WO2022214600A1

WO2022214600A1 - Closed-loop autocalibration method for a computer brain interface device, computer program and computer brain interface device

Info

Publication number: WO2022214600A1
Application number: PCT/EP2022/059282
Authority: WO
Inventors: Bálint VÁRKUTI; Saman HAGH GOOIE; Brian Blischak
Original assignee: CereGate GmbH
Priority date: 2021-04-07
Filing date: 2022-04-07
Publication date: 2022-10-13

Abstract

The present disclosure relates to a closed-loop calibration method for updating a current set of stimulation parameters of a neurostimulation, NS, device such as a computer-brain interface, CBI, device, comprising applying, via a neurostimulation interface device operably connected to the NS/CBI device, a burst sequence of stimulation pulses to a plurality of afferent sensory neurons targeting a sensory cortex area involved with decoding information transmitted by the NS/CBI device wherein the sequence of stimulation pulses is associated with the current set of stimulation parameters and wherein the sequence of stimulation pulses is configured to elicit a bioelectric response in the plurality of afferent sensory neurons. The method further comprises recording, via the neurostimulation interface device, the elicited bioelectric response of the stimulated afferent sensory neurons and deriving, based at least in part on the recorded bioelectric response, a neural excitability profile characterizing a non-linear, dynamic excitation behavior of the plurality of afferent sensory neurons corresponding to the applied sequence of stimulation pulses and adjusting, based on the derived excitability profile at least one stimulation parameter of the current set of stimulation parameters to obtain an updated set of stimulation parameters. A corresponding computer program and NS/CBI device is also part of the present disclosure.

Description

CLOSED-LOOP AUTOCALIBRATION METHOD FOR A COMPUTER BRAIN INTERFACE DEVICE, COMPUTER PROGRAM AND COMPUTER BRAIN

INTERFACE DEVICE

1. Technical field

The present disclosure relates to a closed-loop autocalibration method for a computer brain interface, CBI, device (and other types of neurostimulation equipment) that can be used to adjusts stimulation parameters of the CBI device to ensure - inter alia - robust and consistent information transfer from the CBI device to the brain with high fidelity.

2. Technical background

Several promising approaches for implementing a general-purpose CBI are based on implantable neurostimulation systems that typically include one or more neurostimulation electrodes implanted at a desired stimulation site within or close to the nervous system of a person. A neurostimulator (e.g., an implantable pulse generator (IPG)) typically generates neurostimulation signals that are then applied to the neurostimulation electrodes in order to elicit a neural response (e.g., action potentials) in specific parts of the nervous system. For instance, DE to 2019/202666,

US 2020/0269049 and WO 2020/174051 describe such general-purpose CBI devices and systems that use direct neurostimulation of afferent sensory pathways to communicate abstract conceptual information directly to the brain of an individual.

For such CBI devices and systems to work reliably even in normally behaving (e.g. moving) individuals it has to be ensured that a given neurostimulation signal (or sequence of neurostimulation signals) that for instance is associated with a given piece / block of abstract information to be communicated consistently evokes essentially the same neural response in the brain or nervous system of the individual (e.g. a touch sensation in the left hand associated with a movement instruction or balance support cue etc.). In this context, e.g., due to positional sensitivity a problem occurs, when neurostimulation electrodes that have initially been calibrated to elicit a certain neural response when being provided with a specific neurostimulation signal move relative to the stimulation target (e.g., afferent sensory axons / nerve fibers / neurons terminating in a desired sensory cortex area of the individual). For instance, such a situation may occur when during a movement (e.g., a person changing its body posture from standing to sitting or lying, a person bending over, coughing, etc.) the distance between a spinal cord stimulation electrode and the target nerve fibers in the spinal cord changes. As a result, the same neurostimulation signal that has previously been calibrated for a specific relative orientation and / or distance between electrode and spinal cord nerve fiber will not elicit the same desired neural response. In the prior art this distance is sometimes called dCSF. High bandwidth CBIs typically require complex and fine-tuned neurostimulation signals and thus are affected by such problems more severely than conventional neurostimulation equipment, e.g., for pain treatment etc.

The present disclosure provides autocalibration methods for such CBI devices and similar devices and systems that can run autonomously in an on-line and closed-loop manner and that do not require intervention by trained medical personnel.

In this technical context, EP 3229893 Bi discloses a method of communicating along a neural pathway, comprising stimulating the neural pathway at a first location by delivering a peripheral sensory input, in order to evoke neural responses which propagate along the neural pathway, the neural responses being modulated with data, an implanted device sensing the evoked neural responses at a second location spaced apart from the first location along the neural pathway, and the implanted device demodulating the sensed neural responses to retrieve the data, the data being configured to control or alter the operation of the implanted device.

Further, US 10,568,559 B2 relates to a method for determining a desired location at which to apply a neural therapy, the method comprising, implanting an array of electrodes mounted on a common paddle proximal to neural tissue, where a first plurality of electrodes in the array are configured to provide an electrical stimulus, and a second plurality of electrodes in the array are configured to internally measure neural compound action potential responses, applying a stimulus from the array which evokes a neural compound action potential response in the neural tissue proximal to the array using the first plurality of electrodes, obtaining a plurality of simultaneous respective internal measurements of the neural compound action potential response evoked by applying the stimulus using the second plurality of electrodes, the plurality of simultaneous respective internal measurements being obtained from respective distinct measurement amplifiers each connected to respective distinct electrodes of the second plurality of electrodes and determining from the plurality of internal measurements of the neural compound action potential response a neural sensitivity map of the area alongside the array and determining therefrom a desired location at which to apply a neural therapy in order to suppress undesired / adverse stimulation effects such as undesired paresthesia elicited by the neural therapy.

In addition, US 9,713,720 discloses a neurostimulation system configured for providing therapy to a patient, comprises at least one implantable neurostimulation lead configured for being implanted adjacent target tissue of the patient, and an implantable neurostimulator configured for delivering electrical stimulation energy to the implantable neurostimulation leads in accordance with a set of stimulation parameters, and monitoring circuitry configured for taking at least one measurement indicative of a three-dimensional migration of the neurostimulation leads from a baseline position. The neurostimulation system further comprises at least one controller/processor configured for determining whether the three-dimensional migration of the neurostimulation leads from the baseline position has occurred based on the measurements, and, based on the determined three-dimensional migration, generating a new set of stimulation parameters, and reprogramming the implantable neurostimulator with the new set of stimulation parameters. Effectively, US 9,713,720 proposes that an impedance measurement may be used to infer a lead position.

Further prior art relevant for the technical background of the present invention is provided by US 2013/0253299 Ai, US 9,636,497 B2 and by Yadav, A.P., Li, D. & Nicolelis, M.A.L. : “A Brain to Spine Interface for Transferring Artificial Sensory Information”. Sci Rep 10, 900 (2020).

In the latter it is shown that rats are able to consistently discriminate 3 (and likely 4) distinct burst stimulation patterns applied to dorsal column of the spinal cord via modulation of the burst parameters: Pattern 1: 100 pulses at 333Hz; Pattern 2: 1 pulse; Pattern 3: 100 pulses at 100 Hz; Pattern 4: 5 bursts of 20 pulses each, with inter-burst frequency of 2 Hz and inter pulse frequency of 333 Hz. Moreover, the effects of stimulation patterns were also observed in the theta band (5 Hz - 9.5 Hz) spectral power of local-field-potential, LFP, recordings in the motor cortex (Ml), somatosensory cortex (Si), and striatum (STR) in response to stimulation patterns 1 & 2. US 11,045,129 relates to an implantable device for estimating neural recruitment arising from a stimulus that has a plurality of electrodes. A stimulus source provides stimuli to be delivered from the electrodes to neural tissue. Measurement circuitry obtains a measurement of a neural signal sensed at the electrodes. A control unit is configured to control application of a selected stimulus to neural tissue using the stimulus electrodes and after the selected neural stimulus, apply a probe stimulus having a short pulse width. A remnant neural response evoked by the probe stimulus is measured and the control unit estimates from the remnant neural response a neural recruitment caused by the selected neural stimulus.

Further, US 11,129,991 relates to a system configured to deliver electrical stimulation therapy to a patient, the electrical stimulation therapy comprising a plurality of therapy pulses at a predetermined pulse frequency over a period of time and deliver, over the period of time, a plurality of control pulses interleaved with at least some therapy pulses of the plurality of therapy pulses. The system may also be configured to sense, after one or more control pulses and prior to an immediately subsequent therapy pulse of the plurality of therapy pulses, a respective evoked compound action potential,

ECAP, adjust, based on at least one respective ECAP, one or more parameter values that at least partially defines the plurality of therapy pulses, and deliver the electrical stimulation therapy to the patient according to the adjusted one or more parameter values.

US 11,129,987 relates to an Implantable Pulse Generator, IPG, or External Trial Stimulator, ETS, system that is capable of sensing an ECAP, and in conjunction with an external device is capable of adjusting a stimulation program while keeping a location of a Central Point of Stimulation, CPS, constant. Specifically, one or more features of measured ECAP(s) indicative of its shape and size are determined, and compared to thresholds or ranges to modify the electrode configuration of the stimulation program. Further prior art that forms general technical background of the present disclosure is provided by US 9,872,990, US 10,926,092, US 10,940,316, US 10,960,211. 3. Summary

The methods, devices and systems provided by the prior art have various deficiencies. For instance, they may not allow to perform closed-loop and on-line re-calibration of CBI stimulation parameters or only to a very limited extend. In addition, consistency and long-term stability and / or fidelity of desired artificial sensory perceptions / artificial sensations that are to be elicited in specific sensory cortex areas cannot be ensured with the prior art systems, mainly because in the prior art this technical problem faced by CBI devices does not even arise or does not have the same importance as it has for high bandwidth general purpose CBI applications. In essence, several of the prior art systems discussed above utilize closed-loop methods for detecting neural responses to minimize the occurrence of certain reactions / effects (such as paresthesias, pain etc.). To use an analogy, the prior art systems effectively function similar to audio speakers or headset systems that contain microphones which detect the emitted sound level and will automatically lower the volume if a threshold is reached, thereby protecting the user from unpleasant sensation / high volume sound. By contrast, the stringent fidelity requirements that need to be fulfilled in order to establish a general purpose and high-bandwidth CBI device are of a completely different quality and require a fundamentally different approach.

