention is described herein purely by way of example, and modifications of detail can be made within the scope of the invention. Furthermore, as used herein, and “means plus function” features may be expressed alternatively in terms of their corresponding structure.

[0065] BRIEF DESCRIPTION OF DRAWINGS

[0066] One or more embodiments will now be described, purely by way of example, with reference to the accompanying figures, in which:

[0067] Figure 1 shows a pictorial description of the system invention for closed-loop sensing and therapy to the peripheral nervous system utilising a multi-electrode array, depicted here on the cervical region of the left vagus nerve;

[0068] Figure 2a shows a flow diagram that depicts the method, utilising the system of Fig 1 , of recording biopotentials and controlling therapeutic stimulation using multielectrode recording and variable time delay;

[0069] Figure 2b extends the flow diagram that depicts the method in Figure 2a showing how multiple filtered biopotential signals of the same electrode array in Figure 1 can be combined with variable time delay to produce a therapeutic output by comparison; Figures 3a to 3c shows three flow diagrams showing methods of performing closed-loop therapy, including automatic titration of stimulation and / or the logging and disclosure to patients of biomarker events;

[0070] Figure 4a shows a series recording plot and electrode nerve diagram depicting the output of raw recordings of an evoked compound action potential ex-vivo;

[0071] Figure 4b shows a series of plots depicting various stages in the VSR signal processing method to identify a signal decomposition against nerve fibre conduction velocity;

[0072] Figure 5 shows an electrical schematic for the realisation of electronics for a real time, closed loop, neuromodulation implant utilising the VSR method;

[0073] Figure 6 shows an electrical schematic depicting an implementation of the core biopotential recording and therapeutic actuation;

[0074] Figure 7 shows an alternative monopolar implementation of the biopotential recording electronics;

[0075] Figure 8a shows a mechanical assembly depicting an implementation of the closed loop system;

[0076] Figure 8b shows an exploded view of the mechanical assembly in Figure 8a;

[0077] Figure 9 shows an alternative implementation of the system mechanics where the electrodes and electronics are housed within a single package;

[0078] Figures 10a to 10e show a series of diagrams that depict a large animal experiment that identified a specific biomarker on the vagus nerve during seizure using VSR that can be used to trigger therapeutic stimulation;

[0079] Figure 11 shows a process flow diagram depicting the use of the invention for vagus nerve epilepsy therapy; Figure 12 shows a process flow diagram depicting the use of nested control loops for tonic and on-demand stimulation for closed loop condition management;

[0080] Figure 13 shows a diagram depicting how the device could differentiate between afferent and efferent evoked signals allowing for titration of electrical stimulation for on-target effect while reducing unwanted side effects; and

[0081] Figure 14 shows further examples of electrode arrays that may be used as part of a transducer.

[0082] DETAILED DESCRIPTION

[0083] The closed-loop neuromodulation therapy system described herein represents an improvement over the prior art of incorporating the velocity selective recording (VSR) technique, only known in academic literature with complex lab equipment, in a form suitable for chronic implantable therapy devices. This invention comprises of both a physical system (device) and digital control method required for controlling therapy using physiological biopotentials recorded from nerves in the peripheral nervous system.

[0084] Figure 1 shows a device 100 implanted into a subject 1. The device 100 includes an electronics package 105 having control electronics 106 (a “controller”). The device 100 also includes a transducer 103 having an array of electrodes 102 in relative electrical proximity to the target nerve or nerve bundle 101 of the subject 1 (for example the vagus nerve). The electrodes 102 are in relative electrical isolation to each other. These electrodes 102 should be arranged parallel to the primary axis of nerve conduction. The transducer 103 comprises at least 3 electrodes 102, which allows the control electronics 106 to measure biopotential signals with spatial separation to capture action potentials travelling afferent or efferent of the central nervous system. The transducer 103 includes a mechanical means of affixing the electrodes 102 in electrical proximity, which may also provide electrical isolation between the electrodes 102 and other tissues, such as a silicone sheath surrounding electrodes 102 in a cuff that surrounds all or part of the nerve 101. In one embodiment a flexible lead component 104 would bring electrical signals down insulated conductors to the relatively inflexible electronics package 105 that protects the therapy control electronics 106 from moisture ingress harmful to the electronic integrated circuits and conductors and egress of chemicals harmful to the body. In one embodiment of this this may be a thinwalled titanium ‘can’ that is laser welded shut to provide a hermetic barrier to moisture ingress. The therapy control electronics 106 will contain means to measure electrical activity from more the plurality of electrodes, a method for implementation of the control loop and connection to a therapeutic actuator to perform a therapy titrated or triggered by the electrical recording. A wireless interface 107 that may be electrical, electromagnetic, acoustic or photonic, is required to communicate with an external device 108 for controlling and or titrating therapy. This external device 108 may be a custom product or a common piece of consumer electronics such as a mobile phone or a tablet.

