Systems and methods of training and using a neural decoder

Abstract

A method includes implanting a neural probe configured to record and transmit bioelectrical signals into the brain of a subject. One or more external stimuli having an expected response are provided to the subject. In response to the one or more external stimuli, a recorded bioelectrical signal from the neural probe is received at a computing device in communication with the neural probe. The bioelectrical signal is mapped, at the computing device, to the expected response. The steps of providing, receiving, and mapping are repeated a plurality of times to generate a training dataset of mapped bioelectrical signals. The method further includes, at the computing device, training a neural decoder algorithm based on the training dataset of mapped bioelectrical signals and without external subject feedback.

Description

Attorney Docket No.: AXF-004PC / 139523-5004-WQTITLESYSTEMS AND METHODS OF TRAINING AND USING A NEURAL DECODERCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to and the benefit of U.S. Provisional Patent Application No. 63 / 712,837 filed October 28, 2024, and entitled “Systems and Methods of Training and Using a Neural Decoder,” the contents of which are incorporated by reference herein in their entirety.TECHNICAL FIELD

[0002] The present disclosure generally relates to systems and methods for neural decoding training and uses thereof and, in some embodiments, to systems and methods of training a neural decoder for use with patients having a disorder of consciousness.SUMMARY

[0003] In one embodiment there is a method of training a neural decoder, the method comprising implanting a neural probe into the brain of a subject, the neural probe configured to record and transmit bioelectrical signals, providing one or more external stimuli to the subject, the one or more external stimuli having an expected response, receiving, at a computing device in communication with the neural probe, and in response to the one or more external stimuli, a recorded bioelectrical signal from the neural probe, mapping, at the computing device, the bioelectrical signal to the expected response, repeating the providing, receiving, and mapping steps a plurality of times to generate a training dataset of mapped bioelectrical signals, and at the computing device, training a neural decoder algorithm based on the training dataset of mapped bioelectrical signals and without external subject feedback.

[0004] In another embodiment there is a method of training a neural decoder, the method comprising implanting a neural probe into the brain of a subject, the neural probe configured to record and transmit bioelectrical signals, recording a brain state of the subject, receiving, at a computing device in communication with the neural probe, a recorded bioelectrical signal from the neural probe while the brain state of the subject is being recorded, mapping, at the computing device, the bioelectrical signal to the recorded brain state of the subject, repeating the recording, receiving, and mapping steps a plurality of times to generate a training dataset of mapped bioelectrical signals, and at the computing device, training a neural decoder algorithm based on the training dataset of mapped bioelectrical signals and without external subject feedback. In some embodiments, the method further1DBl / 162944529.6Attorney Docket No.: AXF-004PC / 139523-5004-WQ comprises providing one or more external stimuli to the subject. In some embodiments, the one or more external stimuli have an expected response.

[0005] In some embodiments, the one or more external stimuli comprise an auditory stimulus, an auditory binary question, a non-visual stimulus, a tactile stimulus, a visual stimulus, an olfactory stimulus, a taste stimulus, a proprioceptive stimulus, or a nociceptive stimulus, or any combinations thereof. In some embodiments, the one or more external stimuli are provided to the subject by one or more of the computing device, an artificial intelligence, and a human being, or any combinations thereof. In some embodiments, the one or more external stimuli are provided to the subject in a predefined order and / or at a predefined timing.

[0006] In some embodiments, two or more external stimuli are provided to the subject concurrently.

[0007] In some embodiments, the subject is in a state of cognitive motor dissociation, a vegetative state, a minimally conscious state, a coma, or any combinations thereof.

[0008] In some embodiments, the subject has a disorder of consciousness (DOC) and / or locked- in syndrome.

[0009] In some embodiments, the neural probe comprises a fluorinated elastomer, In some embodiments, the neural probe comprises a perfluorinated elastomer. In some embodiments, the neural probe is implanted through a burr hole with a diameter of less than 2.4 mm.

[0010] In some embodiments, the neural probe comprises a shank configured to be inserted into the brain of the subject. In some embodiments, the shank comprises a plurality of electrodes. In some embodiments, the shank has a number of electrodes per cross-sectional area of the shank that is about 10'7electrodes / micron2to about 102electrodes / micron2. In some embodiments, the electrodes are grouped substantially in two or more regions along the shank. In some embodiments, the grouped electrodes are configured to record the bioelectrical signal from two or more brain regions of the subject. In some embodiments, the two or more brain regions are selected from the cerebral cortex, thalamus, cerebellum, hypothalamus, hippocampus, pituitary gland, brain stem, spinal cord, ventricles, putamen, midbrain, pons, medulla, striatum, amygdala, basal ganglia, pineal gland, frontal lobe, temporal lobe, occipital lobe, parietal lobe, motor strip, sensory strip, Broca’s area, Wernicke’s area, hippocampal formation, hippocampus proper, cingulate gyrus, limbic system, somatosensory cortex, motor cortex, sensory cortex, auditory cortex, prefrontal cortex, entorhinal cortex, visual cortex, perirhinal cortex, insular cortex, primary motor area, supplementary motor area, primary auditory area, auditory association area, primary visual area, and visual association area.DBl / 162944529.6 2Attorney Docket No.: AXF-004PC / 139523-5004-WQ

[0011] In some embodiments, the two or more brain regions comprise a cortex region and the thalamus of the subject.

[0012] In some embodiments, the electrodes are grouped substantially in a first region and a second region along the shank such that, after being inserted into the brain, the electrodes in the first region primarily record the bioelectrical signal from a region of the cerebral cortex of the subject and the electrodes in the second region primarily record the bioelectrical signal from an inner region of the subject’s brain. In some embodiments, the inner region is selected from the thalamus, the hypothalamus, the ventricles, and the cerebellum. In some embodiments, the first region and the second region are about 1 mm to about 120 mm apart along the shank.

[0013] In some embodiments, the step of mapping comprises comparing the bioelectrical signals recorded by the electrodes grouped substantially in two or more regions along the shank. In some embodiments, the step of comparing comprises calculating the delay timings, information flow, and / or cross-complexity of the bioelectrical signals recorded by the electrodes grouped substantially in two or more regions along the shank.

[0014] In some embodiments, the brain state of the subject is an internal event within the brain of the subject. In some embodiments, the internal event within the brain of the subject is selected from a seizure, a sub-clinical seizure, a pre-seizure activity, a cortical spreading depression, a P300 wave, a somatosensory evoked potential, and other sensory evoked potential.

[0015] In some embodiments, the bioelectrical signal comprises single-unit and / or multi-unit neural spikes and / or local field potentials.

[0016] In some embodiments, the training dataset further comprises location information of the neural probe, images of the brain of the subject, video and / or audio recordings of the subject, and / or physiological signals of the subject. In some embodiments, the location information of the neural probe is determined based on characteristics of the bioelectrical signal and / or by brain imaging. In some embodiments, the characteristics of the bioelectrical signal are selected from current source density, bursting patterns, neural spike waveforms, and local field potential waveforms.

[0017] In some embodiments, the subject is a human, a rodent, a pig, a sheep, or a non-human primate.

[0018] In another embodiment there is a system for implementing a method of training a neural decoder, the system comprising a neural probe configured to be implanted into the brain of a subject, the neural probe configured to record and transmit bioelectrical signals; and a computing device in communication with the neural probe, the computing device configured to, in response to one or more external stimuli provided to the subject, receive a recorded bioelectrical signal from the neural probe, DBl / 162944529.6 3Attorney Docket No.: AXF-004PC / 139523-5004-WQ map the bioelectrical signal to an expected response of the external stimulus, generate a training dataset of mapped bioelectrical signals by repeatedly receiving and mapping bioelectrical signals, and based on the training dataset of mapped bioelectrical signals, train a neural decoder algorithm without external subject feedback.

[0019] In another embodiment there is a system for implementing a method of training a neural decoder, the system comprising a neural probe configured to be implanted into the brain of a subject, the neural probe configured to record and transmit bioelectrical signals; and a computing device in communication with the neural probe, the computing device configured to receive a recorded bioelectrical signal from the neural probe, map the bioelectrical signal to a brain state of the subject, generate a training dataset of mapped bioelectrical signals by repeatedly receiving and mapping bioelectrical signals, and based on the training dataset of mapped bioelectrical signals, train a neural decoder algorithm without external subject feedback.

[0020] In another embodiment there is a method of interpreting neural activity of a subject, the method comprising receiving, at a computing device a recorded bioelectrical signal from a neural probe that is implanted into the brain of the subject, decoding, at the computing device via a neural decoder, the bioelectrical signal into an intent and / or brain state of the subject, and transmitting the decoded intent and / or brain state to a communication effector in communication with the computing device.

[0021] In another embodiment there is a use of a method of interpreting neural activity of a subject for the diagnosis, prognosis, and / or treatment of a patient.

[0022] In another embodiment there is a system for implementing a method of interpreting neural activity of a subject, the system comprising a neural probe configured to be implanted into the brain of a subject, the neural probe configured to record and transmit bioelectrical signals; a computing device in communication with the neural probe, the computing device configured to receive a recorded bioelectrical signal from the neural probe and decode, via a trained neural decoder, the bioelectrical signal into an intent and / or brain state of the subject; and a communication effector in communication with the computing device, the communication effector configured to output the decoded intent and / or brain state of the subject.

[0023] In another embodiment, there is a use of a system for implementing a method of interpreting neural activity of a subject for the diagnosis, prognosis, and / or treatment of a patient.BRIEF DESCRIPTION OF THE DRAWINGSDBl / 162944529.6 4Attorney Docket No.: AXF-004PC / 139523-5004-WQ

[0024] The following detailed description of embodiments of the systems and methods of training and using a neural decoder will be better understood when read in conjunction with the appended drawings of exemplary embodiments. It should be understood, however, that the invention is not limited to the precise arrangements and instrumentalities shown.