It is thus a problem underlying the present disclosure to overcome such deficiencies of previous technologies by providing novel autocalibration methods for CBI devices and systems.

Generally, the present invention allows to implement closed-loop and on-line autocalibration of a CBI that is based on observing the excitation behavior /neural activation function of afferent sensory nerve fibers that provide a communication pathway to the brain of an individual. This approach is based on the insight that there exist strong correlations between the highly non-linear bioelectric response of an active stimulated afferent sensory nerve fiber or plurality of such fibers and a corresponding artificial sensory perception / artificial sensation elicited in a sensory cortex area of the individual. In loose analogy to standard candles in astronomy, this non-linear bioelectric response essentially serves as a fingerprint of the afferent sensory nerve fiber that can be measured and used for on-line recalibration of neurostimulation signal parameters for direct neurostimulation of afferent sensory axons (e.g. thalamocortical axons, afferent sensory axons of the brain stem or spinal cord and / or afferent sensoiy axons of the peripheral nervous system) targeting directly or indirectly (i.e. via multi-synaptic afferent pathways) sensory neurons in a target sensory cortex area. In this manner, long-term stability of highly specific, fine-grained, and multi dimensional information transfer to the brain can be ensured.

More specifically, in a 1^st aspect, the present invention provides a method for self- calibrating a computer brain interface, CBI, device of an individual, comprising the following steps: choosing a set of test signal parameters, generating, based on the chosen set of test signal parameters, at least one neurostimulation test signal configured to elicit a bioelectric response in one or more afferent sensory nerve fibers, applying the generated neurostimulation test signal to the afferent sensory nerve fibers via a neurostimulation interface operably connected to or integrated with the CBI device, sensing, via the neurostimulation interface, one or more bioelectric responses of the one or more stimulated afferent sensory nerve fibers, determining, based on the sensed bioelectric responses, whether an excitation behavior of the stimulated afferent sensory nerve fibers with respect to the neurostimulation interface has changed, if the excitation behavior has changed, determining, based on the sensed bioelectric responses, a set of recalibrated neurostimulation signal parameters and operating the CBI device, using the recalibrated neurostimulation signal parameters, to communicate information to the individual, via neurostimulation of the one or more afferent sensory nerve fibers. Throughout the present application the terms artificial sensation and artificial sensory perception are used interchangeably. Both terms indicate that an neural excitation pattern is generated in a population of sensory neurons in a sensory cortex area not in response to a sensory stimulus sensed by one of the natural sensory organs of a person (e.g. by mechanoreceptor cells, inner hair cells of the cochlea, rod cells of the retina, etc.) but artificially via direct neurostimulation of afferent sensory pathways using a neurostimulation interface.

According to a 2^nd aspect, in the 1^st aspect, the method may further comprise generating, based on the determined set of recalibrated signal parameters, a communication neurostimulation signal, configured to elicit an artificial sensation in a sensory cortex area of the individual via stimulating the one or more afferent sensory nerve fibers terminating in the specific sensory cortex area, wherein the artificial sensation is associated with a block of information to be communicated by the CBI device. In this manner, consistency and long-term stability of the elicited artificial sensory perceptions that form the syntactic basis for sensory communication can be maintained regardless of tissue alterations, movement or migration of the neurostimulation electrodes relative to the stimulation target and /or changes in the electrical transfer function of the neurostimulation equipment. In essence, the described method thus provides an active compensation mechanism that ensures continuous optimal information transmission into the nervous system throughout changing factors (e.g. movement of electrodes relative to stimulation target) and states (e.g. changes of neural responsivity).

For instance, in a 3^rd aspect, in the 1^st or 2^nd aspect, determining the set of recalibrated neurostimulation signal parameters may comprise: comparing the sensed bioelectric responses to a set of reference bioelectric responses stored in a memory module of the CBI device or obtained via a communication interface of the CBI device.

In this manner, a precise characterization of the excitation behavior of the afferent sensory nerve fibers (e.g. obtained in a dedicated laboratory upon initial calibration of the CBI device and / or the neurostimulation interface) can serve as a reference for on line recalibration, thereby improving the accuracy of recalibration.

For example, in a 4^th aspect, in the 3^rd aspect, the set of reference bioelectric responses may be associated with a set of artificial sensations that can be elicited by the CBI device via the neurostimulation interface in a sensory cortex area of the individual and that are associated with one or more blocks of information that can be communicated via the CBI device to the individual.

In a 5^th aspect, in the 3^rd or 4^th aspect, the method discussed above may further comprise determining the set of reference bioelectric responses based on one or more of the following: an initial or on-line calibration procedure involving the individual providing subjective feedback on artificial sensations elicited by a set of reference neurostimulation test signals; a plurality of reference calibration measurements performed on a plurality of individuals (e.g. a plurality including or not including the individual for whom the CBI and / or neurostimulation interface is to be calibrated) prior to determining the set of reference bioelectric responses for the individual; and an initial or online calibration procedure involving the individual performing a task with objectifiable outcomes that are supported by the operation of the CBI device and recording stimulation parameters and corresponding bioelectric responses that optimize performance of the task without recording subjective feedback by the individual.

For instance, during an initial calibration procedure conducted in a dedicated laboratory setting a set of reference neurostimulation test signals may be applied via the neurostimulation interface and bioelectric responses of the stimulate afferent sensory nerve fibers such as evoked compound action potentials may be measured (see Fig. 3a discussed in section 4. below). Additionally, the individual may provide, e.g. via a microphone, a graphical user interface or a smart phone application etc., subjective feedback on the type, locus, intensity, quality, etc. of the perceived artificial sensations elicited by the set of reference neurostimulation test signals. This subjective feedback can then be combined and correlated with the measured bioelectric response in order to obtain a mapping between a set of specific bioelectric response of the stimulated afferent sensory nerve fibers and a corresponding set of desired artificial sensations perceived by the individual. Further, since the physiologic and functional structure of afferent sensory pathways such as afferent sensory nerve fibers of the spinal cord is stereotypical and strongly conserved across individuals a plurality of reference calibration measurements performed on a plurality of other individuals as described above, may be used for calibrating the CBI and / or neurostimulation interface of a new individual without having to perform a full initial calibration procedure.

In addition, for CBI devices whose operation is for instance directed to support or enhance an action or movement of the individual (e.g. a balance support CBI, a CBI improving gait, motor- coordination, etc.) it is even possible to perform initial calibration in an objective manner without requiring subjective feedback on perceived sensations. For instance, a person could be instructed, e.g. via smart phone application, to perform a set of calibration actions or movements that are supported by the operation of the CBI device, such as walking a certain distance in a straight line. While walking the CBI device could carry out an on-line reference calibration procedure where the CBI device records specific stimulation parameters and corresponding bioelectric responses that optimize the walking performance.

Accordingly, the present disclosure also provides a method for initial calibration of a computer brain interface, CBI, device of an individual, comprising the following steps: choosing a set of initial test signal parameters, generating, based on the chosen set of test signal parameters, at least one reference neurostimulation test signal configured to elicit a bioelectric response in one or more afferent sensory nerve fibers; applying the generated reference neurostimulation test signals to the afferent sensoiy nerve fibers via a neurostimulation interface operably connected to or integrated with the CBI device; sensing via the neurostimulation interface, one or more bioelectric responses of the one or more stimulated afferent sensoiy nerve fibers; obtaining a subjective or objective feedback signal associated with an artificial sensoiy perception elicited by the applied reference neurostimulation test signal; and correlating the chosen set of initial test signal parameters, the corresponding bioelectric responses and the corresponding elicited artificial sensory perceptions.

For instance, a subjective feedback maybe obtained from a graphical user interface, e.g. provided by a smartphone application with which the individual inputs subjective feedback on artificial sensations elicited by the reference neurostimulation test signals.

Objective feedback signals, on the other hand maybe obtained from one or more sensor devices (e.g. accelerometers, LIDAR, gyroscope sensors, etc.) measuring quantities relating to the behavioral state of the individual. Moreover, in a 6^th aspect, in any one of the 1^st to 5^th aspect, a plurality of different neurostimulation test signals may be generated and applied to the afferent sensory nerve fiber interleaved with a plurality of sensing periods for sensing corresponding bioelectric responses of the afferent sensory nerve fibers. For instance, in a 7^th aspect, in the 6^th aspect, the plurality of different neurostimulation test signals are generated such that one or more test signal parameters are varied in a systematic manner in order to estimate a systematic dependence of the excitation behavior of the afferent sensoiy nerve fibers on the one or more systematically varied test signal parameters. In this manner, even complex, multi-dimensional dependence of the excitation behavior of the one or more afferent sensory nerve fibers can be determined fast and efficiently without loss of accuracy.

In this context, in an 8^th aspect, in the 7^th aspect, it may be advantageous that the one or more test signal parameters are varied in form of an increasing or decreasing ramp and / or that the one or more signal parameters comprise one or more of the following: a spatial activation patter of the neurostimulation interface, a signal amplitude, an inter- pulse frequency, an inter-burst frequency, a pulse width, a wave form shape, a density of pulses within a burst, a signal polarity or a burst duration. In a similar manner, the test signal parameters might also be varied according to any of various known design- of-experiment (DOE) methodologies (e.g. methodologies as used when trying to understand the effect of multiple variables to optimize a chemical reaction / process.)

Further, in an 9^th aspect, in any one of the 1^st to 8^th aspect, determining the set of recalibrated neurostimulation signal parameters may comprise fitting a response function to a plurality of data points, wherein each data point comprises a set of test signal parameters and a corresponding bioelectric response level sensed by the CBI device; and / or determining the set of recalibrated neurostimulation signal parameters may comprise aggregating several bioelectric response recordings for the same chosen set of test signal parameters.