[0085] The exact implementation of the closed loop therapy control algorithm may vary based upon the design constraints of a particular indication and may include digital and analogue signal processing methods. At its core however, there are several elements that must be present to perform velocity selective recording and unlock the benefits of nerve conduction speed and direction information to decode the bodies neural signals. Figure 2a shows one example of how the electronics and method may be implemented. As shown, the device may include up to n electrodes, which each produce a terminal voltage 201 . For each electrode, an amplifier 202 such as an operational amplifier or transistor, is used to increase the amplitude of the electrode terminal voltage 201 and provide a high input impedance to enable to measurement of signals in the 1 uV-10mV range. Subsequently, the analog signal may be digitised with an analog-to-digital converter (ADC - not shown). The device also includes a filter such as a band pass filter 203. The passband may have a low pass value (measured at 3dB attenuation) of between 1 and 500Hz. The passband may have a minimum high pass value of 1 kHz and a maximum at the Nyquist frequency of the analog-to- digital converter. Before or after digitization a phase delay 204 is added to the signal from each spatial recording channel (these channels may be unipolar, bipolar or tripolar in configuration) where the delay factor, u, is a function of the inter-electrode spacing and electrode number. These signals are summed together to form a velocity filtered output signal h(t) 205.

[0086] Figure 2b shows a further improvement to the method discussed above in relation to Figure 2a. In Figure 2b, the same set of analog electrode records are processed using different time delays, In processing step 206, a time delay of ui is used to produce filtered signal hi(t). In processing step 207, a time delay of u2is used to produce filtered signal h2(t). In processing step 208, a time delay of uxis used to produce filtered signal hx(t). These signals can be analysed by the controller 209 to produce the real-time therapeutic output y(t). The controller 209 may utilise methods of comparison of the distribution of the filtered signals against a baseline electrophysiological representation, comparison of one velocity filtered signal against another, ratiometric control according to prominence of features of the distribution of velocity filtered signals such as peak amplitude, peak velocity or distribution shape. As the distribution has a time and velocity variant components extending the analog recorded signals into a second dimension the velocity distribution in time can be analysed as a spectrum with features of power and phase overtime. Spectral analysis techniques can also be utilised by the controller such as applying 2D filters, performing principal component analysis and deconvolution of individual signals from the spectrum. How this signal is used will be described further in relation to Figures 3a-3c.

[0087] The approach above differs from prior art in that a usually one pair of electrodes is be selected to record from, and the resultant recording is a time variant signal at a single point, allowing frequency domain processing but not allowing information about the propagation speed or direction of the signals to be discerned. As peripheral nerve targets usually contain a mixture of afferent and efferent fibres, recording information using a single dimensional recording confounds signals from the point of interest. Fibre types corresponding to different signals are known to have characteristic velocities based upon the axon diameter and amount of myelination. Picking up differences between slower and faster moving fibres can help identify which neural networks are active and discern, for example, between motor action and proprioception. Identifying the directionality and velocity of a neural signal enables the physiological source of the signal to be differentiated from other neural activity that is less correlated with identification of the neural biomarker of interest.

[0088] The application of this method and device can enable ‘closed loop’ therapies to treat many types of conditions.