[0025] In the drawings:

[0026] Fig. l is a diagram illustrating a neural decoding system in accordance with an exemplary embodiment of the present disclosure;

[0027] Fig. 2 is a diagram illustrating a neural probe of the system of Fig. 1 implanted in the brain of a patient;

[0028] Fig. 3 is a diagram illustrating a neural probe of the system of Fig. 1 with electrodes arranged substantially in two groups;

[0029] Fig. 4 is a diagram illustrating two neural probes of the system of Fig. 1 simultaneously implanted in the brain of a patient;

[0030] Fig. 5 is a diagram illustrating a neural probe implanted in the brain of a patient in accordance with another embodiment of the present disclosure;

[0031] Fig. 6 is a flowchart illustrating a method of training a neural decoder in accordance with an exemplary embodiment of the present disclosure;

[0032] Fig. 7 is a flowchart illustrating a method of training a neural decoder in accordance with another exemplary embodiment of the present disclosure;

[0033] Fig. 8 is a flowchart illustrating a method of using a neural decoder in accordance with an exemplary embodiment of the present disclosure;

[0034] Fig. 9 is a flowchart illustrating a method of using a neural decoder in accordance with another exemplary embodiment of the present disclosure;

[0035] Fig. 10 is a flowchart illustrating a method of monitoring a patient with a neural decoder in accordance with an exemplary embodiment of the present disclosure;

[0036] Fig. 11 is a flowchart illustrating a method of monitoring a patient with a neural decoder in accordance with another exemplary embodiment of the present disclosure;

[0037] Fig. 12 is a flowchart illustrating a method of updating and / or retraining a neural decoder in accordance with an exemplary embodiment of the present disclosure;

[0038] Fig. 13 is a diagram illustrating training a neural decoder with a multi-modal dataset in accordance with an exemplary embodiment of the present disclosure;DBl / 162944529.6 5Attorney Docket No.: AXF-004PC / 139523-5004-WQ

[0039] Figs. 14A and 14B illustrate methods and datasets for training a neural decoder with rats in accordance with exemplary embodiments of the present disclosure and show the results of using the trained neural decoder to decode the states of rats; and

[0040] Figs. 15 A to 15D illustrate the methods for and results of using a neural decoder to decode the states of human subjects in accordance with exemplary embodiments of the present disclosure.DETAILED DESCRIPTION

[0041] Patients with brain injuries, who have a disorder of consciousness (DOC), a locked-in syndrome, and / or a cognitive motor dissociation (CMD), and / or who are in a coma, a medically induced coma, a vegetative state, and / or a minimally conscious state may have brain activity which is different from their externally observable status by, e.g., a trained expert. For instance, they may have observable responsiveness and / or reflexes but have no higher-level brain activity associated with the processing of information; while on the other hand, they may have no observable responsiveness and / or reflexes but are cognitively capable of recognizing sensory inputs. In other words, the observable external aspects of a patient may not accurately reflect their internal brain states. In such cases, a neural decoding system can be useful in interpreting neural signals to provide more accurate information about their true brain states to guide their care and reassure clinicians and loved ones. Typical information provided by neural decoding systems includes, e.g., physiological metrics about the patient’s brain state, which may be referred to as Neural Correlates of Consciousness, for comparison of the state of the patient to their disease cohort. Additionally, neural decoding systems utilizing machine learning models can classify or map a patient’s neural data into brain states based on the training with that patient or a group of patients.

[0042] Patients with a DOC who are unresponsive externally to commands may be in a state of CMD. While a patient is in a state of CMD, they may be able to understand auditory paradigms and communicate their corresponding choices via brain activity. However, while the patient is in a state of CMD they are generally unable to generate functional motor output. For example, patients in a state of CMD can be completely unresponsive or only able to achieve minor movements such as the twitch of an eye, finger, and / or toe. Accordingly, a neural decoding system may be useful in interpreting neural signals from the patients and establishing a means of communication therewith. For example, an implantable brain-computer interface (iBCI) including a neural decoder may be implanted in a patient to interpret or translate neural signals from the patient. However, neural decoders require training due to the unique neural signals of patients and conventional systems and methods for doing so require a two-way training. For example, conventional system and methods for training a neural DBl / 162944529.6 6Attorney Docket No.: AXF-004PC / 139523-5004-WQ decoder require a patient to verbally and / or physically respond to an external stimulus in order to map neural signals and train a neural decoder. However, for patients in a state of CMD and / or those who are substantially incapable of expressing physical and / or verbal communication, conventional systems and methods for training a neural decoder are insufficient. Accordingly, aspects of the present disclosure provide systems and methods of training a neural decoder for use in patients who have a DOC, including, without limitation, those who are in a state of CMD and / or substantially incapable of expressing physical, visual and / or verbal communication. The term CMD as used herein may be used interchangeably with the term Convert Consciousness (CC).

[0043] The term “map” and “mapping” as used herein in relation to a neural decoder and bioelectrical signals such as neural signals may generally include the action of the neural decoder in converting bioelectrical signals to intent and / or state (e.g., brain state) and is not limited to a 1 : 1 mapping or any other such constraint on the underlying process and action of the neural decoder.

[0044] Referring to the drawings in detail, wherein like reference numerals indicate like elements throughout, there is shown in Fig. 1 a neural decoding system, generally designated 100, in accordance with an exemplary embodiment of the present invention. In some embodiments, the neural decoding system 100 is configured to train a neural decoder (e.g., a neural decoding model) for use in a subject, including, without limitation, rodents (e.g., mice and rats), large mammals (e.g., pigs and sheep), nonhuman primates, and humans. In some embodiments, the neural decoding system 100 is configured to train a neural decoder (e.g., a neural decoding model) for use in patients that are substantially incapable of expressing physical and / or verbal communication (e.g., patients who have a DOC and / or patients who are in a state of CMD). In some embodiments, the patients are in a coma, a medically induced coma, a vegetative state, and / or a minimally conscious state. In some embodiments, the neural decoding system 100 is configured to train a neural decoder without external feedback from the subject (e.g., the patient). A neural decoder as discussed herein may be a computational model or system that interprets or translates bioelectrical signals (e.g., neural signals) into outputs that may communicate the intent (e.g., movements, speech, text, or other forms of communication) and / or state (e.g., brain state) of the subject, e.g., the patient. In some embodiments, a DOC is diagnosed and / or classified using a clinical scale, e.g., the JFK Coma Recovery Scale- Revised (CRS-R), See, Arch Phys Med Rehabil. 2004;85(12):2020-2029.

[0045] In one embodiment, the system 100 includes one or more computers having one or more processors and memory (e.g., one or more nonvolatile storage devices). In some embodiments, memory or computer readable storage medium of memory stores programs, modules and data structures, or a subset thereof for a processor to control and run the various systems and methods DBl / 162944529.6 "7Attorney Docket No.: AXF-004PC / 139523-5004-WQ disclosed herein. In one embodiment, a non-transitory computer readable storage medium having stored thereon computer-executable instructions which, when executed by a processor, perform one or more of the methods disclosed herein.

[0046] Referring to Figs. 1-2, the system 100 may include a computing device 102, a neural probe 104, and / or a communication effector 106. The computing device 102 may be communicatively coupled to the neural probe 104 and / or the communication effector 106. For example, the computing device 102 may be coupled to the neural probe 104 and / or the communication effector 106 via one or more of a local area network (LAN) and a wide area network (WAN). The computing device 102 may be configured to train and / or execute a neural decoder based on bioelectrical signals recorded by the neural probe 104. The computing device 102 may include a processor and a memory (e.g., a non- transitory computer readable storage medium). The processor may be configured to train and execute a neural decoder model stored in the memory based on bioelectrical signals from the neural probe 104. The computing device 102 may be a computing device suitable for training and executing a neural decoding model.

[0047] The communication effector 106 may be communicatively coupled to the computing device 102 and configured to facilitate the exchange of data to and / or from the computing device 102. In some embodiments, the communication effector 106 is peripheral device coupled to the computing device 102 and configured to receive inputs to the computing device 102 and / or outputs therefrom. In some embodiments, the communication effector 106 is a patient monitor. For example, the communication effector 106 may include one or more sensors, display devices (e.g., monitors), speakers, and / or microphones. The communication effector 106 may be configured to communicate the output of the neural decoder in an audible, visual, and / or physical form. The communication effector 106 may include a voice and / or image generator configured to receive data output from the neural decoder and generate an audible and / or visual output. For example, a neural decoding model executed at the computing device 102 may receive neural signals from the probe 104 and translate those neural signals into text data representing speech of the patient, which the communication effector 106 outputs as visual text on a display device and / or audible output at a speaker. In some embodiments, the communication effector 106 is configured to generate an audible stimulus intended for a patient 10 as discussed in more detail below.

[0048] The neural probe 104 may be implanted into the brain of a subject, e.g., the patient 10, for which neural decoding is desired. The neural probe 104 may be configured to record and transmit neural signals from the brain of the patient 10 to the computing device 102. For example, neural probe 104 may be implanted into the brain of the patient 10 such that it may record bioelectrical signals DBl / 162944529.6 8Attorney Docket No.: AXF-004PC / 139523-5004-WQ(e.g., neural signals) generated by the patient 10 and transmit those signals to the computing device 102. In some embodiments, the neural probe 104 is communicatively coupled to the computing device 102 via an input / output cable 108 (I / O cable 108). Examples of neural probe materials and methods of manufacture thereof suitable for neural decoding applications are described in published international application WO 2022 / 192319, entitled, Fluorinated Elastomers for Brain Probes and Other Applications; published international application WO 2024 / 182406, entitled, Methods to Connect Free-Standing Soft Neural Probes to Active Electronic Devices with High Channel Count Interface; and Non-Patent Literature entitled “Flexible, Biocompatible Photoresists for Fabricating Electronic, in vivo, and Microfluidic Devices”, Proc. SPIE PC 12956, Novel Patterning Technologies 2024, PC129560X (22 April 2024); <https: / / doi.org / 10.1117 / 12.3010475>, the disclosures of each of which are incorporated by reference herein in their entirety.

[0049] In some embodiments, the neural probe 104 is configured to be implanted into the brain of the subject, e.g., the patient 10, via a minimally invasive surgery. For example, the neural probe 104 may be implanted through an opening in the patient’s skull having a maximum dimension of less than about 2.4 mm that extends through the skull of the patient 10. For example, the neural probe 104 may be implanted through a burr hole having a diameter of about 2.4 mm or an incision having a maximum dimension of about 2.4 mm. In some embodiments, the opening in the patient’s skull e.g., a burr hole) may have a maximum dimension of greater than 2.4 mm. In some embodiments, the burr hole may have a maximum dimension of less than or equal to about 5 mm, about 10 mm, about 15 mm or about 20 mm. In some embodiments the insertion hole for implanting the neural probe 104 into the skull of a patient 10 may be made by a craniotomy and be up to about 30 mm, about 50 mm or about 80 mm across. In some embodiments, and as illustrated in Fig. 2, the neural probe 104 includes an anchor bolt 110 for mounting the neural probe to the patient 10. For example, the anchor bolt 110 may extend through a burr hole formed in the patient’s skull and anchor the neural probe 104 to the patient 10. In some embodiments, the anchor bolt is transcutaneous. In some embodiments, the anchor bolt 110 extends from the exterior of the patient’s skin through the skull of the patient 10. In some embodiments, the anchor bolt 110 may enable the neural probe 104 to be temporarily implanted in the patient 10. For example, the anchor bolt 110 may enable the shank 112 to be explanted without requiring any skin-opening surgery. In some embodiments, the neural probe 104 or shank 112 is implanted through an existing cranial access without the need of any additional burr hole or incision. The cranial access may be created during any neurosurgical procedures, including, without limitation, decompressive craniotomy, cerebral drain and / or shunt procedure, brain tumor removal and / or embolization, neuroendoscopy, and / or brain biopsy.DBl / 162944529.6 9Attorney Docket No.: AXF-004PC / 139523-5004-WQ

[0050] The neural probe 104 may include a shank 112 configured to be inserted or implanted into the brain of a subject, e.g., a patient. The shank 112 may be inserted into one or more regions of the brain, depending on, e.g., the needs and / or status of a particular patient. In some embodiments, the one or more brain regions are selected from the cerebral cortex, cerebellum, thalamus, hypothalamus, hippocampus, pituitary gland, brain stem, spinal cord, ventricles, putamen, midbrain, pons, medulla, striatum, amygdala, basal ganglia, pineal gland, frontal lobe, temporal lobe, occipital lobe, parietal lobe, motor strip, sensory strip, Broca’s area, Wernicke’s area, hippocampal formation, hippocampus proper, cingulate gyrus, limbic system, somatosensory cortex, motor cortex, sensory cortex, auditory cortex, prefrontal cortex, entorhinal cortex, visual cortex, perirhinal cortex, insular cortex, primary motor area, supplementary motor area, primary auditory area, auditory association area, primary visual area, and visual association area. The shank 112 may be inserted or implanted in any manners and / or via any trajectories to reach the target regions of the brain. In some embodiments, the shank 112 is inserted or implanted with a neuro-navigation system (e.g., a computer-assisted surgical guidance system) and / or a surgical robot to reach the target regions of the brain.