In this manner, recalibration does not require sampling the full, potentially multi- dimensional parameter space. Instead, known and / or previously derived functional dependencies of the excitation behavior of the one or more afferent sensory nerve fibers can be taken into account and measuring only a subset of data points (e.g. sparsely distributed throughout the parameter space) may be sufficient to obtain precisely recalibrated neurostimulation parameters suitable for neural communication via the CBI device.

In a 10^th aspect, in any one of the 1^st to 9^th aspect, the sensed bioelectric response may correspond to one or more extracellularly sensed action potentials or local field potentials or evoked compound action potentials (ECAPs) elicited by the at least one neurostimulation test signal in the afferent sensory nerve fibers.

Moreover, in an 11^th aspect, in the 9^th to 10^th aspect, the response function may relate two or more different test signal parameters to an excitation threshold of the afferent sensory nerve fiber (e.g. an excitation threshold for eliciting a specific ECAP in the targeted afferent sensory nerve fibers etc.).

The present invention further provides, in a 12^th aspect, a method for operating a CBI device to communicate information to an individual, comprising: transmitting a plurality of sensory messages to a sensory cortex area of the individual via stimulating one or more afferent sensory nerve fibers terminating in the sensory cortex area, and repeatedly carrying out the steps of the calibration method of any of the preceding claims 1 - 10 interleaved with transmitting of the sensory messages using the respective recalibrated neurostimulation signal parameters.

In this manner, the present invention enables a CBI device to constantly update and adapt its intrinsic stimulation parameter configuration without requiring human intervention or even laboratory-based recalibration.

The present invention also provides a computer program, comprising instructions for carrying out the method described above with respect to the 1^st to 12^th aspect, when being executed by processing and neurostimulation circuitry of a neurostimulation device or system.

The present invention further provides, in a 14^th aspect, a CBI device, comprising one or more stimulation and sensing channels adapted to elicit and sense a bioelectric response of one or more afferent sensory nerve fiber terminating (e.g. mono- or multi- synaptically) in a sensory cortex area of an individual, and data and signal processing circuitry configured to carry out the method described above with respect to the 1^st to 12^th aspect.

Such a CBI device may further comprise a memory module operably connected to the data and signal processing circuitry storing a first mapping between one or more artificial sensations that can be elicited by the CBI device in one or more sensory cortex areas of the individual and one or more bioelectric responses; and / or storing a second mapping between a plurality of sets of neurostimulation signal parameters and a plurality of bioelectric responses of the one or more afferent sensory nerve fibers (e.g. recorded upon initial calibration and / or during on-line recalibration as outline above).

Enhanced autocalibration methods and systems incorporating non-linear dynamic properties of neural excitability

While some prior art approaches typically involve comparing neural activity recordings (e.g., ECAP recordings) to a “ threshold ” or a “threshold value ” and adjusting stimulation parameters based on determining whether such a threshold is crossed or not, the concept of the existence of a threshold is a simplification that, in some prior art application scenarios may still be sufficient e.g., for suppressing unwanted neural responses, e.g., for pain management or Parkinson Disease management devices. However, stimulation parameter calibration methods based on this simplification may fail for advanced CBI-paradigms that involve highly consistent high-fidelity stimulation of artificial sensory perceptions as for instance disclosed in US 2020/0269049 and WO 2020/174051. To ensure CBI stimulation fidelity and consistency even in behaving individuals, the inventors of the present disclosure have found that it is instrumental to take into account the fundamental non-linear and dynamic nature of neural excitability when designing stimulation parameter feedback loops.

A key insight of the present disclosure is that already the scientific studies of Hodgkin and Huxley (Nobel prize in physiology 1963) showed that neurons are non-linear dynamical systems, and thus should be treated and interacted with as such, when designing CBI technology and in particular CBI parameter calibration methods and systems. In general, an afferent sensoiy neuron needs to be described by a set of dynamical variables that describe its state and a dynamical law that describes the evolution of the state variables with time (e.g., a set of coupled differential equations). For example, a proper dynamical description of an afferent sensory neuron maybe based on variables describing neuronal dynamics such as a trans-membrane potential, an activation variable of Na+ currents (e.g., ion-channels), an inactivation variable of Na+ currents and activation variable of a fast K+ currents etc. as well as slowly varying adaptation variables, such as an activation of slow voltage- or Ca2+-dependent transmembrane currents. These adaptation variables may change during prolonged neurostimulation and can affect excitability on an intermediate or even on time scales much longer that the duration of a typical action potential and may even change the type of bifurcation behavior and /or phase space topology underlying the excitability of the stimulated afferent sensory neurons used for establishing the CBI channels to the brain.

Embodiments of the present disclosure allow to capture such complex dynamical system behavior of stimulated afferent sensory neurons and to use this information for enhanced closed-loop online calibration of CBI stimulation parameters. For reference, neuronal excitability is extensively reviewed in “Dynamical Systems in Neuroscience: The Geometry of Excitability and Bursting Eugene M. Izhikevich; The MIT Press, 2007.

Naturally, while discussed in the following with reference to a CBI device, the aspects / embodiments of the present disclosure can also be applied to other types of neurostimulation devices, systems and equipment that might likewise benefit from the present disclosure, such as DBS devices or spinal cord stimulation devices, etc. thar are applied for treating neurological conditions or are used for pain management. Thus, wherever the term “CBI” appears in the following it is to be understood that it also covers such other types of neurostimulation devices, systems and equipment.

For instance, in a further aspect of the present disclosure, a closed-loop calibration method for updating a current set of stimulation parameters of a CBI device is provided that comprises applying, via a neurostimulation interface device operably connected to the CBI device, a burst sequence of stimulation pulses to a plurality of afferent sensory neurons (e.g., of the central nervous system, i.e., of the brain and / or the spinal cord) targeting a sensory cortex area involved with decoding information transmitted by the CBI device, wherein the burst sequence of stimulation pulses is associated with the current set of stimulation parameters and wherein the burst sequence of stimulation pulses is configured to elicit a bioelectric response in the plurality of afferent sensory neurons. The method further comprises recording, via the neurostimulation interface device, the elicited bioelectric response of the stimulated afferent sensory neurons and deriving, based at least in part on the recorded bioelectric response, a neural excitability profile characterizing a non-linear, dynamic excitation behavior of the plurality of afferent sensory neurons corresponding to the applied burst sequence of stimulation pulses and adjusting, based on the derived excitability profile at least one stimulation parameter of the current set of stimulation parameters to obtain an updated set of stimulation parameters.

Importantly, as discussed in more detail below, the derivation of a dynamic excitability profile that is enabled by recording the dynamic bioelectric response associated with a burst sequence of stimulation pulses ensures that non-linear dynamical aspects such as sub-critical bifurcation characteristics, phase space topology as well as time scale separation effects, such as slowly varying (as compared to the action potential dynamics) physiologic adaptations modifying action potential dynamics on medium to long time scales can be appropriately taken into account when adjusting stimulation parameters of the CBI device. In this manner, it can be ensured that the CBI device consistently evokes essentially the same cortical response patterns needed for establishing and maintaining a high CBI channel bandwidth, sensory percept fidelity and information complexity that are all instrumental for general purpose CBI performance. For instance, the burst sequence of stimulation pulses may part of a neurostimulation signal or a signal sequence, applied by the CBI device to elicit an artificial sensoiy perception in a sensoiy cortex area receiving input signals from a subset of the plurality of afferent sensory neurons (e.g., for communicating abstract semantic information or for high fidelity sensory substitution or enhancement). In this manner, for example, no dedicated calibration signals are required, and the excitability profile directly corresponds to the actual operation conditions / parameters of the CBI device.

Further, the burst sequence of stimulation pulses may comprise a burst sequence of essentially identical and / or phasic stimulation pulses. Further, the intra-burst pulse frequency may be at least 50 Hz, preferably at least too Hz, more preferably at least 200 Hz and even more preferably at least 250 Hz. For instance, using such pulse parameters allows to characterize the dynamic excitation behavior and certain phase- space properties such as a bifurcation type, phase-space topology, limit-cycle behavior etc. that otherwise could not be taken into account when adjusting stimulation parameters. Further, a typical value for a burst repetition rate may be 1 to 10 Hz and for a pulse count within a burst sequence may be in the range of 3 to 50, preferably in the range of 5 to 25. For instance, to improve characterization of the non-linear excitability behavior of the stimulated neurons, intra-burst frequency should correspond to the refractory period of the neurons and / or burst repetition rate should allow to capture the slow dynamics of physiologic adaptation processes.

Further, recording the elicited dynamic bioelectric response may comprises recording the bioelectric response while the sequence of stimulation pulses is being applied, preferably after each stimulation pulse or continuously during the sequence and wherein deriving the excitability profile is based at least in part on intra-burst variations of the recorded bioelectric response. Further, a sampling rate of the recording may be larger or equal to the inverse of the time duration between two stimulation pulses of the sequence of stimulation pulses, preferably at least twice as large, more preferably at least 10 times as large and even more preferably at least too times as large. For example, preferably, the recording sampling rate maybe at least 30 kHz or at least 100kHz.

Using such recoding techniques and recoding parameters further enhances how the dynamic excitation behavior of the stimulated plurality of afferent sensory neurons can be characterized.

Further, in particular for being more sensitive to slow physiologic adaptation effects and / or for deriving non-trivial properties of the phase space of the non-linear excitation dynamics, in some embodiments, at least two consecutive burst sequences may be applied, and deriving the excitability profile may be based at least in part on analyzing inter-burst variations of the recorded bioelectric response(s). Optionally, stimulation parameters such as intra-burst pulse frequency, amplitude, polarity, etc. may be varied among the at least two consecutive burst sequences.

For instance, varying the pulse frequency (e.g., of a sequence of individually sub-critical pulses) among consecutive burst sequences may allow to characterize non-linear dynamic phenomena such as non-linear resonances, phase-locking, synchronization, Arnold tongues, bifurcation types, etc.

For example, in some embodiments, deriving the dynamic excitability profile may be based at least in part of correlating the recorded bioelectric response with predictions of a non-linear mathematical model of neuronal excitability, comprising model parameters that vary slowly in time to capture physiologic adaptation mechanisms of the stimulated afferent sensory neurons. For similar reasons as above, in some embodiments, at least two burst sequences may be applied and deriving the dynamic excitability profile may comprises analyzing the bioelectric response(s) corresponding to each stimulation pulse within a burst sequence as well as the joint or total bioelectric response corresponding to each burst sequence.