[0089] A first example of a method is shown in Figure 3A. The process starts at 301. Subsequently, at 302 biopotential signals are measured using VSR (e.g., measure a biopotential signal distribution with VSR), as described in relation to Figures 2a and 2b. At 303, the method checks where a biomarker is identified (which will be described in more detail later). If not, then the method returns to 302. If a biomarker is identified, then the method proceeds to 307, where nervous system stimulation is applied and / or adapted.

[0090] A variation on this method is shown in Figure 3B. Steps 301 to 303 are the same as Figure 3A. At 304, the nervous system is stimulated. At 305, the nerve activation (e.g., resulting from the stimulation in 304) is measured. At 306, the method checks whether the stimulation applied in 304 is at a predetermined threshold based on the measurement of the nerve activation. If yes, then the method returns to step 302. If not, the stimulation parameters (e.g., a waveform) is adapted at 307, and then the method returns to 304. In this way, the device 100 provides automatic titration of the output stimulus by measuring the nerve activation after stimulation 304 via a measurement of evoked compound action potential (eCAP) or similar. Utilising the VSR technique can mitigate issues of stimulation artifacts typically present in eCAP recordings as these occur on all electrodes almost instantaneously, the VSR technique nulls common mode noise. This can be seen on the recording 405 in Figure 4a as the sharp dip 406 just after t=0, this recording was taken using a multi-electrode cuff with 9 elements 403 arranged along the vagus nerve bundle 401 surrounded by an in insulative silicone sheath 404. The propagation delay caused by a travelling eCAP 402 can be seen on the graph 409, but the monopolar recording signals show a complex convolution of faster and slower fibre potentials. In Figure 4a, the compound action potential fast fibres are shown at 407, and the compound action potential slower fibres are shown at 408.

[0091] Another variation on the method is shown in Figure 3C. Steps 301 to 304 are the same as Figure 3B. At step 308, due to the identification of the biomarker in step 303, the presence of the biomarker is logged in a device memory. At step 309, the device 100 connects to the external device 108. At step 310, the data from the device memory is synchronized with a patient healthcare record stored in the external device 108.

[0092] In another example, the device 100 may simply record the biopotentials at step 302 and record changes over time to allow a patient or healthcare practitioner to adapt the treatment based on the patient’s electronic log data. In other words, this method would proceed according to the method in Figure 3C but the signals measured in 302 would be stored in the device memory in 308. This would be accessible when the device 100 connects to an external electronic device 108, this may connect in-turn to other forms of electronic healthcare records, either on a local data store or on the cloud. On-demand stimulation of the nervous system may be included by controlling the neuromodulation output in response to the identified biomarker as a trigger for stimulation, or amplitude I frequency or pulse width modulation, instead or in addition to other sources of stimulation control such as patient or clinician remote devices, or a time based ‘tonic’ stimulation (see figure 12).

[0093] By using the velocity selective recording method (Figures 2a and 2b) it can be seen in Figure 4b that by applying the velocity transform for various propagation delays 412 to the biopolar signals 410, several velocity transforms 411 stand out against the noise that correspond to the physiological conduction velocities and propagation direction. Several methods can be used to identify this nerve activity against the background: a discrete Fourier transform method 413 with a 1.5kHz centre frequency, 800Hz bin width, is presented against an absolute peak max square method 414 with magnitude of signals against the signal velocity. It can be seen in both cases there is a clearly defined peak of activity between 10-80m / s in the positive direction indicating the recruitment of myelinated A and B fibres from the stimulus. This is expected as these groups form the majority of fibres in the vagus nerve, myelinated fibres are more actively recruited by electrical stimulation and the signal amplitude from the smaller diameter, slower C fibres is less than that of A and B fibres. Other methods to identify or characterise neural activity in a velocity filtered signal include: the sum of squares, mean of squares, wavelet transformations, peak finding, deconvolution, other Fourier transforms such as FFTs and other frequency domain analyses.

[0094] The implementation of the electrical control system is critical to the ability of an implantable device to function for a long period of time in the human body. One example of the electronics package 500 will now be described with reference to Figure 5. The electronics package 500 includes microelectronics components 506-517 that will be described later in more detail. The electronics package 500 also includes a battery 518 to supply power to the microelectronics components 506-517. The battery 518 may be rechargeable. The battery 518 may be a primary cell. The electronics package 500 comprises contacts 505 to facilitate connection with the flexible lead 104. The electronics package 500 comprises a communication antenna 504 to facilitate connection with the external device 108. The electronics package 500 includes an inductive charging coil 503 to allow for charging of the battery 518.