[0051] In some embodiments, the shank 112 has a length from about 1 mm to about 200 mm and / or a width or diameter from about 0.001 mm to about 100 mm. In some embodiments, the shank 112 has a length that is about 1 mm, about 2 mm, about 4 mm, about 8 mm, about 10 mm, about 15 mm, about 20 mm, about 30 mm, about 40 mm, about 50 mm or more. In some embodiments, the shank 112 has a length that is about 200 mm, about 180 mm, about 160 mm, about 140 mm, about 120 mm, or about 100 mm or less. In some embodiments, the shank 112 has a width or diameter that is about 0.001 mm, about 0.01 mm, about 0.1 mm, about 0.5 mm, about 1 mm, about 2 mm, about 4 mm, about 8 mm, about 16 mm, about 32 mm, about 64 mm or more. In some embodiments, the shank 112 has a width or diameter that is about 100 mm, about 80 mm, about 60 mm, about 40 mm, about 20 mm, about 10 mm, about 5 mm or less. Combinations of these ranges are possible. For instance, in some embodiments, the shank 112 has a length that is about 10 mm to about 100 mm and a width or diameter that is about 0.1 mm to about 10 mm.

[0052] The shank 112 may house one or a plurality of electrodes that are configured to detect the bioelectrical signals generated from neurons in the brain of the subject, e.g., the patient 10. In some embodiments, the shank 112 includes about 2 electrodes, about 4 electrodes, about 8 electrodes, about 16 electrodes, about 32 electrodes, about 64 electrodes, about 128 electrodes, about 256 electrodes, about 512 electrodes, about 1024 electrodes or more than 1024 electrodes embedded therein. The shank 112 may have a number of electrodes per cross-sectional area of the shank 112 that is about 10’ 7 electrodes / micron2to about 102electrodes / micron2. In some embodiments, the shank 112 has a DB1 / 162944529.6 10Attorney Docket No.: AXF-004PC / 139523-5004-WQ number of electrodes per cross-sectional area of the shank 112 that is about 10'6electrodes / micron2, about 10'5electrodes / micron2, about 10'4electrodes / micron2, about 10'3electrodes / micron2, about 10" 2 electrodes / micron2, about 10'1electrodes / micron2, about 10° electrodes / micron2, or about 101electrodes / micron2. The electrodes may be arranged in any configurations. In some embodiments, the electrodes are arranged in a single column or multiple columns. In some embodiments, the electrodes are arranged in grids. In some embodiments, the electrodes are arranged in groups of 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, or more electrodes spaced out along the length and / or width of the shank 112. In some embodiments, the electrodes are arranged without following any regular patterns. In some embodiments, the electrodes are arranged in configurations to optimize the yield of both single unit recordings and spatial sampling of local field potentials. In some embodiments, the electrodes are grouped substantially in two or more regions along the length of the shank 112. In some embodiments, the electrodes are arranged substantially in 2 or more groups along the length of the shank 112 to sample bioelectrical signals from one or more regions at different depths of the brain as disclosed herein. In some embodiments, the electrodes are arranged substantially in about 2 groups, about 4 groups, about 6 groups, about 8 groups, or about 10 groups or more along the length of the shank 112.

[0053] Referring to Fig. 3, the electrodes housed on the shank 112 may be arranged substantially in 2 groups: the first group of electrodes 113a and the second group of electrodes 113b. In some embodiments, the electrodes are arranged substantially in 2 groups along the length of the shank 112 such that, after being inserted or implanted into the brain, the first group of electrodes 113a primarily samples bioelectrical signals from a target region of the cerebral cortex of the patient’s brain and the second group of electrodes 113b primarily samples bioelectrical signals from an inner target region of the patient’s brain, including, without limitation, the thalamus, the hypothalamus, the ventricles, and / or the cerebellum. The target region of the cerebral cortex may be any regions disclosed herein, including, without limitation, the somatosensory cortex, the motor cortex, sensory cortex, the auditory cortex, the prefrontal cortex, the entorhinal cortex, the visual cortex, the perirhinal cortex, and the insular cortex. In some embodiments, the first group of electrodes 113a primarily samples bioelectrical signals from the parietal and / or prefrontal cortices of the patient’s brain and the second group of electrodes 113b primarily samples bioelectrical signals from the thalamus. The locations of and distances between the first group of electrodes 113a and the second group of electrodes 113b may depend on, e.g, the way the neural probe 104 and / or shank 112 is implanted, the size of the brain, and / or the locations of the target regions. In some embodiments, the first group of electrodes 113a substantially locates around about 1 mm to about 20 mm from the base of the shank 112 (e.g., the end of the shank 112 mounted to the anchor bolt 110), and the second group of electrodes 113bDBl / 162944529.6 1 1Attorney Docket No.: AXF-004PC / 139523-5004-WQ substantially locates near the tip of the shank 112. In some embodiments, the first group of electrodes 113a substantially locates around about 1 mm, about 2 mm, about 4 mm, about 6 mm, about 8 mm, about 10 mm, about 15 mm, or about 20 mm from the base of the shank 112. In some embodiments, the distance along the length of the shank 112 between the first group of electrodes 113a and the second group of electrodes 113b is from about 1 mm to about 120 mm. In some embodiments, the distance along the length of the shank 112 between the first group of electrodes 113a and the second group of electrodes 113b is about 1 mm, about 5 mm, about 10 mm, about 20 mm, about 40 mm, about 60 mm, about 80 mm, about 100 mm, or about 120 mm.

[0054] The shank 112 may be mounted to the anchor bolt 110 and extend outwardly therefrom into the brain of the subject, e.g., the patient 10. In some embodiments, the neural probe 104 includes a connector 114 for communicatively coupling the shank 112 to the I / O cable 108. The connector 114 may be coupled to the anchor bolt 110 opposite the shank 112. In some embodiments, the connector 114 is in communication with the shank 112 such that bioelectrical signals detected by the electrodes embedded in the shank 112 may be transmitted to the computing device 102 via the I / O cable 108. In some embodiments, the shank 112 and connector 114 are fixedly coupled to the anchor bolt 110. In some embodiments, the connector 114 includes one or more embedded electronics such as, but not limited to, a transducer, a low-power mixed-signal electronic circuit and / or an application-specific integrated circuit (ASIC). In some embodiments, the embedded electronics of the connector 114 may be configured for amplifying, multiplexing, and / or digitizing neural signals recorded via the shank 112. In some embodiments, the neural probe 104 may include a plurality of shanks 112. For example, the neural probe 104 may include a plurality of shanks 112 coupled to the anchor bolt 110 and / or connector 114. In some embodiments, the neural probe 104 includes about 2 shanks 112, about 4 shanks 112, about 6 shanks 112, about 8 shanks 112, about 10 shanks 112 or more coupled to the anchor bolt 110 and / or connector 114. In some embodiments, the neural probe 104 and the anchor bolt 110 are physically separated from the connector 114. In some embodiments, the neural probe 104 and / or the anchor bolt 110 include one or more wireless communications devices and / or one or more signal processing circuits embedded therein. In some embodiments, the connector 114 includes a wireless communication module in communication with the neural probe 104 and / or the anchor bolt 110. In some embodiments, the wireless communication between the connector 114 and the neural probe 104 and / or the anchor bolt 110 is local. For instance, the wireless communication may be within the local environment of the brain of a patient. As another example, the wireless communication may be within the room or space where the patient is located.DBl / 162944529.6 12Attorney Docket No.: AXF-004PC / 139523-5004-WQ

[0055] In some embodiments, the neural probe 104 is a minimally invasive, minimally damaging, removable neural probe that is generally soft and can be implanted at a depth into the brain of a subject, e.g., a patient, and configured to maintain connection with the same neurons for extended periods of time. In some embodiments, the neural probe 104 is comprised of a generally soft material configured to minimize tissue encapsulation around the probe 104 and that is anti -biofouling and / or anti-sticking. In some embodiments, the probe 104 is comprised of a fluorinated elastomer. For example, the shank 112 may be comprised of one or more layers of fluorinated elastomer. In some embodiments, the fluorinated elastomer has an elastic modulus of less than or equal to about 106Pa. In some embodiments, the fluorinated elastomer has an elastic modulus substantially equal to the elastic modulus of a tissue into which the shank 112 is configured to be embedded (e. ., brain tissue). In some embodiments, the shank 112 includes a layer of conductive material that forms the one or more electrodes, and which is sandwiched between layers of a fluorinated elastomer. In some embodiments, the fluorinated elastomer is a perfluorinated elastomer. In some embodiments, the shank 112 has an overall elastic modulus of less than or equal to about 106Pa. In some embodiments, the shank 112 has an elastic modulus that is substantially equal to the elastic modulus of a tissue into which the shank 112 is configured to be embedded (e.g., brain tissue). In some embodiments, the shank 112 has a flexural rigidity that is about ICT13Nm to about 10'5Nm. In some embodiments, the shank 112 has a flexural rigidity that is about 10'14Nm, about 10'13Nm, about 10'12Nm, about 10’11Nm, about 10’10Nm, about 10'9Nm, about 10'8Nm, about 10'7Nm, or about 10‘6Nm. In some embodiments, the shank 112 is configured to remain in contact with a tissue or organ for at least one week, at least 2 weeks, at least 4 weeks, at least one month, at least 2 months, at least 4 months, at least 6 months, at least a year or more continuously without provoking a substantial immune response, fibrotic response, encapsulation, scar tissue formation, and / or tissue necrosis.

[0056] In some embodiments, the neural probe 104 is configured to provide stimulation to the subject, e.g., the patient 10, via one or more of the electrodes embedded therein. For example, an electrical signal may be transmitted to the shank 112 such that an electrical stimulation is delivered to the brain of the patient 10. In some embodiments, the connector 114 is configured to transmit electrical signals to the electrodes embedded in the shank 112. In some embodiments, the computing device 102 may transmit an electrical signal to the neural probe 104 to cause the shank 112 to deliver an electrical stimulation to the brain of the patient 10.

[0057] Referring to Fig. 4, in some embodiments, two or more neural probes 104 may be implanted in a subject, e.g., a patient. As illustrated in Fig. 4, two neural probes 104a, 104b each being generally the same as neural probe 104 discussed above are shown implanted in the brain of a patient DB1 / 162944529.6 13Attorney Docket No.: AXF-004PC / 139523-5004-WQ10. In some embodiments, the two or more neural probes 104a, 104b may be implanted through the same burr hole and share a connection to the computing device 102. The neural probes 104a, 104b may each be communicatively coupled to computing device 102 in generally the same manner as the neural probe 104 discussed above. In some embodiments, the computing device 102 is configured to receive bioelectrical signals (e.g., neural signals) from each of the neural probes 104a, 104b and train and / or execute a neural decoding model based on the received bioelectrical signals. In some embodiments, each of the neural probes 104a, 104b are implanted into the brain of the patient 10 in generally the same manner as discussed above with regards to neural probe 104. For example, each of the neural probes 104a, 104b may be implanted through respective burr holes having a diameter of less than about 2.4 mm. In some embodiments, there may be more than two neural probes 104 implanted into the patient and in communication with the computing device 102. In some embodiments, there may be between one to about 20 neural probes 104 implanted into the patient 10 simultaneously and in communication with the computing device 102. The one or more neural probes 104a, 104b may be physically separated from the connector 114 and communicate with the connector 114 wirelessly, as described above. In some embodiments, the connector 114 coordinates the communication of signals to and from the one or more neural probes 104a, 104b.