Additionally of alternatively deriving the excitability profile may comprises analyzing variations among the recorded bioelectric responses within one burst sequence and / or among consecutive burst sequences. To further improve, at least in some embodiments, the characterization of the non linear excitability properties of the stimulated neurons, deriving the excitation profile may also comprise extracting temporal variations or dynamics of recording signal parameters or derived metrics from a plurality of subsequent recordings of the elicited bioelectric response and / or continuous recordings and / or classifying the excitation behavior of a subset of the stimulated afferent sensory neurons may use a closed set of discrete categories and based at least in part on the derived excitability profile.

For instance, such classification into excitability categories maybe based on a metric (such as an absolute value or levels) for one or more signal parameters extracted from the recorded bioelectric responses. In addition, such classification may also be based on dynamical properties such as the type of bifurcation behavior (e.g., sub-critical Poincare- Andronov-Hopf bifurcation, saddle-node bifurcation etc.) underlying the non-linear behavior of the stimulated afferent sensory neurons. In other examples, such classification may also quantify in a discrete manner the distance (e.g., in phase space) from generating an action potential. Optionally, classification may also be based at least in part on analyzing a temporal variation or dynamic of the excitability profile within one burst sequence of stimulation pulses and / or among consecutive burst sequences of stimulation pulses. In some aspects, the elicited bioelectric response(s) may comprise one or more compound action potentials, CAPs, and deriving the excitability profile may comprise determining one or more of: an Ni / P2 amplitude; a number of detectable peaks or minima, a measure of synchrony (in time) among the CAP responses within the sequence or among subsequent sequences of stimulation pulses and / or a delay between a stimulation pulse and the corresponding CAP response.

Further, adjusting the at least one stimulation parameter may comprise, at least in some embodiments, comparing the derived excitability profile with a reference excitability profile.

For example, the reference excitability profile may includes one or more of the following information: an amplitude of a reference bioelectric response, intra-burst variations among bioelectric responses corresponding to single stimulation pulses within a burst sequence, intra-burst variations of the bioelectric response corresponding to the first and the last stimulation pulse within a burst stimulation sequence and inter-burst variations of the bioelectric response.

As discussed in detail in 17/224,953 such a reference bioelectric response maybe stored in a memory module of the CBI device or obtained via a communication interface of the CBI device. Accordingly, in some embodiments, the reference excitability profile may correspond to a specific artificial sensory perception corresponding to a set of reference stimulation parameters associated with the reference excitability profile.

Further, a neurostimulation signal or signal sequence may be applied to a subset of the afferent sensoiy neurons using the updated stimulation parameters wherein the neurostimulation signal may be configured to elicit an artificial sensory perception / percept in a sensory cortex area receiving afferent sensoiy input from the stimulated subset of afferent sensory neurons.

Further, in some embodiments, the calibration method discussed above may incorporate other types physiological signals such as myogenic potentials or recordings from cortical areas as feedback information.

In a further aspect, the present disclosure also provides a computer program (e.g., stored on a memory device or memoiy medium) comprising instructions for cariying out a method according to any of the embodiments discussed above, when these instructions are executed by data and signal processing circuitry of a computer brain interface device, e.g., operably connected to an IPG via a wireless communication interface, such as Bluetooth. Further a computer brain interface, CBI, device is provided that comprises data and signal processing circuitry for carrying out a method according to any of the embodiments discussed above, e.g., when carrying out the instructions of a computer program as discussed above. In some aspects, such a CBI device may comprise one or more stimulation and sensing channels adapted to elicit and sense a bioelectric response of one or more afferent sensory neurons terminating (e.g. mono- or multi-synaptically) in a sensory cortex area. Such a CBI device may further comprise a memory module operably connected to the data and signal processing circuitry storing a first mapping between one or more artificial sensations that can be elicited by the CBI device in one or more sensory cortex areas of the individual and one or more bioelectric responses and / or storing a second mapping between a plurality of sets of neurostimulation signal parameters and a plurality of bioelectric responses of the one or more afferent sensory neurons (e.g. recorded upon initial calibration and / or during on-line recalibration as outline above).

4. Short Description of the Figures

Various aspects of the present disclosure are described in more detail in the following by reference to the accompanying figures. These figures show:

Fig. 1 a diagram illustrating an individual being equipped with a CBI device according to an embodiment of the present invention;

Fig. 2 a functional block circuit diagram illustrating a CBI device according to an embodiment of the present invention;

Fig. 3a a diagram illustrating a set of bioelectric responses recorded from an afferent sensory nerve fiber bundle upon initial calibration of a neurostimulation interface driven by a CBI device according to an embodiment of the present invention;

Fig. 3b a diagram illustrating a plurality of systematic measurements of bioelectric responses of an afferent sensory nerve fiber depending on two different neurostimulation parameters recorded during initial calibration of a CBI device according to an embodiment of the present invention for characterizing the excitation behavior of the stimulated afferent sensory nerve fiber with respect to the neurostimulation interface driven by the CBI device;

Fig. 3c a diagram illustrating a change in the excitation behavior of afferent sensory nerve fibers with respect to a neurostimulation interface driven by a CBI device according to an embodiment of the present invention;

Fig. 4 a diagram illustrating the operation of a CBI device executing an on-line autocalibration method according to an embodiment of the present invention in an interleaved manner with actual information transmissions to the brain of an individual;

Fig. 5 a diagram illustrating a first example of a neurostimulation test signal and a corresponding bioelectric response sensing period executed as part of an autocalibration method according to an embodiment of the present invention;

Fig. 6 a diagram illustrating a second example of a neurostimulation test signal and corresponding bioelectric response sensing periods executed as part of an autocalibration method according to an embodiment of the present invention;

Fig. 7 a diagram illustrating a third example of a neurostimulation test signal and corresponding bioelectric response sensing periods executed as part of an autocalibration method according to an embodiment of the present invention; Fig. 8 a diagram illustrating a fourth example of a neurostimulation test signal and corresponding bioelectric response sensing periods executed as part of an autocalibration method according to an embodiment of the present invention;

Fig. 9 a diagram illustrating a fifths example of a neurostimulation test signal and corresponding bioelectric response recordings executed as part of an autocalibration method according to an embodiment of the present invention;

Fig. lo a diagram illustrating an individual being equipped with a CBI device according to aspects of the present disclosure; Fig. 11 a functional block circuit diagram illustrating a CBI device according to aspects of the present disclosure;

Fig. 12 a diagram illustrating a set of dynamic bioelectric responses recorded from afferent sensory neurons according to aspects of the present disclosure.

Fig. 13 a diagram illustrating examples of temporally varying excitability profiles according to some aspects of the present disclosure;

Fig. 14 a diagram illustrating a basic example of intra-burst recording according to aspects of the present disclosure;

Fig. 15 a diagram illustrating a basic example of inter-burst recording according to aspects of the present disclosure. 5. Detailed Description of some exemplary embodiments

In the following, some exemplary embodiments of the present disclosure are described in more detail, with reference to a CBI device that can be interfaced with neurostimulation electrodes such as spinal cord stimulation electrodes and / or DBS electrodes, e.g., via an intermediate neurostimulation device. However, the present disclosure can also be used with any other neurostimulation interface that is capable of stimulating afferent sensory neurons (e.g., axons, nerve fibers, etc.) of the central or peripheral nervous system targeting directly or indirectly a sensory cortex area of an individual.

While specific feature combinations are described in the following paragraphs with respect to the exemplary embodiments of the present disclosure, it is to be understood that not all features of the discussed embodiments have to be present for realizing the disclosure, which is defined by the subject matter of the claims. The disclosed embodiments may be modified by combining certain features of one embodiment with one or more technically and functionally compatible features of other embodiments. Specifically, the skilled person will understand that features, components and / or functional elements of one embodiment can be combined with technically compatible features, components and / or functional elements of any other embodiment of the present disclosure as long as covered by the invention specified by the appended claims.

Moreover, the various modules of the devices and systems disclosed herein can for instance be implemented in hardware, software or a combination thereof. For instance, the various modules of the devices and systems disclosed herein may be implemented via application specific hardware components such as application specific integrated circuits, ASICs, and / or field programmable gate arrays, FPGAs, and / or similar components and / or application specific software modules being executed on multi purpose data and signal processing equipment such as CPUs, DSPs and / or systems on a chip, SOCs, or similar components or any combination thereof.

For instance, the various modules of the CBI devices discussed herein above may be implemented on a multi-purpose data and signal processing device (e.g., a smart phone) configured for executing application specific software modules and for communicating with various sensor devices and / or neurostimulation devices or systems via conventional wireless communication interfaces such as a NFC, a WIFI and / or a Bluetooth interface.

Alternatively, the various modules of the CBI devices discussed in the present application may also be part of an integrated neurostimulation apparatus, further comprising specialized electronic circuitry (e.g. neurostimulation signal generators, amplifiers etc.) for generating and applying the determined neurostimulation signals to a neurostimulation interface of the individual (e.g. a multi-contact electrode, a spinal cord stimulation electrode, peripheral sensory nerve stimulation electrode etc.) and for recoding the bioelectric responses as disclosed herein. As discussed above the present disclosure may be realized in situations where the perceptual channels of a general-purpose CBI are not calibrated via subject- experimenter interactions. Instead, the CBI stimulation parameters can be self- calibrated by tapping into the neural activity of the tissue in vicinity of the stimulation interface. For instance, the level of induced bioelectric activation can be measured by interleaving recording bioelectric responses elicited by burst sequences of stimulation pulses.

In some examples, special test waveforms maybe defined by modulating various aspect of the waveform in bursting mode. The modulated parameters of the waveform may include but are not limited to: a spatial activation pattern of the electrode contacts, an amplitude, an inter-pulse frequency, an inter-burst frequency, a pulse width, a wave form shape (e.g. mono-phasic, biphasic with symmetric or with long active discharge period, multiphasic, etc.), a density of pulses within a burst or a burst duration. In an exemplary stimulation paradigm, a few symmetric pulses (e.g., in a range of 4 - 9 pulses) are delivered within short bursts (e.g., lasting 40 ms - 60 ms) to convey information related to intensity of sensation. The intensity can then be varied at a second measurement of loci point in time by changing density of pulses per burst while keeping pulse numbers constant i.e. shortening duration but increase intra-burst frequency and vice-versa.