[0095] Typically, the electronics are housed within a hermetic package 501 to protect the electronics from moisture ingress and the body from egress of toxic substances. The hermetic package 501 may be constructed from metals such as stainless steel or titanium, or ceramics such as zirconia and silicon oxides. Alternatively, the electronics package 500 may not have the hermetic package 501 but with barriers to liquid transport such as parylene coatings, silicon nitride, epoxy castings or silicone over mouldings 502, a mixture of these packaging types are often present to house different components. In Figure 5, the connections to the lead contacts 505, communications antenna 504 and inductive charging coil 503 are housed in a non-hermetic package 502, which may improve electromagnetic performance or connections to the electrodes. Whereas the microelectronics components 506-517 and battery 518 are housed within hermetic package 501. The microelectronics components 506-517 will now be described in more detail. In one implementation, the main control loop may run in a core, programmable, processor 510 that communicates to an analog-front-end 508 that provides the interface to the electrodes. The processor 510 and analog-front-end 508 may be discrete components or an Application Specific Integrated Circuit (ASIC). The processor 510 may be referred to herein as a microprocessor or controller. The processor 510 may be a field programmable gate array (FPGA). The analog-front- end 508 may be connected to the case, which can be configured as a system ground, anode or cathode. It is preferable to connect the recording and stimulation electronics to the electrodes in electrical contact with the neural tissue via DC blocking capacitors 506, which filter out DC offsets for the signal processing chain and also provide protection from charge buildup on stimulation electrodes causing patient harm. The microelectronics components include a timekeeping module 509 (e.g., crystal oscillator or similar) to control recording and stimulation intervals, a non-volatile memory 512 for storage of program memory and physiological data that is retained if the battery 518 depletes below the operating level. To control the battery charging and discharge, a battery management system 517 is used. The battery management system 517 may be connected to a system power management module 515, which may be connected to DC system voltage supply rails 516. The battery management system 517 may be connected to the inductive coil 503, optionally via power electronics such as a rectifier, tank and overvoltage protection 514. The battery 518 may be connected to a ground reference, which may be a floating ground reference.

[0096] Communication to the external device 108 is achieved by the communication antenna 504 which may receive and communicate program or condition information over a radio-frequency, inductive coupling, optical or acoustic link. The signals may be provided between the antenna 504 and the processor 510 with a communications module 511. Other sensors 513 may be present such as a temperature sensor, accelerometer, or any other suitable type of sensor. While the above description refers to several components, it will be appreciated that the functions may instead be provided by a single processor or controller. The analog-front-end 508 may be configured in various ways to enable the recording of the multiplicity of electrodes. An example is included in Figure 6, where an array of electrodes 602 are arranged along a nerve conduction direction 601. The array of electrodes 602 are connected to pre-amplifiers 603 that may implement a bandpass filter reducing high frequency components above a 10kHz neural firing bandwidth and velocity selective filter aliasing. This amplified signal is then fed into an array of difference amplifiers 604 such as instrumentation amplifiers or operational amplifiers, to obtain a bipolar signal of electrode(n) - electrode(n-l) signal (where n is the number of electrodes). These n-1 signals may be multiplexed with a multiplexer 605 to reduce the number of analog to digital convertors 606, the signal acquisition rate for an ADC multiplexed in this way would need to be n times greater than one that can obtain digital values for the different bipolar signal channels simultaneously. For example, a device with 5 electrodes and a 3 mm inter-electrode spacing would require a >16.7kHz simultaneous sample rate or a >66.7kHz sequential sampling rate to get the full VSR filtering specificity for signals of speeds up to 50 m / s.