[0058] Referring to Fig. 5, in some embodiments, there may be one or more neural probes 204 chronically or permanently implanted into a subject, e.g., a patient. The neural probe 204 may be generally similar to neural probe 104 except that it is configured to be chronically or permanently implanted in a patient 10 and is configured to facilitate wireless communications with the computing device 102. The neural probe 204 may include a shank 212 generally the same as shank 112 discussed above. Further, there may be multiple shanks 212 similar to shank 112 connected to the same neural probe 204 and anchor bolt 210. The neural probe 204 may include an anchor bolt 210 coupling the shank 212 to the connector 214. The anchor bolt 210 may be similar to the anchor bolt 110 except that it does not extend exteriorly from the skin of the patient 10. For example, the anchor bolt 210 is embedded partially within the patient’s skull and does not extend exteriorly through the patient’s skin. In some embodiments, the anchor bolt 210 does not extend entirely through the patient’s skull.

[0059] The neural probe 204 may include a connector 214 that is generally similar to connector 114 except that it may be configured to be implanted in the subject, e.g., the patient 10. In some embodiments, the connector 214 and / or anchor bolt 210 include at least one power source embedded therein and configured to provide power to the neural probe 204. In some embodiments, the power source is a rechargeable power source and the connector 214 may include corresponding circuitry and / or devices for facilitating charging. For example, the connector 214 may include a charging DB1 / 162944529.6 14Attorney Docket No.: AXF-004PC / 139523-5004-WQ circuitry and an induction charging receiving (e.g., a receiver coil). As another example, the connector 214 may include a charging port accessible at an exterior of the skull of the patient 10 such that a power cable may be connected to the charging port to provide power to the rechargeable power source. In further embodiments, the connector 214 is configured to provide power to the neural probe 204 while receiving power from an external source. For example, the connector 214 may not include a power source and may only provide power to the neural probe 204 while receiving power from an external source (e.g., via a power cable and / or magnetic induction). In other embodiments, the power source is a primary cell battery and the connector 214 and / or anchor bolt 210 may be configured to regulate a voltage of the primary cell battery.

[0060] In some embodiments, the connector 214 includes one or more wireless communication devices and / or one or more signal processing circuits embedded therein. The connector 214 may include a wireless communications module in communication with the computing device 102. In some embodiments, the connector 214 includes one or more of a radio frequency (RF) device, an ultrasound communicator, a near-field communication (NFC) device, or any other suitable wireless communications device such that data may be transmitted between the neural probe 204 and computing device 102. In some embodiments, the connector 214 includes an embedded ASIC and / or a low-power mixed-signal electronic circuit. In some embodiments, the connector 214 may be configured for amplifying, multiplexing, digitizing, and / or encoding, via a wireless communication protocol, neural signals recorded via the shank 212. The connector 214 may be coupled to the shank 212 such that neural signals recorded by the shank 212 may be transmitted to the computing device 102 via the connector 214. In some embodiments, the system 100 may include the neural probe 104 and / or neural probe 204. In some embodiments, the system 100 may include a combination of the neural probe 104 and neural probe 204. In some embodiments, the system 100 includes a plurality of neural probes 204. In some embodiments, the system 100 includes a plurality of each of the neural probes 104 and neural probe 204.

[0061] Referring to Fig. 6, there is illustrated a method of training a neural decoder, generally designated 300, in accordance with an exemplary embodiment of the present disclosure. The method 300 may include the step 302 of implanting a neural probe into the brain of a subject, e.g., a patient. In some embodiments, the patient is substantially incapable of expressing physical and / or verbal communication. In some embodiments, the patient has a DOC and / or is in a state of CMD. In some embodiments, the patient is in a coma, a medically induced coma, a vegetative state, and / or a minimally conscious state. In some embodiments, the neural probe is configured to record and transmit bioelectrical signals. For example, and as illustrated in Fig. 1, the neural probe 104 is DB1 / 162944529.6 15Attorney Docket No.: AXF-004PC / 139523-5004-WQ implanted into the brain of the patient 10. The neural probe 104 is configured to record and transmit bioelectrical signals (e.g., neural signals) from the brain of the patient 10 to the computing device 102 as discussed above. In some embodiments, the neural probe 104 is implanted through a burr hole with a diameter of less than 2.4 mm. In some embodiments, the neural probe 104 is comprised of a fluorinated elastomer. For example, the shank 112 may be comprised of one or more layers of fluorinated elastomer as discussed above. In some embodiments, the fluorinated elastomer is a perfluorinated elastomer. In some embodiments, the step 302 is performed by implanting the neural probe 204 discussed above with regards to Fig. 5.

[0062] The method 300 may include the step 304 of providing an external stimulus to the subject, e.g., the patient 10. In some embodiments, the external stimulus has an expected or known response. An external stimulus having a known or expected response may be provided to the patient 10 while the neural probe(s) 104 are implanted in the patient 10. In some embodiments, the external stimulus is provided to the patient 10 via the communication effector 106 and / or an external agent 12 (e.g., a computer device that may transmit that communication via an appropriate medium (SMS, email, etc ), an artificial intelligence (Al) agent, a healthcare practitioner, a family member of the patient, or any other human agent). In some embodiments, the external stimulus is a non-visual stimulus. In some embodiments, the external stimulus is an auditory stimulus. For example, an auditory stimulus in the form of a question with a known answer (e.g., “Is your name Tom?”, “Is it July?”) or a prompt for an action (e.g., “Move your left hand”, “Wiggle your big toe on your left foot”) may be provided to the patient. In some embodiments, the external stimulus is an auditory binary question. For example, the external stimulus may be an audible question with ‘yes’ and ‘no’ as the possible answers. In some embodiments, the external stimulus includes a statement, such as, for example, the name of the patient 10, the name of a relative of the patient 10, or a passage (e.g., a sentence, paragraph, or musical song) known to the patient 10.

[0063] In some embodiments, the external stimulus may include an audible tone output via the communications effector 106 a predetermined number of times followed by a question of whether the tone was played that predetermined number of times. For example, if the tone is played twice the audible question to the patient may be to move their index finger if the tone was played twice, with the expected response being movement or attempted movement of the patient’s index finger. In some embodiments, the external stimulus may be a tactile stimulus. For example, the external stimulus may include applying pressure to a part of the patient’s body (e.g., hands, feet, legs, chest, arms, or neck). In some embodiments, the external stimulus may be electrical stimulation on the surface of the skin of the patient 10. In some embodiments, the external stimulus may be in the form of ocular stimulus. DB1 / 162944529.6 16Attorney Docket No.: AXF-004PC / 139523-5004-WQFor example, the external stimulus may include one or more flashes of colored light in front of the eyes or eyelids of the patient 10. In some embodiments, the external stimulus may be an olfactory stimulus. For example, the external stimulus includes a smell presented to the nose of the patient. In some embodiments, the external stimulus may be a combination of one or more of the types of external stimuli described above.

[0064] Providing an external stimulus with a known or expected response to the patient may be particularly beneficial in instances where the patient is unable to provide external feedback to the response. For example, the patient may be in a state of CMD, have a DOC, be in a vegetative state, be in a minimally conscious state, and / or have locked-in syndrome. In such states, the patient may not be capable of providing external feedback in the form of movement or an audible answer to the external stimulus. Accordingly, providing an external stimulus with a known or expected response may aid in mapping neurological signals detected in the patient’s brain as discussed in more detail below.

[0065] The method 300 may include the step 306 of receiving, at a computing device in communication with the neural probe, and in response to the external stimulus, a recorded bioelectrical signal from the neural probe. For example, in response to the external stimulus provided to the patient 10, the neural probe(s) 104 implanted in the patient may record a resulting neural signal via electrodes embedded in the probe(s) 104. The neural probe(s) 104 may, in response to recording a neural signal, be configured to transmit the neural signal to the computing device 102. For example, neural signals recorded by the shank 112, 212 of the neural probes 104, 204, respectively, are transmitted to the computing device 102 via the respective connectors 114, 214, as discussed above.

[0066] The method 300 may include the step 308 of mapping, at the computing device, the bioelectrical signal to the expected response. The computing device 102 may be configured to receive a neural signal recorded at step 306 and automatically map that neural signal to the expected response from the external stimulus provided to the patient 10 in step 304. For example, an auditory stimulus provided to a patient 10 named Tom in the form of the prompt ‘Move your left index finger if your name is Tom’ is performed at step 304. Further to this example, the neural probe 104 records one or more neural signals following the auditory stimulus and transmits the neural signal(s) to the computing device 102. In response to receiving the neural signal the computing device 102 automatically maps the neural signal with intended movement of the patient’s left index finger because the expected response to the auditory stimulus is movement of the left index finger. In some embodiments, the computing device 102 automatically maps the neural signal with the intended external response from the patient 10 without the patient exhibiting said response. For example, in step 308 the computingDB1 / 162944529.6 17Attorney Docket No.: AXF-004PC / 139523-5004-WQ device 102 maps the neural signal without any visible motor response from the patient 10. In some embodiments, the computing device 102 is configured to assign the mapped bioelectrical signal to a training dataset of mapped bioelectrical signals. In some embodiments, the step 308 of mapping comprises comparing the bioelectrical signals recorded by the electrodes grouped substantially in two or more regions along the shank 112, 212. In some embodiments, the step 308 of mapping comprises calculating the delay timings, information flow, and / or cross-complexity of the bioelectrical signals recorded by the electrodes grouped substantially in two or more regions along the shank 112, 212. In some embodiments, the step 308 of mapping comprises comparing the bioelectrical signals recorded by the electrodes on two or more shanks 112, 212 and / or two or more probes 104a, 104b. In some embodiments, the step 308 of mapping comprises calculating the delay timings, information flow, and / or cross-complexity of the bioelectrical signals recorded by the electrodes on two or more shanks 112, 212 and / or two or more probes 104a, 104b.

[0067] In some embodiments, the method 300 includes verifying the accuracy of the mapped neural signal. In instances where a subject, e.g., a patient, displays signs of behavioral responses, the accuracy of the mapped neural signal may be measured using, for example, individualized quantitative behavior assessment techniques. In instances where a subject, e.g., a patient, does not display any behavioral response, the accuracy of a mapped neural signal may be assessed using perturbation to external stimulus that are intended to trigger a neural response that is similar to the unperturbed external stimulus. For example, a perturbation to the external stimulus “Move your left index finger if your name is Tom” may be “Move your left middle finger if your name is Tom.” Further to this example, the corresponding neural activity may be different than for the external stimulus “Move your left index finger if your name is Tom”, and the quantitative difference in the recorded neural activity may represent a measure of accuracy. Accuracy of the mapped neural signals, in some embodiments, is assessed by repeating the external stimulus to the patient multiple times (e.g., 2 times, 5 times, 10 times, or more than 10 times), to determine the variability of the neural response to the same stimulus. Accuracy of the neural mapping may be improved by repeating the same external stimulus in different environmental conditions, such as at different times of the day, at different speeds, at different levels of amplitude (e.g., for an auditory stimulus, different dB levels). In some embodiments, accuracy of neural mapping(s) may be determined via a combination of repeating the same external stimulus, providing perturbation stimulus and / or individualized quantitative behavior assessment techniques.