For instance, neural recordings / sensing of bioelectric responses may take place by ramping stimulation signal bursts in repetition, aggregate frequency power pre- and post-pulse for each step of the ramp across repeated bursts then create differential response profile to pulses with varied intensity for the same purpose, so that the CBI device can estimate the neural excitation behavior of the stimulated afferent sensory nerve fibers by fitting a response function to the amplitude of the ECAP or theta frequency band of the ECAPs taking into account the response at every intensity increment. As stated above the excitation behavior can also be estimated not only by varying the amplitude of the burst in a ramp by also by changing other parameters of the stimulation such as frequency, pulse width, as well as the inter burst intervals, for example. The estimated dynamic excitability profile then allows to determine optimal stimulation parameters which are adequate to generate desired level of activity in the target tissue thereby stabilizing the intensity, locus and / or quality of artificial sensory perceptions in the targeted sensory cortex area. This may be achieved, for example, by determining the highest value parameter coefficients which crucially contribute to determination of sensation intensity.

Figure 1 illustrates a person / individual 100 that is equipped with a CBI device as described in section 3 above. In the illustrated embodiment, the CBI is implemented via direct neurostimulation of afferent sensory nerve fibers in the spinal cord 106 via one or more multi-contact electrodes 104 driven by an IPG 102 that may be operatively connected to or integrated with a CBI device as disclosed herein.

For establishing a perceptual communication channel to the brain of the individual too the CBI device must be calibrated such that neurostimulation signals generated by the CBI device and applied via the IGP 102 and the multi-contact electrode 104 elicit one or more action potentials 108 in one or more afferent sensory nerve fibers of the spinal cord 106 targeting (e.g. via multi-synaptic afferent sensory pathways) one or more sensory cortex areas 110 of the individual where the one or more action potentials 108 generate artificial sensory perceptions that can be used to communicate with the individual too. As discussed in detail in US 2020/0269049, fully incorporated herein by reference, artificial sensory perceptions that are elicited in a sensory cortex area (e.g. a sensory cortex area processing touch sensations on the left or right hand) can be associated with any kind of abstract information that is intelligible (i.e. consciously or subconsciously) by the individual. Figure 2 shows an exemplary CBI device according to an embodiment of the present invention. In this embodiment, the CBI device comprises an integrated neurostimulation and sensing module 230 (e.g. comprising a neuronal signal generator and an output amplifier as well as a sensing amplifier and an analog to digital converted) that is connected to a plurality of output signal leads 270 and a plurality of separate or identical sensing signal leads 280 that may be interfaced with a neurostimulation interface of the individual (e.g. a multi-contact spinal cord stimulation electrode such as the electrode 104 shown in Fig. 1). The CBI device may further comprise a communication antenna 260 operably connected to a communication interface module 210, configured for wireless communication (e.g. via NFC, Bluetooth, or a similar wireless communication technology).

The communication interface module 210 may be configured, for example, to receive one or more sensor signals from one or more sensors (not shown; e.g. acceleration signals obtained form an accelerometer etc.) and / or control information from a control device such as a remote control or a smart phone. The communication interface module 210 is operably connected to a data / signal processing module 220 configured to generate one or more neurostimulation signals and /or signal parameters (e.g. waveform, pulse shape, amplitude, frequency, burst count, burst duration etc.) for generating the one or more neurostimulation signals. For instance the processing module 220 may access a data storage module 240 configured to store a plurality of relations, specific for the individual, associating a plurality of neurostimulation signals (or parameters used for generating a plurality of neurostimulation signals) with a plurality of corresponding pieces of information to be communicated to the individual.

The generated neurostimulation signals and / or the signal parameters are input into the integrated neurostimulation and sensing module 230 that may be configured to process (e.g. modulate, switch, amplify, covert, rectify, multiplex, phase shift, etc.) the one or more neurostimulation signals generated by the processing module 220 or to generate the one or more neurostimulation signals based on the signal parameters provided by the processing module 220.

The generated and processed neurostimulation signals are then output by the neurostimulation and sensing module 230 and can be applied to one or more electric contacts of a neurostimulation electrode (e.g. a DBS electrode or spinal cord stimulation electrode as shown in Fig. l) via output leads 270. The CBI device of Fig. 2 may also comprise a rechargeable power source 250 that, for instance maybe wirelessly charged via a wireless charging interface 265.

As discussed above, the data / signal processing module 220 may be further configured to, e.g. in conjunction with the data storage module 240 and the neurostimulation and sensing module 230, to execute an on-line autocalibration method as discussed in section 3 above. For example, it may generate one or more neurostimulation test signals (for examples see Figs. 5 - 9 below) configured to elicit a bioelectric response in one or more afferent sensory nerve fibers such as an evoked (compound) action potential in one or more afferent sensory nerve fibers of the spinal cord 106 shown in Fig. 1. The neurostimulation test signal(s) (e.g. a combined intensity and burst-duration ramp; see Fig. 8) may then be applied via output stimulation leads 270 to a neurostimulation interface such as the most caudal contact 112 of the multi-contact electrode 104 shown in Fig. 1. The neurostimulation and sensing module 230 may then sense, via the neurostimulation interface (e.g. via the most rostral contact 114), a bioelectric response 108 of the stimulated afferent sensory nerve fiber of the spinal cord

106.

Based on the sensed bioelectric response, the excitation behavior of the stimulated afferent sensory nerve fibers with respect to the neurostimulation interface can then be estimated by the neurostimulation and sensing module 230 and / or the processing module 220. As discussed above (e.g. see section 3), based on the sensed bioelectric responses, a set of recalibrated neurostimulation signal parameters can then be determined and stored in the data storage module 240 for later use, e.g. for operating the CBI device to transmit information via the afferent sensory nerve fibers of the spinal cord 106 to a sensory cortex area 110 of the individual too.

Figure 3a illustrates exemplary reference bioelectric responses 310, 320, 330 (e.g. extracellularly sensed ECAPs) of a sub-population of afferent sensory nerve fibers (e.g. of the spinal cord 106; see Fig. 1) sensed and recorded upon initial calibration of a neurostimulation interface (e.g. the multi-contact spinal cord stimulation electrode 104 shown in Fig. 1) driven by a CBI device (see Fig. 2) according to an embodiment of the present invention. The illustrated bioelectric responses are sensed after neurostimulation using different sets of stimulation parameters such as different values for stimulus strength and duration. The bioelectric response 310 corresponds to a set of stimulation parameters that result in a sub-threshold stimulation of the targeted afferent sensory nerve fibers (see data points indicated with a symbol in Fig. 3b and Fig. 3c). Consequently, no action potentials are elicited, and no artificial sensation can be elicited in the sensoiy cortex area(s), in which the targeted nerve fibers ultimately terminate. During initial calibration such a stimulation signal would thus not trigger individual to provide positive subjective feedback. The waveform of the bioelectric response 310 could then be stored in a memoiy module of the CBI device as a reference example of a sub-threshold bioelectric response that should be avoided when cariying out the autocalibration method as described above.

The bioelectric response 320 corresponds to a combination of stimulation parameters resulting in a supra-threshold stimulation of the targeted nerve fiber (see data points indicated with an “x” symbol in Fig. 3b and Fig. 3c) and eliciting a desired artificial sensoiy perception in a sensory cortex area of the individual, such as a mild tingling touch sensation on the left index finger that is clearly perceivable by the individual but is not unpleasant or painful. The waveform of the bioelectric response 320 could then be stored in a memoiy module of the CBI device as a reference example of a desired supra-threshold bioelectric response when cariying out the autocalibration method as described above.

Here it is important to note that that the terms “sub-threshold” & “supra-threshold” are tied to axonal activation (i.e. action potential generation in the targeted nerve fiber(s)) but are not necessarily tied to conscious perception. In other words, in some embodiments, a CBI device as disclosed herein may provide behavioral benefits (such as balance support or gait improvement cues) by generating supra-threshold spinal cord activation for which, however, the artificial sensations remain sub-conscious, e.g. after training and sematic calibration of the CBI device.

In such a configuration, the sematic calibration of the CBI device would be done conventionally with consciously reported sensations and objective behavioral tests and then one could gradually diminish the intensities (or different signal parameters) until the subject does no longer report any conscious perception but a behavioral benefit still persists. In other words, due to training (similar as when learning Braille script or Morse code) it is possible that the subject does not report the same conscious percepts anymore but can still intelligibly process the communicated information.

The bioelectric response 330 corresponds to a combination of stimulation parameters resulting in a supra-threshold stimulation of the targeted nerve fiber (see data points indicated with an “0” symbol in Fig. 3b and Fig. 3c) and eliciting a stronger (e.g. different or undesired) artificial sensory perception (e.g. too strong, wrong type of sensation, wrong locus of sensation, etc.) in a sensory cortex area of the individual, such as an unpleasant touch sensation on the abdomen of the individual. The waveform of the bioelectric response 330 could then be stored in a memory module of the CBI device as a reference example of a desired supra-threshold bioelectric response having an alternate sematic meaning or as an undesired supra-threshold bioelectric response when carrying out the autocalibration method as described above.

Naturally, several different bioelectric responses such as the response 320 may result in similar or essentially identical artificial sensations that may all be used for communicating the same block of information to the individual. The more response waveforms that are labeled with “desired / undesired” are stored in memory the better the on-line autocalibration method discussed above performs.

In order to serve as references for later use in an autocalibration method as discussed above, the stimulation parameters are thus associated, during initial calibration of the CBI device, to a threshold for eliciting an active, non-linear bioelectric response of the respective nerve fiber (e.g. an bioelectric response such as an ECAP having a specific intensity and signal shape). The excitation threshold of a nerve fiber may depend, inter alia, on the electric transfer function of the neurostimulation equipment, on the distance and relative orientation between stimulation contact and nerve fiber, the electric properties of the tissue surrounding the stimulation site, and the bioelectric properties (e.g. Na-ion and K-ion channel density) of the targeted nerve fiber etc.