[0097] While delays can be added in analog electronics before digitisation, it is most useful to control these in the digital domain with the processor 607, delays may be applied in discrete time steps by shifting samples in memory, triggering sequential acquisitions, or by amounts less than a single timestep via phase shifting, interpolation or frequency domain manipulation. The processor 607 will also contain the control loop software to adapt or trigger the stimulation, examples of therapeutic stimulation could be the control of drug from a pump 610 or acoustic energy from an ultrasound transducer 611 or light from a light source. A particular implementation would utilise the same (or a subset of the) electrodes 602 as used for recording to apply electrical stimulation. This requires a digital to analog converter (DAC) 608 that may turn the digital stimulation command into a current or voltage controlled output. The external stimulation may use multiple channels for different waveforms on different electrodes 602, or a multiplexer circuit 609 to switch between electrodes 602 or steer current between electrodes 602. Figure 7 depicts an alternative implementation of the recording hardware. As with the example above, the electrodes 702 are arranged along a nerve conduction direction 701. However, in this example, the signals are recorded using high precision analog to digital converters 703 with a reference electrode 705 situated away from the neural bioelectrical activity. Conversion to biopolar channels can be undertaken in the processor 704 as part of the signal processing chain. In practice, input impedance to the ADCs 703 and small amplitude of the bioelectronic signals promote the use of voltage buffers and amplifiers between the electrodes and converters, though for simplicity these are not shown in Figure 7.

[0098] Figures 8a and 8b show an implementation of the device 800. As discussed previously in relation to Figure 5, the electronics are housed within an electronics package 804. A transducer 820 is connected to the electronics package 804 by a flexible, detachable lead 803. The transducer 820 includes an electrode cuff array 801 . The electrode array 801 is housed within an insulating flexible polymer cuff 802 that wraps circumferentially around the target nerve maintaining mechanical stability and electrical continuity between the nerve activity and the electrodes. In this implementation the connectors for the lead-electrode contacts 807 (that provide mechanical security and protection from liquid ingress), and a wound metallic coil 808 (for inductive communications and battery charging), are housed within a cast or moulded ‘header’ 806 made from a non-conductive material. The device 800 has a feature 809 in this header 806 for a surgeon to affix to a patient anatomy to mitigate migration.

[0099] As mentioned above in relation to Figure 5, the electronic components are housed within a hermetic electronics enclosure 805. To pass electrical signals into the hermetic electronics enclosure 805 an array of brazed ceramic and metal feed throughs are used that connect to the lead contacts 807 and printed circuit board (PCB) 811. This PCB 811 contains all the passive components and integrated circuits 814 to read bioelectronic signals and control an electrical stimulation output via a programmed control loop. Off-board a rechargeable battery 812 is held within a moulded carrier 813 and also connected to the PCB 811 . An alternative device 900 is depicted in Figure 9. In this mechanical configuration the electrodes 902 and control electronics 904 are included within the same package with a small, non-rechargeable battery 906, and an RF antenna 907 is used for communication with an external device. In order for the electrodes 902 to be arranged adjacent to a target nerve site 901 , the device 900 is implanted at nearby location 903 in the subject. This type of monolithic package may be preferable as it minimizes the invasiveness of the surgery required to implant the device 900. However it offers limited room for battery 906 and therefore may not be suitable for long-term use. A charging circuit could be included to extend the device lifetime via wireless charging. A challenge with this approach is that the requirements for space at the target nerve site 901 may not allow a single package to be mechanically affixed in place there. In this implementation the package is not hermetically sealed, but a secondary barrier to moisture ingress 905 is included for biocompatibility and electronics lifetime.

[0100] This technology can be applied to the treatment of various conditions, but one example of such an application is the treatment of epilepsy via closed loop vagus nerve stimulation. Figures 10a and 10b depict an experimental setup used to test the device 1004 and method described previously. An acute epilepsy model was used with an anesthetised pig 1001 , which had seizures induced and measured via the traditional method of electrocorticographic electrodes 1005 on the surface of the brain 1002. Nine electrode cuffs were implanted on the left and right cervical vagus nerves 1003 and bioelectrical activity was recorded using the electrodes 1005 at baseline and seizure events. The results are shown in Figure 10c. The method described previously was then used to obtain average velocity spectra at baseline and seizure events, and the results are shown in Figures 10d and 10e. It can be seen that a prominent peak at 7-15 m / s appears during seizure activity. This peak can be attributed to Ay (A gamma) or AG (A delta) nerve fibres.