[0068] The method 300 may include the step 310 of repeating the providing, receiving, and mapping steps a plurality of times to generate a training dataset of mapped bioelectrical signals. For example, the steps 304-308 may be repeated any number of times until a sufficient dataset of mapped DB1 / 162944529.6 18Attorney Docket No.: AXF-004PC / 139523-5004-WQ bioelectrical signals for training a neural decoder is generated. For example, the subject, e.g., the patient 10, may be subjected to a series of external stimuli, as described above in relation to step 304. The external stimulus may generate a pattern of neural electrical activity in the brain of the patient 10. In some embodiments, subjecting the patient 10 to a series of external stimuli results in the neural probe 104, 204 recording mappings of the neural activities that are specific to each external stimulus. A set of the recorded mappings of neural activity may constitute a training dataset. In some instances, when the patient 10 is subjected to an external stimulus, features of the neural activity recorded by the neural probe(s) 104, 204 may be compared to the training dataset, by computing device 102, to predict a potential conscious response from the patient to the external stimulus, as discussed in more detail below.

[0069] If a sufficient dataset is not generated, the method 300 may be repeated from step 304 to step 310. If a sufficient dataset has been generated the method may progress from step 310. Sufficient mapping may be a subjective metric defined by the users of the system 100, such as, but not limited to, clinicians or family members to the patient. In some embodiments, sufficient mapping may be achieved when a neural response to a known external stimulus can be attributed by the neural decoder to the correct external stimulus from the dataset with a specific percentage of success. For example, a neural decoder with sufficient mapping dataset may give the correct prediction more than 99% of the time, more than 95% of the time, more than 90% of the time, or more than 75% of the time. In some embodiments, a sufficient dataset is achieved when the percentage of false predictions (e.g., the error rate) is below a specific percentage. For example, a sufficient dataset may be achieved when the error rate is below 1%, below 2%, below 5%, below 10%, or below 25%.

[0070] In response to a sufficient dataset being generated at step 310, the method 300 may progress to step 312, which includes, at the computing device, training a neural decoder algorithm based on the training dataset of mapped bioelectrical signals and without external patient feedback. The computing device 102 may be configured to input the training dataset of mapped bioelectrical signals to a model for training a neural decoding algorithm and output a trained neural decoding algorithm. As discussed above, the computing device 102 maps bioelectrical signals to an expected response to the external stimulus without requiring external feedback from the patient 10. Accordingly, the training dataset of mapped bioelectrical signals generated by the computing device 102 may be generated without external feedback from the patient. By generating a trained neural decoding algorithm without receiving external feedback from the patient, the system 100 and method 300 of the present disclosure may be particularly useful for patients who are incapable of providingDBl / 162944529.6 19Attorney Docket No.: AXF-004PC / 139523-5004-WQ external feedback, which in conventional systems and methods would be required for training a neural decoder.

[0071] The system 100 may be configured to execute the method 300 to generate a trained neural decoding algorithm that may be used to interpret and / or translate neural signals from the subject, e.g., the patient 10. The computing device 102 may store the trained neural decoder algorithm in memory and execute the neural decoder algorithm in response to receiving neural signals from probe(s) 104. In some embodiments, the computing device 102 is configured to input neural signals received from a neural probe 104 to the neural decoder algorithm and output the translation thereof to the communications effector 106.

[0072] Referring to Fig. 7, there is illustrated a method of training a neural decoder, generally designated 400, in accordance with an exemplary embodiment of the present disclosure. The method 400 may include the step 402 of implanting a neural probe into the brain of a subject, e.g., a patient. The step 402 may be generally similar to the step 302 described above, and the neural probe may be any of the neural probes disclosed herein (e.g., the neural probe(s) 104, 204). The method 400 may include the step 404 of recording the state of the patient. The state of the patient may be recorded by any of the methods known in the art, including, without limitation, through clinical observation of the patient by a bedside examination and / or questionnaire, through monitoring of physiological signals of the patient, through medical imaging (e.g., neuroimaging), and / or through video and / or audio monitoring of the patient. In some embodiment, the recordings are themselves passed through a neural network or decoder to approximate a state of the patient that is used in the method 400 at step 404. In some embodiments, the state is a brain state. In some embodiments, the state is a neural decoder classification of brain states, including, without limitation, wakefulness, sleep, local brain integration, global brain integration, brain death, and anesthesia, and different stages and / or depths thereof. For instance, the state may include different sleep states, including, without limitation, REM, deep sleep, NREM 1, NREM 2, and NREM 3. In some embodiments, the brain states are not externally observable. In some embodiments, the brain state is an internal event within the brain of a patient. The internal event may include, without limitation, a seizure, a sub-clinical seizure, a pre-seizure activity, a cortical spreading depression, a P300 wave, a somatosensory evoked potential, and other sensory evoked potential.

[0073] The method 400 may include the step 406 of receiving, at a computing device in communication with the neural probe, and while the state of the subject, e.g., the patient, is being recorded, a recorded bioelectrical signal from the neural probe. For example, while the state of the patient 10 is being recorded, the neural probe(s) 104, 204 implanted in the patient 10 may record a DBl / 162944529.6 20Attorney Docket No.: AXF-004PC / 139523-5004-WQ bioelectrical signal (e.g., neural signal) via electrodes embedded in the probe(s) 104, 204. The neural probe(s) 104, 204 may, in response to recording a bioelectrical signal, be configured to transmit the bioelectrical signal to the computing device 102. For example, neural signals recorded by the shank 112, 212 of the neural probes 104, 204 respectively may be transmitted to the computing device 102 via the respective connectors 114, 214, as discussed above. The method 400 may include the step 408 of mapping, at the computing device, the bioelectrical signal to the recorded state of the patient. The computing device 102 may be configured to receive a bioelectrical signal recorded at step 406 and automatically map that bioelectrical signal to the recorded state of the patient in step 404. In some embodiments, the computing device 102 is configured to assign the mapped bioelectrical signal to a training dataset of mapped bioelectrical signals. In some embodiments, the step 408 of mapping comprises comparing the bioelectrical signals recorded by the electrodes grouped substantially in two or more regions along the shank 112, 212. In some embodiments, the step 408 of mapping comprises calculating the delay timings, information flow, and / or cross-complexity of the bioelectrical signals recorded by the electrodes grouped substantially in two or more regions along the shank 112, 212. In some embodiments, the step 408 of mapping comprises comparing the bioelectrical signals recorded by the electrodes on two or more shanks 112, 212 and / or two or more probes 104a, 104b. In some embodiments, the step 408 of mapping comprises calculating the delay timings, information flow, and / or cross-complexity of the bioelectrical signals recorded by the electrodes on two or more shanks 112, 212 and / or two or more probes 104a, 104b. In some embodiments, the method 400 includes verifying the accuracy of the mapped neural signal, as described above in the method 300. The method 400 may include the step 410 of repeating the recording, receiving, and mapping steps a plurality of times to generate a training dataset of mapped bioelectrical signals. For example, the steps 404-408 may be repeated any number of times until a sufficient dataset of mapped bioelectrical signals for training a neural decoder is generated. The sufficiency of the dataset may be determined by any metrics. For example, a neural decoder with a sufficient mapping dataset is achieved when it gives the correct prediction more than 99% of the time, more than 95% of the time, more than 90% of the time, or more than 75% of the time. In some embodiments, a sufficient dataset is achieved when the percentage of false predictions (e.g., the error rate) is below 1%, below 2%, below 5%, below 10%, or below 25%.

[0074] In some embodiments, the method 400 may include a step of providing one or more external stimuli to the subject. In some embodiments, the one or more external stimuli having an expected response. In some embodiments, the one or more external stimuli are helpful for mapping the recorded bioelectrical signal to the state of a subject, e.g., a patient.DB1 / 162944529.6 21Attorney Docket No.: AXF-004PC / 139523-5004-WQ

[0075] Referring to Fig. 8, there is illustrated a method of using a neural decoder, generally designated 500, in accordance with an exemplary embodiment of the present disclosure. The method 500 may occur following the method 300, discussed above with regards to Fig. 6. For example, the method 500 may occur using the neural decoding algorithm that was generated in step 312 of method 300. The method 500 may include the step 502 of providing an external stimulus to the subject, e.g., the patient. The external stimulus provided in step 502 may be generally similar to a stimulus described above with regards to step 302, except the external stimulus in step 502 may not include an expected response. For example, the external stimulus provided in step 502 may be any type or form of stimulus provided to the patient 10. For instance, the external stimulus may be an auditory stimulus, visual stimulus, tactile stimulus, olfactory stimulus, taste stimulus, proprioceptive stimulus, nociceptive stimulus, or any combinations thereof. The external stimulus may be an external event, including, without limitation, a therapeutic intervention. In some embodiments, the therapeutic intervention comprises administering pharmaceuticals (e.g., amantadine) to the patient 10. In some embodiments, the therapeutic intervention comprises physical rehabilitation actions performed with and / or to the patient 10. In some embodiments, the therapeutic intervention comprises a stimulation applied to the patient 10 to facilitate their recovery. The stimulation may be applied electrically, mechanically, ultrasonically, magnetically, and / or optically. The stimulation may be applied directly to the patient 10 or via an implanted or external systems. In some embodiments, the stimulation is applied to the brain of the patient 10. In some embodiments, the stimulation is applied to a peripheral nerve of the patient 10. The stimulation may be applied to any peripheral nerves of the patient 10. In some embodiments, the stimulation is applied to a spinal nerve or cranial nerve of the patient 10. In some embodiments, the stimulation is applied to a peripheral nerve selected from the sciatic nerve, femoral nerve, ulnar nerve, radial nerve, median nerve, tibial nerve, peroneal nerve, vagus nerve, facial nerve, phrenic nerve, optic nerve, olfactory nerve, oculomotor nerve, trochlear nerve, trigeminal nerve, abducens nerve, facial nerve, vestibulocochlear nerve, glossopharyngeal nerve, accessory nerve, and hypoglossal nerve. In some embodiments, a plurality of external stimuli is provided to the patient 10. In some embodiments, the plurality of external stimuli is played sequentially. In some embodiments, the plurality of external stimuli is preferentially paired together.

[0076] The method 500 may include the step 504 of receiving, at a computing device in communication with a neural probe, a recorded bioelectrical signal from the neural probe. As discussed above with regards to Figs. 1-5, the neural probe(s) 104, 204 may be configured to detect bioelectrical signals in the brain of the patient and transmit the detected bioelectrical signals to the computing device 102. For example, in response to the external stimulus from step 502, a neural probe DBl / 162944529.6 22Attorney Docket No.: AXF-004PC / 139523-5004-WQ104 records a bioelectrical signal in the brain of the patient 10 and transmits it to the computing device 102.