By recording, upon initial calibration of the CBI device, the wave-form of bioelectric responses of the nerve fibers and by relating them activation threshold and a plurality of desired artificial sensations the CBI device can later be auto-calibrated in an on-line manner by observing the bioelectric responses of the nerve fiber alone without the need to record cortical activity patterns and /or without the individual participating in a laboratory-based calibration procedure. Fig. 3b shows a diagram illustrating a plurality of systematic measurements of bioelectric responses of a sub-population of afferent sensory nerve fibers depending on two different neurostimulation parameters recorded / sensed during initial calibration of a CBI device according to an embodiment of the present invention. By such a collection of measurements across a two or higher-dimensional stimulation parameter space the excitation behavior of the stimulated afferent sensory nerve fibers with respect to the neurostimulation interface driven by the CBI device can accurately be characterized.

For instance, data points are sampled from the parameter space and bioelectric response (e.g. ECAPs) are recorded. Some parameter combination elicit no responses (data points indicated with

some elicit responses that the individual has reported as sensation intensity level 1 (data points indicated with “x”) and some as sensation intensity level 2 or higher (data points indicated with “0”)· It is simpler to retune the function if subjective perceptions of subjects are clearly mapped to the bio response signatures (e.g. during an initial calibration sessions) but this is not critically relevant. More important is that the dotted-line 340 (or a hyperplane for higher dimensional parameter spaces) delineates the dividing line between response and no response and this line is altered upon e.g. moving the implant or moving the spine in a certain way thereby changing the excitation behavior of the targeted afferent sensory nerve fiber. Further, such functions can be calculated (e.g. to differentiate x & 0 intensity levels), so that shells of dividing lines can be pictured. In a certain sense they are also isolines - in the sense that alle parameters that trigger a response along that isoline trigger similar intensity responses. In the following a numerical example is given to illustrate the underlying concepts: Example 1: Initial Calibration Session:

- 0.5 mA; 500 ms burst duration: No Subjective Response / No ECAP

- 2.5 mA; 500 ms burst duration: No Subjective Response / Weak ECAP

For example, such bioelectric responses may be associated during a sematic training session with a too weak stimulation that should be avoided. Accordingly, prior to re calibration using the autocalibration method described above, this combination of stimulation parameters will not be used by the CBI device. - 2.5 mA; looo ms burst duration: Response - Intensity weak perceived / Middle ECAP

For example, such a bioelectric response maybe associated during a sematic training session with a piece / block of information to be communicated via the CBI device (e.g. the letter “A”).

3.5 mA; 500 ms burst duration: Response - Intensity middle perceived / Strong ECAP

For example, such a bioelectric response maybe associated during a sematic training session with a further piece / block of information to be communicated via the CBI device (e.g. the letter “B”).

- 3.5 mA; 1000 ms burst duration: Response - Intensity very strong perceived /

Very Strong ECAP.

For example, such a bioelectric response may be associated during a sematic training session with an unpleasant artificial sensation to be avoided. Accordingly, prior to re calibration using the autocalibration method described above, this combination of stimulation parameters will not be used by the CBI device.

Figure 3c shows a diagram illustrating a change in the excitation behavior of the afferent sensory nerve fibers on which the initial calibration procedure resulting in Fig. 3a was performed. For instance, in this example, the position of the test signal stimulation contact moved with respect to the targeted afferent sensory nerve fiber. In the shown example, the dotted delimiting line 304 shifted towards the origin of the diagram as compare to the reference calibration procedure of Fig. 3b. In such a configuration, not changing stimulation parameters may well lead to an overstimulation of the targeted afferent sensory nerve fiber thereby degrading the quality of the corresponding perceptual communication channel.

However, the present invention allows, based on the reference bioelectric responses recorded for Fig. 3b, to recalibrate stimulation parameters (e.g. intensity and / or pulse duration, etc.) and thereby maintain the quality of the corresponding perceptual communication channel. Figure 4 shows an exemplary sequence of operation of a CBI device executing an on line auto-calibration method according to an embodiment of the present invention in an interleaved manner with actual information transmissions to the brain of an individual. In this automated routine channel checks are interleaved with blocks of information transmission. Specifically, channel check periods 420 are interleaved with data transmission periods 420. Each channel check period 420 involves application of neurostimulation test signals and recording of a bioelectric response of the stimulated afferent sensory nerve fiber as discussed above. For instance, the examplary test signal - sensing sequences illustrated in Figs. 5 - 9 maybe used during such a channel check period 420.

Based on the sensed bioelectric response of the stimulated afferent sensory nerve fiber(s) a current activation function of the nerve fiber can be determined and compared to a reference activation function (see Fig. 4). If deviations from the reference activation function(s) are detected stimulation parameters can be re- calibrated 430. For instance, a set of recalibrated stimulation parameters (intensity, duration, pulse width, etc.) maybe determined and then be used for a subsequent data transmission 410. In this manner, the intensity, quality, and / or locus of the corresponding artificial sensory perceptions can be stabilized. For instance, the present excitation behavior (see Fig. 3c above) can be estimated by utilizing bioelectric response measurements using at least two types of stimulation parameters such as the illustrated intensity and pulse duration.

The activation curve can however include other modalities or other dimensions (e.g. multi-dimensional activation curves). Importantly, a full re-sampling of the activation curve is not absolutely necessary since in many configurations a sparse sampling approach indicating a rheobase and chronaxie values would be enough to estimate the activation function without having to move through parameter space in brute force. As a result, the properties such as the channel bandwidth of the corresponding perceptual communication channel can be maintained even in normally behaving subjects during a broad range of daily activities. Figure 5 illustrates a test signal / recording configuration where the CBI device delivers (e.g. via a neurostimulation module; see Fig. 2) or commands an implanted stimulator to deliver whole bursts of test stimuli then waits for detection and recording of the induced bioelectric responses (e.g. action potentials, ECAPs, etc.) after the last pulse stimuli is applied. In this first example, stimulation parameters during burst stimulation 510 remain constant and bioelectric response (e.g. ECAP) measurement 520 takes place after the last pulse iteration within the burst.

Figure 6 shows a test signal / recording sequence where the intensity of the neurostimulation test signal 610 is ramped and bioelectric response (e.g. ECAP) measurement 620 takes place after each pulse iteration within the burst.

Figure 7 shows a test signal / recording sequence where the intensity of the neurostimulation test signal 710 is kept constant and the pulse duration is ramped. As in Fig. 6 bioelectric response (e.g. ECAP) measurement 720 takes place after each pulse iteration 710 within the burst.

Figure 8 shows a combined intensity and pulse duration ramp, that for instance may be used to efficiently estimate a two-dimensional activation function (e.g. see Fig. 3b and 3c). As in Fig. 6 and Fig. 7 bioelectric response (e.g. ECAP) measurement 820 takes place after each pulse iteration 810 within the burst sequence.

Figure 9 illustrates how averaging across multiple channel check sequences can improve data quality and thus make the estimation of the current activation function more precise and noise tolerant.

Figure 10 illustrates a person / individual iooc that is equipped with a CBI device as described in section 3 “Summary” above. In the illustrated embodiment, the CBI is implemented via direct neurostimulation of afferent sensory nerve fibers / neurons in the spinal cord io6x via one or more multi-contact electrodes 104X driven by an IPG i02x that may be operatively connected to or integrated with a CBI device as disclosed herein.

For establishing a perceptual communication channel to the brain of the individual loox the CBI device typically is calibrated such that neurostimulation signals generated by the CBI device and applied via the IGP i02x and the multi-contact electrode 104X elicit one or more action potentials io8x in one or more afferent sensory nerve fibers of the spinal cord io6x targeting (e.g. via multi-synaptic afferent sensory pathways) one or more sensory cortex areas nox of the individual where the one or more action potentials io8x generate artificial sensory perceptions that can be used to communicate with the individual iooc. As discussed in detail in US 2020/0269049, fully incorporated herein by reference, artificial sensory perceptions that are elicited in a sensoiy cortex area (e.g., a sensory cortex area processing touch sensation on the left or right hand) can be associated with any kind of abstract information that is intelligible (i.e., consciously or subconsciously) by the individual.

While Fig. 10 shows orthodromically recoding, bioelectric responses may also be recorded differently, such as antiorthodromically. Figure 11 shows an exemplary CBI device according to an embodiment of the present invention. In this embodiment, the CBI device comprises an integrated neurostimulation and sensing module 230X (e.g. comprising a neuronal signal generator and an output amplifier as well as a sensing amplifier and an analog to digital converted) that is connected to a plurality of output signal leads 270X and a plurality of separate or identical sensing signal leads 28ox that may be interfaced with a neurostimulation interface of the individual (e.g. a multi-contact spinal cord stimulation electrode such as the electrode 104X shown in Fig. 10). The CBI device may further comprise a communication antenna 2όoc operably connected to a communication interface module 2iox, configured for wireless communication (e.g., via NFC, Bluetooth, or a similar wireless communication technology).

The communication interface module 2iox may be configured, for example, to receive one or more sensor signals from one or more sensors (not shown; e.g., acceleration signals obtained form an accelerometer etc.) and / or control information from a control device such as a remote control or a smart phone. The communication interface module 2iox is operably connected to a data / signal processing module 220x configured to generate one or more neurostimulation signals and /or signal parameters (e.g., waveform, pulse shape, amplitude, frequency, burst count, burst duration etc.) for generating the one or more neurostimulation signals. For instance, the processing module 220x may access a data storage module 240X configured to store a plurality of relations, specific for the individual, associating a plurality of neurostimulation signals (or parameters used for generating a plurality of neurostimulation signals) with a plurality of corresponding pieces of information to be communicated to the individual. The generated neurostimulation signals and / or the signal parameters are input into the integrated neurostimulation and sensing module 230X that may be configured to process (e.g., modulate, switch, amplify, covert, rectify, multiplex, phase shift, etc.) the one or more neurostimulation signals generated by the processing module 220x or to generate the one or more neurostimulation signals (e.g., burst sequences of stimulation pulses as discussed in the present disclosure) based on the signal parameters provided by the processing module 220x.