[0101] Comparison of the primary biomarker to other features of the velocity time space can be used to confirm the validity of the neural recording in-vivo. For example, the 30-50 m / s afferent peaks are seen in Figure 10e during both baseline activity and seizure events. This is known to correspond to respiratory activity which is commonly associated with firing of Ap (A beta) and A6 (A delta) afferent nerve fibres; presence of these peaks provides additional confidence in the validity of the recorded data.

[0102] This novel seizure biomarker could be utilized a trigger for electrical stimulation to disrupt the seizure pathway, and a method for use in epilepsy is depicted in Figure 11. At step 1101 , a measurement interval may be programmed. At step 1102, the electrode array measures vagus nerve biopotentials . A measurement interval is preferable over another embodiment that uses continuous recording, since reduce overall device power consumption is reduced. At step 1103 the device checks for the presence of a seizure biomarker (using the methods described above). If no biomarker is detected, then the device continues to measure the vagus nerve biopotentials at set measurement intervals (step 1102). If the device detects a biomarker for seizure over the patient baseline (which may be statically programmed or rely on a moving average approach or similar) then therapeutic stimulation is applied to the patient nervous system at step 1104. This may be the release of an anti-convulsant medication, electrical or acoustic stimulation of the target nerve. At step 1105, the device logs this activity to non-volatile memory. In this way, at step 1106, when a connection to an external device is established the data could be displayed or synchronised with the patient healthcare record to allow the patient, healthcare practitioner or caregiver to monitor the condition. Nonvolatile memory is advantageous for a rechargeable device as it allows the device / seizure log to persist during a discharge state and be interrogated after the device is re-energised. In a primary cell battery device volatile memory could be used instead as the memory module would have power between device connections, however this embodiment suffers because the relative cost and power consumption of volatile memory is higher than non-volatile.

[0103] Figure 12 depicts a more fully featured embodiment of a closed loop nerve stimulation method. At step 1201 , therapy settings are initially programmed via the external device wireless interface. At step 1202 the device enters an adaptive therapy control loop, which has two main activity intervals. The first is a ‘tonic’ stimulation interval, where an applied stimulus is given at a set interval, for example ten, 5mA amplitude, 10uS bipolar pulses every 5 minutes. This may be delivered in an open-loop fashion but in this embodiment the device would measure the nerve activation after stimulus by recording the evoked compound action potential (eCAP) on electrodes not used for the stimulation. Based on this measurement, at step 1203, the device may adapt the stimulation settings in the program memory to titrate the stimulus to the correct threshold. This allows the stimulation to be adjusted based on the subject’s response, which may allow for a more optimal level of stimulation to be applied.

[0104] The second activity interval is a biomarker measurement interval, shown at step 1204, where naturally occurring biopotentials are measured at the biomarker measurement interval. At step 1204, a stimulus is applied in response to the presence, loss, amplitude or composition of this biomarker. Similar to the tonic stimulation control loop, the biomarker-controlled stimulus can be adjusted utilising eCAP recording to titrate the stimulus to correct thresholds and control for on-target / vs-off target affects or changes in the nerve response at the time. At step 1206, biomarker events and eCAP response data could be logged in the device memory and this used to identify long-term macro trends and adapt the stimulation intervals or other control loop settings.

[0105] One specific implementation of the eCAP stimulation titration is depicted in Figure 13. A multi-electrode array comprising a plurality of electrodes 1303 is implanted in electrical proximity to the target nerve bundle 1301. As described previously VSR can be utilised to differentiate between evoked activity; afferent activity 1302 towards the central nervous system (CNS), and efferent activity 1305, from the CNS towards the periphery. If an electrical stimulus is applied at a central location 1306 and recordings are taken using multiple electrodes afferent 1307 and efferent 1308 of the stimulus 1306, the relative amplitude and electrical characteristics of the action potentials can be compared. The stimulation waveform and selection of electrodes could therefore be adapted to target the preferred physiology with the neurostimulation and mitigate off-target side effects. For vagus nerve stimulation this may be targeting the brain while reducing stimulation of the central organs. Alternative embodiments of this could utilise single electrode recordings and not VSR. However even with a good insulating layer 1304, direct stimulation artifacts would dominate the signal performance, and thus VSR mitigates this by filtering out signals that simultaneously occur at different electrodes.