[0077] The method 500 may include the step 506 of mapping, at the computing device, via the trained neural decoding algorithm, the bioelectrical signal to a response to the external stimulus. The computing device 102 may be configured to automatically execute the trained neural decoding algorithm on a received bioelectrical signal received from a neural probe 104, 204. In some embodiments, executing the trained neural decoding algorithm causes the bioelectrical signal to be assigned to a response from the patient 10 to an external stimulus. For example, the trained neural decoding algorithm receives the bioelectrical signal (e.g., neural signal) and translates the signal into a data output indicating one or more words the neural decoding algorithm predicts the patient 10 is attempting to speak. In some embodiments, the mapped response provides information about the state of the patient, and executing the trained neural decoding algorithm causes determining and reporting the mapped response as a metric of the state of the patient (e.g., the state of the patient’s brain health). In some embodiments, the mapped response is used for diagnosis or assisting in diagnosis of the brain state of a subject, e.g., a patient. In some embodiments, the mapped response is used for prognosis or assisting in prognosis of a patient. In some embodiments, the mapped response is used for treatment of a patient. The method 500 may include the step 508 of outputting the mapped response at a communication effector. The computing device 102 may be configured to transmit a mapped response generated by the trained neural decoding algorithm to the communication effector 106 such that the mapped response is output audibly, visually, and / or physically. For example, and continuing from the above example, the computing device 102 causes a speaker of the communication effector 106 to output audibly the words determined by the trained neural decoding algorithm. As another example, the computing device 102 causes a monitor of the communication effector 106 to output visually the state determined by the trained neural decoding algorithm.

[0078] In some embodiments, the method 500 may include outputting at one or more of the computing device and communication effector a confidence value associated with the mapped response. The computing device 102 may be configured to calculate a confidence value representing how likely a mapped response, generated by the trained neural decoding algorithm, is to being correct. In some embodiments, the computing device 102 is configured to calculate the confidence value passively as an estimate over many samples while the neural decoding algorithm is in use. In some embodiments, the passive calculation may include a number of sample observations of output of the neural decoder and / or as an inherent parameter of the neural decoder, such as a statistical model. In some embodiments, the computing device 102 is configured to calculate the confidence value actively DBl / 162944529.6 23Attorney Docket No.: AXF-004PC / 139523-5004-WQ by conducting a set of test and answer questions and in response to the set of test and answer questions generating a statistical sample. The confidence value may be used by the external agent (e. ., a healthcare provider) to contextualize the reliability of the responses provided by the system 100 from the patient.

[0079] Referring to Fig. 9, there is illustrated a method of using a neural decoder, generally designated 600, in accordance with an exemplary embodiment of the present disclosure. The method 600 may occur following method 400. The method 600 may include the step 602 of providing an external stimulus to the subject, e.g., the patient 10. The external stimulus provided in step 602 may be any external stimuli. In some embodiments, the external stimulus provided in step 602 is generally similar to a stimulus described above with regards to step 502. In some embodiments, the external stimulus in step 602 provides information about the state of the patient 10. For instance, the external stimulus may be a sequence of tones with a pattern with a tone changed at the end of the pattern (e.g., an audio oddball paradigm), that elicits local brain responses or global brain responses depending on the sequence of the patterns.

[0080] The method 600 may include the step 604 of receiving, at a computing device in communication with a neural probe, a recorded bioelectrical signal from the neural probe. As discussed above, the neural probe(s) 104, 204 may be configured to detect bioelectrical signals in the brain of the patient 10 and transmit the detected bioelectrical signals to the computing device 102. For example, in response to the external stimulus from the step 602, a neural probe 104 records a bioelectrical signal in the brain of the patient 10 and transmits it to the computing device 102.

[0081] The method 600 may include the step 606 of mapping, at the computing device, via the trained neural decoding algorithm, the bioelectrical signal to a state of the patient 10. The computing device 102 may be configured to automatically execute the trained neural decoding algorithm on a received bioelectrical signal received from a neural probe 104, 204. In some embodiments, executing the trained neural decoding algorithm causes the bioelectrical signal to be assigned to a state of the patient 10. For example, the trained neural decoding algorithm receives the bioelectrical signal (e.g., neural signal) and translates the signal into a data output indicating the state of the patient the neural decoding algorithm predicts.

[0082] The method 600 may include the step 608 of outputting the mapped state at a communication effector (e.g., a patient monitor). The computing device 102 may be configured to transmit a mapped state generated by the trained neural decoding algorithm to the communication effector 106 such that the mapped state is output audibly, visually and / or physically. In some embodiments, the step 608 of outputting the mapped state at the communication effector triggers an DBl / 162944529.6 24Attorney Docket No.: AXF-004PC / 139523-5004-WQ alert and / or a status update. In some embodiments, the method 600 may include outputting at one or more of the computing device and communication effector a confidence value associated with the mapped state. The computing device 102 may be configured to calculate a confidence value representing how likely a mapped state, generated by the trained neural decoding algorithm, is to being correct. Referring to Fig. 10, there is illustrated a method of monitoring a subject, e.g., a patient, with a neural decoder, generally designated 700, in accordance with an exemplary embodiment of the present disclosure. The method 700 may include the step 702 of monitoring, via a computing device in communication with a neural probe, bioelectrical signals of a patient. For example, the computing device 102 is configured to receive, via neural probe(s) 104, 204 bioelectrical signals from the brain of the patient 10. In some embodiments, the monitoring of step 702 is performed generally continuously. For example, neural probes 104, 204 may be configured to transmit bioelectrical signals to the computing device 102 as the signals are detected.

[0083] The method 700 may include the step 704 of determining, at the computing device, if an intent is detected in the bioelectrical signal. The intent may refer to an intent (e.g., an intent to move or intent to speak) of the patient 10 in which the neural probe(s) 104, 204 are embedded. In some embodiments, the computing device 102 is configured to determine if a received bioelectrical signal is associated with an intent of the patient 10. In some embodiments, the computing device 102 is configured to execute the trained neural decoding algorithm on a bioelectrical signal to determine if there is a patient intent associated with the bioelectrical signal.

[0084] In some embodiments, computing device 102 is configured to detect intent based on monitored neural activity over a period of time. The computing device 102 may be configured to detect intent in neural activity by comparing neural activity over a period of time to preceding neural activity. For example, the computing device 102 may be configured to compare the neural activity within a rolling window of one to ten seconds to the neural activity occurring one to two minutes prior. Further to this example, the computing device 102 may determine that neural activity within the rolling window of time is higher than preceding neural activity and automatically determines that intent has been detected in the bioelectrical signal (e.g., neural activity) of the patient.

[0085] In some embodiments, the computing device 102 is configured to determine volitional intent of a patient via the neural decoding algorithm. In some embodiments, the neural decoding algorithm is trained, at least partially, on a clinical assessment for using a command-following task. For example, the neural decoding algorithm may be trained to determine volitional intent based on a clinical assessment of covert consciousness of the patient. An example of a clinical assessment of convert consciousness is described in, Non-Patent Literature to Edlow BL, Fins JI. Entitled DBl / 162944529.6 25Attorney Docket No.: AXF-004PC / 139523-5004-WQ“Assessment of Covert Consciousness in the Intensive Care Unit: Clinical and Ethical Considerations.” J Head Trauma Rehabil. 2018 Nov / Dec;33(6):424-434. doi:10.1097 / HTR.0000000000000448. PMID: 30395042; PMCID: PMC6317885;<https: / / pmc.ncbi.nlm.nih.gov / articles / PMC6317885 / >, the disclosure of which is incorporated by reference herein in their entirety.

[0086] In some embodiments, the neural decoding algorithm is trained at least partially based on data generated during the clinical assessment. For example, a clinician may determine volitional intent of a patient 10 by assessing datasets from fMRI or EEG and comparing them to datasets from healthy patients to make a prediction and generate corresponding data. The data generated by the clinician may be provided to the neural decoding algorithm as a training input such that the neural decoding algorithm is configured to detect neural activity based on the data generated during the clinical assessment performed by the clinician. In some embodiments, a plurality of command-following tasks may be performed during the clinical assessment and a dataset of volitional intent generated therefrom is provided to train the neural decoding algorithm. The resulting trained neural decoding algorithm may be configured to generate an output that is a prediction of volitional intent, based on a specific neural activity pattern measured by the probe 104 in the patient’s brain. The neural decoding algorithm’s output may be a quantitative assessment of the likelihood of volitional intent. The computing device 102 may be configured to determine intent based on the likelihood output by the neural decoding algorithm. For example, the likelihood output may be a percentage between 0% and 99%, where a percentage equal to or greater than 50% corresponds to the neural activity being volitional, and where a percentage below 50% corresponds to the neural activity not representing a volitional intent. It is appreciated that this threshold may be set at any value based on model training and clinical experience.

[0087] In response to no intent being detected at step 704, the method 700 may return to the step 702. In response to an intent being detected at step 704 the method 700 may proceed to the step 706 of mapping, at the computing device via a trained neural decoding algorithm, the bioelectrical signal to the intent of the patient. For example, the computing device 102 executes the neural decoding algorithm, which is trained as discussed above, to map the bioelectrical signal to the intent of the patient 10. In some embodiments, the method 700 includes the step 708 of outputting the mapped intent at a communication effector in communication with the computing device. For example, the computing device 102 may cause the mapped intent to be output via communication effector 106 in generally the same manner as described above with regards to step 508 of method 500.DBl / 162944529.6 26Attorney Docket No.: AXF-004PC / 139523-5004-WQ

[0088] Referring to Fig. 11, there is illustrated a method of using a neural decoder, generally designated 800, in accordance with an exemplary embodiment of the present disclosure. The method 800 may occur following the method 400, discussed with regards to Fig. 7. For example, the method 800 may occur using the neural decoding algorithm that is generated in step 412 of method 400.

[0089] The method 800 may include the step 802 of monitoring, via a computing device in communication with a neural probe, bioelectrical signals of a subject, e.g., a patient. The step 802 may be generally similar to the step 702 described above. For example, the monitoring of step 702 may be performed generally continuously, with neural probes 104 being configured to transmit bioelectrical signals to the computing device 102 as the signals are detected.

[0090] The method 800 may include the step 804 of determining, at the computing device, if a state is detected in the bioelectrical signal. In some embodiments, the computing device 102 is configured to determine if a received bioelectrical signal is associated with a state of the patient 10. In some embodiments, the computing device 102 is configured to execute the trained neural decoding algorithm on a bioelectrical signal to determine if there is a patient state associated with the bioelectrical signal. In some embodiments, the bioelectrical signal is a combination of multimodal neural signals. For instance, the bioelectrical signal may be a combination of neural spikes and neural local field potentials. In some embodiments, the neural spikes are single-unit neural spikes. In some embodiments, the neural spikes are multi-unit neural spikes. In some embodiments, the computing device 102 is configured to execute the trained neural decoding algorithm on a multimodal data to determine if there is a patient state associated with a bioelectrical signal. In some embodiments, the multimodal data is a combination of neural spikes, neural local field potentials, imaging of the brain (e.g., structural imaging from neuroimaging), video and / or audio recordings, and / or other physiological signals of the patient.