The generated and processed neurostimulation signals are then output by the neurostimulation and sensing module 230X and can be applied to one or more electric contacts of a neurostimulation electrode (e.g., a DBS electrode or spinal cord stimulation electrode as shown in Fig. 10) via output leads 27OX. The CBI device of Fig. 11 may also comprise a rechargeable power source 250X that, for instance may be wirelessly charged via a wireless charging interface 265X.

As discussed above, the data / signal processing module 220x may be further configured to, e.g., in conjunction with the data storage module 240X and the neurostimulation and sensing module 230X, to execute a closed-loop, on-line autocalibration method as discussed and detail above and below. For example, it may generate one or more burst sequences of stimulation pulses (for examples see Fig. 14 and Fig. 15 below) configured to elicit a bioelectric response in one or more afferent sensory nerve fibers / neurons such as an evoked (compound) action potential in one or more afferent sensory nerve fibers / neurons of the spinal cord io6x as shown in Fig.

10.

The burst sequences of stimulation pulses may then be applied via output stimulation leads 270X to a neurostimulation interface such as the most caudal contact H2x of the multi-contact electrode 104X shown in Fig. 10. The neurostimulation and sensing module 230X may then sense, via the neurostimulation interface (e.g., via the most rostral contact 114.x), a bioelectric response io8x of the stimulated afferent sensory nerve fiber of the spinal cord io6x.

Based on the sensed bioelectric response(s), the excitation behavior of the stimulated afferent sensory nerve fibers / neurons with respect to the neurostimulation interface can then be estimated by the neurostimulation and sensing module 230X and / or the processing module 220x. As discussed above (e.g., see section 3 “Summary”), based on the sensed bioelectric responses, a dynamic excitability profile can be derived and used for closed-loop stimulation parameter adaptation and / or stored in data storage module 240X for later use, e.g., for determining slowly varying physiologic adaptation processes as discussed above.

Figure 12 illustrates exemplary bioelectric responses 3iox, 320X, 330X (e.g., extracellularly sensed (E)CAPs) of a sub-population of afferent sensory nerve fibers / neurons (e.g., of the spinal cord io6x; see Fig. 10) sensed and recorded during application of a burst sequence of stimulation pulses according to aspects of the present disclosure (e.g., applied via the multi-contact spinal cord stimulation electrode 104X shown in Fig. 10) driven by a CBI device (see Fig. 11) according to an embodiment of the present disclosure. The illustrated bioelectric responses are sensed / recorded while several (e.g., consecutive) pulses within a burst sequence (see for example Fig. 14 below) are applied. Although the stimulation parameters for each pulse are kept constant, the bioelectric response changes substantially due to the non-linear nature of neuronal excitability as discussed above.

Based on such recordings of bioelectric responses the temporal dynamics of neuronal excitability can be derived and used for deriving the excitability profiles discussed in detail above.

Figure 13 illustrates three examples of such excitability profiles. On the x-axis pulse progression within a burst or several subsequent bursts is indicated. On the y-axis an excitation profile parameter such as the amplitude of the first peak or the difference of the second peak and the first valley or the delay between pulse and first peak or any other suitable recording signal parameter or metric as discussed above is plotted as function of burst progression.

The three exemplary traces 4iox, 420X and 43OX may correspond to three different burst sequences each using different stimulation parameters. For instance, trace 410 may correspond to a set of parameters that do not result in (compound) action potential generation, trace 420X may correspond to a set of parameters that may result in in-consistent excitation behavior and trace 430X may correspond to a set of parameters that consistently evoke (compound) action potentials in a subset of the stimulated plurality of afferent sensory nerve fibers / neurons. In other situations, the three traces may also be recorded in subsequent stimulation trials with essentially identical pulse parameters, e.g., in situation where slow physiologic changes fundamentally shift the dynamic excitation behavior of the stimulated neurons.

As can be seen from Fig. 13 deriving a whole excitability profile is necessary to determine with high probability which set of stimulation parameters actually results in consistent action potential generation and thus should be used for operating the CBI device to transmit information to the brain. For instance, by just comparing the initial part of the three traces would not allow to characterize the excitation behavior and could thus result in wrongly adjusted stimulation parameters.

Auto-calibration of the perceptual channels of the CBI device can then be achieved by using a neural interface capable of stimulation and recording from the neural tissue. The derived excitability profiles and their dynamics may be compared after each individual stimulation pulse within a burst and / or between bursts in a trial to automatically determine the effectiveness of stimulation settings and establish various sensation levels within perceptual channels. Given similar stimulation parameters in each burst, the inter-burst dynamics of the excitability profile 430X exhibit a distinct shape compared to an undesired excitability profile as function of burst progression. It should be noted that although profile 420X may (locally) exhibit a higher intensity response, the stimulation cannot maintain an increasing profile evolution.

Figure 14 illustrates an intra-burst sequence recording configuration where the CBI device delivers (e.g., via a neurostimulation module; see Fig. 11) or commands an implanted stimulator to deliver a burst 510X of essentially identical stimulation pulses and records the induced bioelectric responses (e.g., action potentials, ECAPs, etc.) while the burst is applied. For instance, the induced bioelectric responses may be recorded 520X after the first and after the last stimulation pulse 530X. In other embodiments, recordings may take place after each stimulation pulse within the burst or throughout in an essentially continuous manner (e.g., with a sampling rate of 100kHz) as discusses above.

Figure 15 illustrates an inter-burst sequence recording configuration where the CBI device delivers (e.g., via a neurostimulation module; see Fig. 11) or commands an implanted stimulator to deliver a sequence of bursts 6iox of essentially identical stimulation pulses and records 620X the induced bioelectric responses (e.g., action potentials, ECAPs, resting potential, depolarization, etc.) after each burst sequence and optionally, also while each burst sequence is applied as illustrated in Fig. 14 and discussed above. Such a stimulation and recording sequence may enable the CBI to detect slowly varying variables that might affect neuronal excitability on medium to long time scales as also discussed above.

Claims

1. A method for self-calibrating a computer brain interface, CBI, device of an individual, comprising the following steps: selecting a set of test signal parameters; generating, based on the chosen set of test signal parameters, at least one neurostimulation test signal configured to elicit a bioelectric response in one or more afferent sensory nerve fibers; applying the generated neurostimulation test signal to the afferent sensory nerve fibers via a neurostimulation interface operably connected to or integrated with the CBI device; sensing via the neurostimulation interface, one or more bioelectric responses of the one or more stimulated afferent sensory nerve fibers; determining, based on the sensed bioelectric responses, whether an excitation behavior of the stimulated afferent sensory nerve fibers with respect to the neurostimulation interface has changed; when the excitation behavior has changed, determining, based on the sensed bioelectric responses, a set of recalibrated neurostimulation signal parameters; and operating the CBI device, using the recalibrated neurostimulation signal parameters, to communicate information to the individual, via neurostimulation of the one or more afferent sensory nerve fibers.

2. The method of claim 1, further comprising: generating, based on the determined set of recalibrated signal parameters, a communication neurostimulation signal, configured to elicit an artificial sensation in a sensoiy cortex area via stimulating the one or more afferent sensory nerve fiber terminating in the specific sensoiy cortex area, wherein the artificial sensation is associated with a block of information to be communicated by the CBI device.

3. The method of claim 1 or 2, wherein determining the set of recalibrated neurostimulation signal parameters comprises: comparing the sensed bioelectric responses to a set of reference bioelectric responses stored in a memory module of the CBI device or obtained via a communication interface of the CBI device.

4. The method of claim 3, wherein the set of reference bioelectric responses is associated with a set of artificial sensations that can be elicited by the CBI device via the neurostimulation interface in a sensory cortex area of the individual and that are associated with one or more blocks of information that can be communicated via the CBI device to the individual.

5. Method of claim 3 or 4, further comprising: determining the set of reference bioelectric responses based on one or more of: an initial or on-line calibration procedure involving the individual providing subjective feedback on artificial sensations elicited by a set of reference neurostimulation test signals; a plurality of reference calibration measurements performed on a plurality of individuals prior to determining the set of reference bioelectric responses for the individual; and an initial or online calibration procedure involving the individual performing one or more tasks with objectifiable outcomes that are supported by the operation of the CBI device and recording stimulation parameters and corresponding bioelectric responses that optimize performance of the task without recording subjective feedback by the individual.

6. The method of any of the claims 1 - 5, wherein a plurality of different neurostimulation test signals are generated and applied to the afferent sensory nerve fibers interleaved with a plurality of sensing periods of corresponding bioelectric responses of the afferent sensory nerve fibers.

7. The method of claim 6, wherein the plurality of different neurostimulation test signals are generated such that one or more test signal parameters are varied in a systematic manner in order to estimate a systematic dependence of the excitation behavior of the afferent sensory nerve fibers on the one or more systematically varied test signal parameters.

8. The method of claim 7, wherein the one or more test signal parameters are varied in form of an increasing or decreasing ramp; and / or wherein the one or more signal parameters comprise one or more of the following: a spatial activation pattern of the neurostimulation interface, a signal amplitude, an inter-pulse frequency, an inter-burst frequency, a pulse width, a wave form shape, a density of pulses within a burst, signal polarity or a burst duration.

9. The method of any of the preceding claims, wherein determining the set of recalibrated neurostimulation signal parameters comprises fitting a response function to a plurality of data points, wherein each data point comprises a set of test signal parameters and a corresponding bioelectric response level sensed by the CBI device; and / or wherein determining the set of recalibrated neurostimulation signal parameters comprises aggregating several bioelectric response recordings for the same chosen set of test signal parameters.

10. The method of any of the preceding claims, wherein the sensed bioelectric responses correspond to one or more extracellularly recorded action potentials and / or local field potentials and /or evoked compound action potentials elicited by the at least one neurostimulation test signal.

11. The method of any of the preceding claims 9 or 10, wherein the response function relates two or more different test signal parameters to an excitation threshold of the afferent sensory nerve fiber.