[0106] Different configurations of transducers can be used to achieve the desired electrical proximity to a nerve target 1401 and some example embodiments are shown in Figure 14. To achieve implantation through a percutaneous stylet or similar surgical tool, the transducer may comprise a linear configuration of electrodes 1402. The cylindrical electrodes 1402 are separated by insulating lead material in a narrow aspect ratio and the lead axis is implanted parallel to the target nerve axis. The transducer may be provided by a cuff or wrap configuration 1403; this may be preferable to the transducer 1402 by providing electrical insulation between the electrode and surrounding tissues as well as more electrode surface area to transduce electrical activity. However, this comes at the disadvantage of implantation complexity as the nerve has to be excavated to have this array wrapped around and mechanically affixed in place.

[0107] The transducer 1404 comprises a paddle or flat matrix array of electrodes. The transducer 1405 comprises a multi-electrode array arranged on a cuff. The matrix array may be a 4 by 4 array (as shown), but could have different numbers of electrodes in either dimension, such as a 4 by 5 array or a 5 by 5 array. These electrode arrays are examples of utilising multiple electrodes spaced in more than one dimension to provide an ability to electronically configure which electrodes VSR is being used to record electrophysiological activity. The active electrodes 1406 can be selected from the paddle to accommodate angular alignment tolerance with the target nerve 1401 to reduce the complexity of surgery at the expense of a more complex electromechanical assembly. A selection of partially circumferential electrodes 1407 in the multi-electrode cuff 1405 may be made to provide selectivity for recording activity certain parts of the nerve bundle over others. For example, if the nerve is about to bifurcate towards two different organs on the left and right side of the body, if therapy was to be triggered off the response of one but not the other recording from the right side and not the left could be advantageous.

[0108] While the foregoing is directed to exemplary embodiments of the present invention, it will be understood that the present invention is described herein purely by way of example, and modifications of detail can be made within the scope of the invention. Furthermore, one skilled in the art will understand that the present invention may not be limited by the embodiments disclosed herein, or to any details shown in the accompanying figures that are not described in detail herein or defined in the claims. Indeed, such superfluous features may be removed from the figures without prejudice to the present invention.

[0109] Moreover, other and further embodiments of the invention will be apparent to those skilled in the art from consideration of the specification, and may be devised without departing from the basic scope thereof, which is determined by the claims that follow.

Claims

CLAIMS1 . An implantable device for interfacing with a nerve of a human or animal subject, the device comprising: a transducer comprising a sensing array having three or more electrodes arranged along an axis; an actuator configured to output a stimulus; a controller in electrical communication with the electrodes and the actuator, wherein the controller is configured to: receive a plurality of signals from the sensing array; process the signals using velocity selective recording (VSR) to determine a current electrophysiological representation of the subject, wherein VSR uses a plurality of measurement delays to provide a first plurality of velocity filtered signals representative of velocity data of a current signal travelling through the nerve; compare the current electrophysiological representation to a baseline electrophysiological representation, wherein the baseline electrophysiological representation comprises a second plurality of velocity filtered signals representative of velocity data of a baseline signal travelling through the nerve; based on the comparison, identify an electrophysiological biomarker; and control the actuator to output said stimulus based on the electrophysiological biomarker.

2. The implantable device of claim 1 , wherein the current electrophysiological representation comprises a distribution obtained using the first plurality of velocity filtered signals, and the baseline electrophysiological representation comprises a distribution obtained using the second plurality of velocity filtered signals.

3. The implantable device of claim 2, wherein the comparison between the current and baseline electrophysiological representations includes identifying adifference in the distribution at one or more velocities, and identifying the electrophysiological biomarker based on this difference.

4. The implantable device of any preceding claim, wherein the current electrophysiological representation is used by the controller to adjust one or more parameters of the stimulus applied to the subject.