[0091] In some embodiments, the state is a brain state. In some embodiments, the state is a neural decoder classification of brain states, including, without limitation, wakefulness, sleep, local brain integration, global brain integration, brain death, and anesthesia, and different stages and / or depths thereof. For instance, the state may include different sleep states, including, without limitation, REM, deep sleep, NREM 1 , NREM 2, and NREM 3. In some embodiments, the brain states are not externally observable. In some embodiments, the state detected at the step 804 is an internal event within the brain of a patient. The internal event may include, without limitation, a seizure, a sub-clinical seizure, a pre-seizure activity, a cortical spreading depression, a P300 wave, a somatosensory evoked potential, and other sensory evoked potential.DBl / 162944529.6 27Attorney Docket No.: AXF-004PC / 139523-5004-WQ

[0092] In response to no state being detected at the step 804, the method 800 may return to the step 802. In response to a state being detected at the step 804 the method 800 may proceed to the step 806 of mapping, at the computing device via a trained neural decoding algorithm, the bioelectrical signal to the state of the patient. For example, the computing device 102 may execute the neural decoding algorithm, which is trained as discussed above, to map the bioelectrical signal to the state of the patient 10. In some embodiments, the method 800 includes the step 808 of outputting the mapped state at a communication effector (e.g., a patient monitor) in communication with the computing device. For example, the computing device 102 may cause the mapped state to be output via communication effector 106 in generally the same manner as described above with regards to step 608 of method 600. In some embodiments, the method 800 outputs surrogate metrics that are useful to an external agent (e.g., a clinician, a healthcare provider, or a decision-making agent). For instance, the method 800 may output surrogate metrics that may be useful to an external agent for diagnosis or assisting in diagnosis of the brain state of a patient. As another example, the method 800 may output surrogate metrics that may be useful to an external agent for prognosis or assisting in prognosis of a patient. In some embodiments, the method 800 outputs surrogate metrics for treatment of a patient. The surrogate metrics may include, without limitation, complexity of signal, loop activity, domain power, and / or Neural Correlates of Consciousness. In some embodiments, the step 808 of outputting the mapped state and / or surrogate metrics at the communication effector triggers an alert and / or a status update. The alert and / or status update may be different from the continuous monitoring described above. The method 800 may function in an automated mode of operation. In some embodiments, the method 800 functions in an automated mode of operation whereby the external agent directs the outputs to different views. In some embodiments, the outputs are available for being viewed simultaneously.

[0093] Referring to Fig. 12, there is illustrated a method of updating and / or retraining a neural decoding algorithm, generally designated 900, in accordance with an exemplary embodiment of the present disclosure. In some embodiments, the method 900 is executed following method 300 of training of the neural decoding algorithm as discussed above with regards to Fig. 6. In some embodiments, the method 900 is executed following one or more of the methods 500, 700 as discussed above with regards to Figs. 6 and 7.

[0094] The method 900 may include the step 902 of receiving, at a computing device, a request to retrain a neural decoding algorithm. For example, the computing device 102 receives the request to retrain the neural decoding algorithm. In some embodiments, the request is input at the computing device 102 or transmitted from a computing device in communication therewith. For example, a user DBl / 162944529.6 28Attorney Docket No.: AXF-004PC / 139523-5004-WQ(e.g., a patient or ahealthcare practitioner) may input at the computing device 102 a request to retrain and / or update the neural decoding algorithm. In some embodiments, the computing device 102 is configured to generate the retraining request based on detected performance of the neural decoding algorithm. For example, the computing device 102 may be configured to track the confidence value of the outputs of the neural decoding algorithm over a period of time. Based on the tracked confidence values the computing device 102 may be configured to determine whether the neural decoding algorithm is performing below a desired performance threshold and initiate retraining of the neural decoding algorithm. For example, if an average of tracked confidence values over a period of time falls below a predetermined threshold, the computing device 102 may be configured to automatically initiate retraining of the neural decoding algorithm. In some embodiments, the computing device 102 is configured to initiate retraining of the neural decoding algorithm according to a predetermined schedule. For example, the computing device 102 may be configured to initiate the retrain or update of the neural decoding algorithm every 7 days.

[0095] In some embodiments, the method 900 includes the step 904 of analyzing bioelectrical signal data to the communication output performance of the neural decoding algorithm over a predetermined period of time. In some embodiments, the computing device 102 and / or a database in communication with the computing device 102 may store time-stamped records of inputs (e.g., bioelectrical signals) and the corresponding outputs of the neural decoding algorithm. The computing device 102 may be configured to, based on at least a portion of the time-stamped records falling within the predetermined period of time, analyze the performance of the neural decoding algorithm. For example, the computing device 102 may determine the performance of the neural decoding algorithm based on the confidence value associated with the outputs of the neural decoding algorithm over the predetermined period of time.

[0096] In some embodiments, the method 900 includes the step 906 of retraining the neural decoding algorithm. In some embodiments, the computing device 102 is configured to execute the step 906 via method 300 discussed above. In some embodiments, the step 906 is only executed in response to the performance of the neural decoding algorithm being below a predetermined threshold. The computing device 102 may be configured to determine that the performance of the neural decoding algorithm, analyzed in step 904, is below a predetermined threshold and execute the step 906. The system 100 may be configured to retrain and / or update the neural decoder such that the performance of the neural decoding algorithm increases over time.

[0097] In some embodiments, aspects of the system 100 and / or methods 300, 400, 500, 600, 700, and 800 may be combined with the method 900 when retraining or updating the performance of the DBl / 162944529.6 29Attorney Docket No.: AXF-004PC / 139523-5004-WQ neural decoder algorithm. In some embodiments, the method 900 may include data collected from multiple patients other than the patient that the neural decoding algorithm is intended for. For example, the computing device 102 may be configured to incorporate data collected from systems 100 associated with a plurality of different patients into the retraining and / or updating of a neural decoding algorithm.

[0098] In some embodiments, method 900 includes quantifying the change in the patient’s health state. For example, the computing device 102 may be configured to determine a complexity of the bioelectric signals available to the neural decoder and determine whether the bioelectric signals are getting more or less complex. The computing device 102 may be configured to associate the increasing and / or decreasing complexity of bioelectrical signals with improvement and / or worsening of the patient’s health.

[0099] In some embodiments, the neural decoder algorithms and associated methods of training / use thereof, as discussed herein, may be configured to output a set of learnt 1 : 1 mappings between bioelectric signals and intent or state that is output to a communication effector. In one embodiment, the neural decoder of system 100 is a foundational model that, having been provided a corpus of bioelectric data from one or more subjects, e.g., patients, which may or may not include the subject, e.g., the patient for which the neural decoder is intended, is able to, without undergoing training as per method 300 or method 400, conduct the steps in one or more of methods 500, 600, 700, and 800 to map bioelectric signal to response, intent, and / or state of the subject, e.g., the patient (e.g., a zero-shot decoding). In some embodiments, the neural decoder of system 100 is trained on a corpus of bioelectric data from one or more subjects, e.g., patients, that does not include the subject, e.g., the patient, for which the neural decoder is in use and is able to, while undergoing either no training (e.g., a zero-shot functionality), one instance of training (e.g., a one-shot functionality), or a small amount of training (e.g., few-shot functionality) with data from the subject, e.g., the patient for which the neural decoder is in use, map bioelectrical signals to the response, intent, and / or state of the subject, e.g., the patient. In some embodiments, the neural decoder of system 100 is trained on a corpus of bioelectric data including data from one or more patients with different conditions, including, without limitation, epilepsy, DOC, and other neurological conditions. In some embodiments, the neural decoder of system 100 is trained on data from one or more subjects across species, including, without limitation, rodents (e.g., mice and rats), large mammals (e.g., pigs and sheep), non-human primates, and human, and is able to map bioelectrical signals to the response, intent, and / or state of a human patient.DBl / 162944529.6 30Attorney Docket No.: AXF-004PC / 139523-5004-WQ

[0100] In some embodiments, the neural decoding systems disclosed herein (e.g., the neural decoding system 100) provide multiple methods of interrogating the response, intent, and / or state of a subject, e.g., a patient. The methods disclosed herein may be combined with one another in any manner. For instance, a method with a combination of the methods 300 and 400 may comprise the steps of implanting a neural probe into the brain of a patient, providing an external stimulus to the patient, receiving a recorded bioelectrical signal from the neural probe, recording a state of the patient, mapping the bioelectrical signal to an expected response to the stimulus, mapping the recorded state of the patient to the expected response to the stimulus, determining the sufficiency of the mapping, optionally repeating the steps of providing, receiving, recording, and mapping, and training a neural decoder based on the mapped signals and / or states. As another example, a method with a combination of the methods 500 and 600 is disclosed. In some embodiments, the method with a combination of the methods 500 and 600 occurs after the method with a combination of the methods 300 and 400. The method with a combination of the methods 500 and 600 may comprise the steps of providing an external stimulus to a patient, receiving a recorded bioelectrical signal from the neural probe, mapping the bioelectrical signal to a response to the stimulus, outputting the mapped response at a communication effector, and outputting a state of the patient (e.g., by using the mapped response as a metric of the state of the patient or by using the neural decoder trained in the method with a combination of the methods 300 and 400) at the communication effector.[00101J In some embodiments, the methods disclosed herein are conducted in an automated mode (e.g., without any input or interference from an external agent). In some embodiments, the methods disclosed herein are conducted with input from an external agent (e.g., clinicians or healthcare providers). In some embodiments, the input is directed for certain external stimuli to be effectuated. In some embodiments, the external stimuli are effectuated in a predefined order and / or timing.

[0102] In some embodiments, the methods disclosed herein further comprise a step of analyzing cross-region brain activities (e.g., bioelectrical signals recorded from different brain regions). The step of analyzing cross region brain activities may include, without limitation, detecting information in the bioelectrical signals that is characteristic of feedforward or feedback neural activity between cortical layers or any two or more regions of the brain. In some embodiments, a neural probe 104 comprising a shank 112 with the electrodes housed on the shank 112 arranged substantially in two or more groups along the length of the shank 112 is used in any of the methods disclosed herein. For instance, the electrodes may be arranged substantially in a first group and a second group along the shank 112, as described above with respect to Fig. 3. In some embodiments, the methods disclosed herein comprise a step of comparing the bioelectrical signals from the first group of electrodes and the bioelectrical DB1 / 162944529.6 31Attorney Docket No.: AXF-004PC / 139523-5004-WQ signals from the second group of electrodes. In some embodiments, the first group of electrodes and the second group of electrodes are implanted within different regions of the brain of a patient. For instance, the first group of electrodes may be in the cortex of the patient while the second group of electrodes may be in the thalamus of the patient. In some embodiments, two or more neural probes (e.g., 104a, 104b as described in Fig. 4) are used in any of the methods disclosed herein. In some embodiments, the two or more neural probes are implanted in different regions of the brain of a patient. For instance, one neural probe may be implanted in a cortical column while another neural probe may be implanted in a different cortical column, or one neural probe may be implanted in a cortical layer while another neural probe may be implanted in a different cortical layer. In some embodiments, the methods disclosed herein comprises the step of comparing the bioelectrical signals from one neural probe and the bioelectrical signals from another neural probe. The step of comparing may comprise calculating the delay timings, information flow, cross-complexity, and / or other information metrics of the recorded bioelectrical signals.