12. The method of any of the preceding claims , the method further comprising: repeating the method for self-calibrating the CBI device to obtain updated sets of recalibrated neurostimulation signal parameters; and communicating information to the individual via neurostimulation of the one or more afferent sensory nerve fibers using the updated sets of recalibrated neurostimulation signal parameters.

13. A computer program comprising instructions executable by a processor and neurostimulation circuitry of a neurostimulation device to: select a set of test signal parameters; generate, based on the selected set of test signal parameters, at least one neurostimulation test signal configured to elicit a bioelectric response in one or more afferent sensory nerve fibers; apply the generated neurostimulation test signal to the afferent sensory nerve fibers via a neurostimulation interface operably connected to or integrated with the neurostimulation device; sense via the neurostimulation interface, one or more bioelectric responses of the one or more stimulated afferent sensory nerve fibers; determine, based on the sensed bioelectric responses, whether an excitation behavior of the stimulated afferent sensory nerve fibers with respect to the neurostimulation interface has changed; when the excitation behavior has changed, determine, based on the sensed bioelectric responses, a set of recalibrated neurostimulation signal parameters; and operate the neurostimulation device, using the recalibrated neurostimulation signal parameters, to communicate information to the individual via neurostimulation of the one or more afferent sensory nerve fibers.

14. The computer program of claim 13, wherein the instructions are further executable to cause the neurostimulation device to: generate, based on the determined set of recalibrated signal parameters, a communication neurostimulation signal, configured to elicit an artificial sensation in a sensory cortex area via stimulating the one or more afferent sensory nerve fiber terminating in the specific sensory cortex area, wherein the artificial sensation is associated with a block of information to be communicated by the neurostimulation device.

15. The computer program of claim 13 or 14, wherein in determining the set of recalibrated neurostimulation signal parameters, the instructions are executable to cause the neurostimulation device to: compare the sensed bioelectric responses to a set of reference bioelectric responses stored in a memory module of the neurostimulation device or obtained via a communication interface of the neurostimulation device.

16. The computer program of claim 15, wherein the set of reference bioelectric responses is associated with a set of artificial sensations that can be elicited by the neurostimulation device via the neurostimulation interface in a sensory cortex area of the individual and that are associated with one or more blocks of information that can be communicated via the neurostimulation device to the individual.

17. A computer-brain-interface, CBI, device, comprising: a neurostimulation interface comprising one or more stimulation and recording channels adapted to elicit and record a bioelectric response of one or more afferent sensory nerve fibers terminating in a sensory cortex area of an individual; and data and signal processing circuitry, wherein the CBI device is configured to: select a set of test signal parameters; generate, based on the selected set of test signal parameters, at least one neurostimulation test signal configured to elicit the bioelectric response in the one or more afferent sensory nerve fibers; apply the generated neurostimulation test signal to the afferent sensory nerve fibers via the neurostimulation interface; and sense, by the neurostimulation interface, one or more bioelectric responses of the one or more stimulated afferent sensory nerve fibers; determine, based on the sensed bioelectric responses, whether an excitation behavior of the stimulated afferent sensory nerve fibers with respect to the neurostimulation interface has changed; when the excitation behavior has changed, determine, based on the sensed bioelectric responses, a set of recalibrated neurostimulation signal parameters; and using the neurostimulation interface, communicate information to the individual using the recalibrated neurostimulation signal parameters via neurostimulation of the one or more afferent sensory nerve fibers.

18. The CBI device of claim 17, further comprising a memory module operably connected to the data and signal processing circuitry, wherein the memory stores one or both of: a first mapping between one or more artificial sensations that can be elicited by the CBI device in one or more sensory cortex areas of the individual and one or more bioelectric responses; and a second mapping between a plurality of sets of neurostimulation signal parameters and a plurality of bioelectric responses of the one or more afferent sensory nerve fibers.

19. The CBI device of claim 17, wherein the CBI device is further configured to: determine the set of reference bioelectric responses based on one or more of: an initial or on-line calibration procedure involving the individual providing subjective feedback on artificial sensations elicited by a set of reference neurostimulation test signals; a plurality of reference calibration measurements performed on a plurality of individuals prior to determining the set of reference bioelectric responses for the individual; and an initial or online calibration procedure involving the individual performing one or more tasks with objectifiable outcomes that are supported by the operation of the CBI device and recording stimulation parameters and corresponding bioelectric responses that optimize performance of the task without recording subjective feedback by the individual.

20. The CBI device of claim 17, wherein a plurality of different neurostimulation test signals is generated and applied to the afferent sensory nerve fibers interleaved with a plurality of sensing periods of corresponding bioelectric responses of the afferent sensory nerve fibers.

21. A closed-loop calibration method for updating a current set of stimulation parameters of a neurostimulation, NS, device or a computer-brain interface, CBI, device, comprising applying, via a neurostimulation interface device operably connected to the NS or CBI device, a burst sequence of stimulation pulses to a plurality of afferent sensory neurons targeting a sensory cortex area involved with decoding information transmitted by the NS or CBI device; wherein the sequence of stimulation pulses is associated with the current set of stimulation parameters; and wherein the sequence of stimulation pulses is configured to elicit a bioelectric response in the plurality of afferent sensory neurons; recording, via the neurostimulation interface device, the elicited bioelectric response of the stimulated afferent sensory neurons; deriving, based at least in part on the recorded bioelectric response, a neural excitability profile characterizing a non-linear, dynamic excitation behavior of the plurality of afferent sensory neurons corresponding to the applied sequence of stimulation pulses; and adjusting, based on the derived excitability profile at least one stimulation parameter of the current set of stimulation parameters to obtain an updated set of stimulation parameters.

22. The method of claim 21, wherein the burst sequence of stimulation pulses is part of a neurostimulation signal or signal sequence, applied by the NS or CBI device to elicit an artificial sensoiy perception in a sensory cortex area receiving input signals from at least a subset of the plurality of afferent sensory neurons; and / or wherein the burst sequence of stimulation pulses comprises a burst sequence of essentially identical and / or phasic stimulation pulses; and / or wherein an intra-burst pulse frequency is at least 50 Hz, preferably at least too Hz, more preferably at least 200 Hz and even more preferably at least 250 Hz.

23. The method of claim 21 or 22, wherein recording the elicited bioelectric response comprises: recording the bioelectric response while the burst sequence of stimulation pulses is being applied, preferably after each stimulation pulse or continuously during the sequence; and wherein deriving the excitability profile is based at least in part on analyzing intra-burst variations of the recorded bioelectric response.

24. The method of any of the preceding claims 21 to 23, wherein a sampling rate of the recording is equal or larger than the inverse of a time duration between two stimulation pulses of the burst sequence of stimulation pulses, preferably at least twice as large, more preferably at least 10 times as large and even more preferably at least too times as large.

25. The method of any of the preceding claims, wherein at least two consecutive burst sequences are applied, and wherein deriving the excitability profile is based at least in part on analyzing inter-burst variations of the recorded bioelectric response; and optionally, wherein stimulation parameters, preferably the pulse frequency, are varied among the at least two consecutive burst sequences.

26. The method of any of the preceding claims, wherein at least two burst sequences are applied; and wherein deriving the excitability profile comprises analyzing the bioelectric response corresponding to each stimulation pulse within a burst sequence and the bioelectric response corresponding to each burst sequence; and / or wherein deriving the excitability profile comprises analyzing variations among the recorded bioelectric responses within one burst sequence and / or among consecutive burst sequences.

27. The method of any of the preceding claims 21 to 26, wherein deriving the excitation profile comprises extracting temporal variations or dynamics of recording signal parameters or derived metrics from a plurality of subsequent recordings of the elicited bioelectric response and / or continuous recordings; and / or wherein the method further comprises classifying the excitation behavior of a subset of the stimulated afferent sensory neurons using a closed set of discrete categories and based at least in part on the derived excitability profile.

28. The method of claim 27, wherein classification is based at least in part on analyzing a temporal variation or dynamic of the excitability profile within one burst sequence of stimulation pulses and / or among consecutive burst sequences of stimulation pulses.

29. The method of any of the preceding claims 21 to 28, wherein deriving the excitability profile is based at least in part on correlating the recorded bioelectric response with predictions of a non-linear mathematical model of neuronal excitability, comprising model parameters that vary slowly in time to capture physiologic adaptation mechanisms of the stimulated afferent sensory neurons.

30. The method of any of the preceding claims 21 to 29, wherein the elicited bioelectric response comprises one or more compound action potentials, CAPs, and wherein deriving the excitability profile comprises determining one or more of: an Ni / P2 amplitude; a number of detectable peaks or troughs; a measure of synchrony among the bioelectric responses recorded for the burst sequence or among subsequent burst sequences of stimulation pulses, a delay between a stimulation pulse and the corresponding CAP.

31. The method of any of the preceding claims 21 to 30, wherein adjusting the at least one stimulation parameter comprises comparing the derived excitability profile with a reference excitability profile, wherein optionally, the reference excitability profile includes one or more of the following information: an amplitude of a reference bioelectric response, intra-burst variations among bioelectric responses corresponding to single stimulation pulses within a burst sequence, intra-burst variations of the bioelectric response corresponding to the first and the last stimulation pulse within a burst stimulation sequence and inter-burst variations of the bioelectric response.

32. The method of claim 31, wherein the reference excitability profile corresponds to a specific artificial sensory perception corresponding to a set of reference stimulation parameters associated with the reference excitability profile.

33· The method of any of the preceding claims 21 to 32, further comprising applying a neurostimulation signal or signal sequence to at least a subset of the afferent sensoiy neurons using the updated stimulation parameters wherein the neurostimulation signal or signal sequence is configured to elicit an artificial sensory perception in a sensory cortex area receiving afferent sensoiy input from the stimulated subset of afferent sensory neurons.

34. A computer program comprising instructions for carrying out the method of any of the preceding claims, when being executed by data and signal processing circuitry of a neurostimulation or computer brain interface device or system.

35. A neurostimulation, NS, or computer brain interface, CBI, device or system comprising data and signal processing circuitiy for carrying out the method of any of the preceding claims 21 to 34 when carrying out the instructions of the computer program of claim 34.