5. The implantable device of any preceding claim, wherein the plurality of signals are received from the sensing array at a biomarker measurement interval.

6. The implantable device of any preceding claim, wherein the controller is further configured to control the actuator to emit a background stimulation to a nerve.

7. The implantable device of any preceding claim, wherein, after controlling the actuator to emit the stimulus, the controller is further configured receive a plurality of signals from the sensing array to measure nerve activation in response to the emitted stimulus.

8. The implantable device of any preceding claim, wherein the actuator is provided by at least one electrode, and wherein the at least one electrode is controlled to emit an electrical signal.

9. The implantable device of claim 8 when dependent upon claim 7, wherein the at least one electrode is provided in a centre of the sensing array, and received signals from the remaining electrodes either side of the at least one electrode are processed by the controller to determine an electrophysiological representation for both afferent and efferent signals.

10. The implantable device of claim 9, wherein the controller is further configured to adjust at least one of: the emitted electrical signal to maximise afferent stimulation and / or minimise efferent stimulation; and the emitted electrical signal to minimize afferent stimulation and / or maximize efferent stimulation.

11. The implantable device of any preceding claim, wherein the actuator is provided by at least one of: a light source, a drug pump, and an ultrasound transducer.

12. The implantable device of any preceding claim, wherein the controller is further configured to update the baseline electrophysiological representation based on the current electrophysiological representation.

13. The implantable device of any preceding claim, further comprising a data storage unit configured to record data corresponding to identification of the electrophysiological biomarker.

14. The implantable device of any preceding claim, further comprising a wireless interface configured to transmit data to and / or from the device.

15. The implantable device of preceding claim, further comprising a rechargeable battery and an interface for connecting a battery-charging device.

16. The implantable device of any preceding claim, wherein the transducer is provided by a cuff configured to substantially surround the nerve, during use.

17. The implantable device of any preceding claim, wherein the transducer comprises an electrode matrix having one or more additional electrodes arranged off the axis of the three or more electrodes.

18. The implantable device of any preceding claim, wherein the controller is housed in an electronics package surrounded by a protective housing.

19. The implantable device of claim 18, wherein the transducer is connected to the electronics package by a flexible lead.

20. The implantable device of claim 18, wherein the transducer is provided as part of the electronics package.21 . The implantable device of any preceding claim, wherein the transducer is configured to provide the actuator and preferably wherein actuator is provided by at least one electrode in the sensing array.

22. A method of controlling the implantable device of any preceding claim, the method comprising: receiving a plurality of signals from the sensing array; processing the signals using velocity selective recording (VSR) to determine a current electrophysiological representation of the subject, wherein the processing using VSR uses a plurality of measurement delays to provide a first plurality of velocity filtered signals representative of velocity data of a current signal travelling through a nerve; comparing the current electrophysiological representation to a baseline electrophysiological representation, wherein the baseline electrophysiological representation comprises a second plurality of velocity filtered signals representative of velocity data of a baseline signals travelling through the nerve; based on the comparison, identifying an electrophysiological biomarker; and determining a stimulus to be applied by the actuator based on the electrophysiological biomarker.

23. The method of claim 22, wherein processing the signals using VSR comprises: delaying each signal from each electrode or each pair of electrodes by a time period that is dependent upon: the distance of the electrode along the longitudinal axis of the transducer, and a target signal measurement velocity; summing the delayed signals together to produce a phase aligned signal for the target signal measurement velocity.

24. The method of claim 23, wherein phased aligned signals are produced over a range of target signal measurement velocities whereby to provide a velocity spectrum.

25. The method of any of claims 22 to 24, wherein the VSR processing is based on bipolar data obtained from pairs of electrodes in the sensing array.

26. The method of any of claims 22 to 25, wherein the identified electrophysiological biomarker corresponds to the onset or occurrence of a neurological event.

27. The method of claim 26, wherein the neurological event is an epileptic seizure.

28. The method of any of claims 22 to 27, wherein the electrophysiological biomarker is identified based on an increase in nerve signals travelling from 5 to 15 m / s, preferably from 7 to 10 m / s.

Images