[0103] The step of comparing may be preceded by a step of determining the location of the implanted neural probe and / or electrodes within the brain of a subject, e.g., a patient. In some embodiments, the location of the implanted neural probe and / or electrodes is determined on the basis of any information from pre or post operative brain imaging and / or any characteristics of the recorded bioelectrical signal, including, without limitation, current source density, bursting patterns, neural waveforms, and local field potential waveforms.

[0104] In some embodiments, the data input for training (e.g., the training dataset) and / or using the neural decoder is multimodal, as illustrated in Fig. 13, in accordance with an exemplary embodiment of the present disclosure. In some embodiments, the data input comprises neural spikes, neural local field potentials, location information of the shanks and / or electrodes implanted into the brain, structural imaging of the brain from neuroimaging, video and / or audio recording of patients, physiological signals of patients, and a combination thereof.

[0105] Illustration of Subject Technology as Clauses:

[0106] Various examples of aspects of the disclosure are described as numbered clauses (1, 2, 3, etc.) for convenience. These are provided as examples, and do not limit the subject technology. Identifications of the figures and reference numbers are provided below merely as examples and for illustrative purposes, and the clauses are not limited by those identifications.

[0107] Clause 1. A method of training a neural decoder, the method comprising: implanting a neural probe into the brain of a subject, the neural probe configured to record and transmit bioelectrical signals; providing one or more external stimuli to the subject, the one or more external stimuli having DBl / 162944529.6 32Attorney Docket No.: AXF-004PC / 139523-5004-WQ an expected response; receiving, at a computing device in communication with the neural probe, and in response to the one or more external stimuli, a recorded bioelectrical signal from the neural probe; mapping, at the computing device, the bioelectrical signal to the expected response; repeating the providing, receiving, and mapping steps a plurality of times to generate a training dataset of mapped bioelectrical signals; and at the computing device, training a neural decoder algorithm based on the training dataset of mapped bioelectrical signals and without external subject feedback.

[0108] Clause 2. A method of training a neural decoder, the method comprising: implanting a neural probe into the brain of a subject, the neural probe configured to record and transmit bioelectrical signals; recording a brain state of the subject; receiving, at a computing device in communication with the neural probe, a recorded bioelectrical signal from the neural probe while the brain state of the subject is being recorded; mapping, at the computing device, the bioelectrical signal to the recorded brain state of the subject; repeating the recording, receiving, and mapping steps a plurality of times to generate a training dataset of mapped bioelectrical signals; and at the computing device, training a neural decoder algorithm based on the training dataset of mapped bioelectrical signals and without external subject feedback.

[0109] Clause 3. A method of interpreting neural activity of a subject using the neural decoder trained by the method of clause 1 or 2, the method comprising: receiving, at a computing device a recorded bioelectrical signal from a neural probe that is implanted into the brain of the subject; decoding, at the computing device via the neural decoder trained by the method of clause 1 or 2, the bioelectrical signal into an intent and / or brain state of the subject; and transmitting the decoded intent and / or brain state to a communication effector in communication with the computing device.

[0110] Clause 4. The method of clause 2, further comprising providing one or more external stimuli to the subject, the one or more external stimuli having an expected response.

[0111] Clause 5. The method of clause 3, wherein the recorded bioelectrical signal is recorded in response to one or more external stimuli provided to the subject.

[0112] Clause 6. The method of clause 3 or 5, wherein the communication effector comprises one or more sensors, monitors, speakers, and / or microphones.

[0113] Clause 7. The method of any one of clauses 3, 5, and 6, wherein the steps of receiving, decoding, and transmitting are performed continuously to monitor the subject.

[0114] Clause 8. The method of any one of clauses 1, and 4-7, wherein the one or more external stimuli are provided to the subject in a predefined order and / or at a predefined timing.

[0115] Clause 9. The method of any one of clauses 1, and 4-8, wherein the one or more external stimuli comprise an auditory stimulus, an auditory binary question, a non-visual stimulus, a tactile DBl / 162944529.6 33Attorney Docket No.: AXF-004PC / 139523-5004-WQ stimulus, a visual stimulus, an olfactory stimulus, a taste stimulus, a proprioceptive stimulus, or a nociceptive stimulus, or any combinations thereof.

[0116] Clause 10. The method of any one of clauses 1, and 4-9, wherein two or more external stimuli are provided to the subject concurrently.

[0117] Clause 11. The method of any one of clauses 1-10, wherein the subject is in a state of cognitive motor dissociation, a vegetative state, a minimally conscious state, or a coma, or any combinations thereof.

[0118] Clause 12. The method of any one of clauses 1-11, wherein the subject has a disorder of consciousness (DOC) and / or locked-in syndrome.

[0119] Clause 13. The method of any one of clauses 1-12, wherein the neural probe is implanted through a burr hole with a diameter of less than 2.4 mm.

[0120] Clause 14. The method of any one of clauses 1-13, wherein the neural probe comprises a shank configured to be inserted into the brain of the subject, optionally wherein the shank comprises a plurality of electrodes, optionally wherein the shank has a number of electrodes per cross-sectional area of the shank that is about 10'7electrodes / micron2to about 102electrodes / micron2, optionally wherein the electrodes are grouped substantially in two or more regions along the shank, optionally wherein the grouped electrodes are configured to record the bioelectrical signal from two or more brain regions of the subject, optionally wherein the two or more brain regions are selected from cerebral cortex, thalamus, cerebellum, hypothalamus, hippocampus, pituitary gland, brain stem, spinal cord, ventricles, putamen, midbrain, pons, medulla, striatum, amygdala, basal ganglia, pineal gland, frontal lobe, temporal lobe, occipital lobe, parietal lobe, motor strip, sensory strip, Broca’s area, Wernicke’s area, hippocampal formation, hippocampus proper, cingulate gyrus, limbic system, somatosensory cortex, motor cortex, sensory cortex, auditory cortex, prefrontal cortex, entorhinal cortex, visual cortex, perirhinal cortex, insular cortex, primary motor area, supplementary motor area, primary auditory area, auditory association area, primary visual area, and visual association area.

[0121] Clause 15. The method of clause 14, wherein the two or more brain regions comprise a cortex region and the thalamus of the subject.

[0122] Clause 16. The method of clause 14, wherein the electrodes are grouped substantially in a first region and a second region along the shank such that, after being inserted into the brain, the electrodes in the first region primarily record the bioelectrical signal from a region of the cerebral cortex of the subject and the electrodes in the second region primarily record the bioelectrical signal from an inner region of the subject’s brain, optionally wherein the inner region is selected from theDBl / 162944529.6 34Attorney Docket No.: AXF-004PC / 139523-5004-WQ thalamus, the hypothalamus, the ventricles, and the cerebellum, optionally wherein the first region and the second region are about 1 mm to about 120 mm apart along the shank.

[0123] Clause 17. The method of clause 14, wherein the step of mapping and / or the step of decoding comprise comparing the bioelectrical signals recorded by the electrodes grouped substantially in two or more regions along the shank, optionally wherein the step of comparing comprises calculating the delay timings, information flow, and / or cross complexity of the bioelectrical signals recorded by the electrodes grouped substantially in two or more regions along the shank.

[0124] Clause 18. The method of any one of clauses 1-17, wherein two or more probes are implanted into the brain of the subject, and wherein the step of mapping and / or the step of decoding comprise comparing the bioelectrical signals recorded by electrodes on the two or more probes, optionally wherein the step of comparing comprises calculating the delay timings, information flow, and / or cross complexity of the bioelectrical signals recorded by the electrodes on the two or more probes.

[0125] Clause 19. The method of any one of clauses 2-18, wherein the brain state of the subject is an internal event within the brain of the subject, optionally wherein the brain state of the subject is selected from a seizure, a sub-clinical seizure, a pre-seizure activity, a cortical spreading depression, a P300 wave, a somatosensory evoked potential, and other sensory evoked potential.

[0126] Clause 20. The method of any one of clauses 1-19, wherein the neural probe comprises a fluorinated elastomer, optionally wherein the neural probe comprises a perfluorinated elastomer.

[0127] Clause 21. The method of any one of clauses 1, and 3-20, wherein the one or more external stimuli are provided to the subject by one or more of: the computing device, an artificial intelligence, and a human being, or any combinations thereof.

[0128] Clause 22. The method of any one of clauses 1-21, wherein the bioelectrical signal comprises neural spikes and / or local field potentials.

[0129] Clause 23. The method of any one of clauses 1, 2, 4, and 8-22, wherein the training dataset further comprises location information of the neural probe, images of the brain of the subject, video and / or audio recordings of the subject, and / or physiological signals of the subject.

[0130] Clause 24. The method of clause 23, wherein the location information of the neural probe is determined based on characteristics of the bioelectrical signal and / or by brain imaging, optionally wherein the characteristics of the bioelectrical signal are selected from current source density, bursting patterns, neural spike waveforms, and local field potential waveforms.

[0131] Clause 25. The method of any one of clauses 1-24, wherein the subject is a human, a rodent, a pig, a sheep, or a non-human primate.DBl / 162944529.6 35Attorney Docket No.: AXF-004PC / 139523-5004-WQ

[0132] Clause 26. A system for implementing the method of clause 1, the system comprising: a neural probe configured to be implanted into the brain of a subject, the neural probe configured to record and transmit bioelectrical signals; and a computing device in communication with the neural probe, the computing device configured to: in response to one or more external stimuli provided to the subject, receive a recorded bioelectrical signal from the neural probe, map the bioelectrical signal to an expected response of the one or more external stimuli, generate a training dataset of mapped bioelectrical signals by repeatedly receiving and mapping bioelectrical signals, and based on the training dataset of mapped bioelectrical signals, train a neural decoder algorithm without external subject feedback.

[0133] Clause 27. A system for implementing the method of clause 2, the system comprising: a neural probe configured to be implanted into the brain of a subject, the neural probe configured to record and transmit bioelectrical signals; and a computing device in communication with the neural probe, the computing device configured to: receive a recorded bioelectrical signal from the neural probe, map the bioelectrical signal to a brain state of the subject, generate a training dataset of mapped bioelectrical signals by repeatedly receiving and mapping bioelectrical signals, and based on the training dataset of mapped bioelectrical signals, train a neural decoder algorithm without external subject feedback.

[0134] Clause 28. A system for implementing the method of clause 3, the system comprising: a neural probe configured to be implanted into the brain of a subject, the neural probe configured to record and transmit bioelectrical signals; a computing device in communication with the neural probe, the computing device configured to: receive a recorded bioelectrical signal from the neural probe, and decode, via the neural decoder trained by the method of clause 1 or 2, the bioelectrical signal into an intent and / or brain state of the subject; and a communication effector in communication with the computing device, the communication effector configured to output the decoded intent and / or brain state of the subject.

[0135] Clause 29. The system of clause 27, wherein the recorded bioelectrical signal is recorded in response to one or more external stimuli provided to the subject, the one or more external stimuli having an expected response.

[0136] Clause 30. The system of clause 28, wherein the recorded bioelectrical signal is recorded in response to one or more external stimuli provided to the subject, optionally wherein the one or more external stimuli are provided to the subject in a predefined order and / or at a predefined timing.

[0137] Clause 31. The system of clause 28 or 30, wherein the system is configured to continuously interpret neural activity of the subject to monitor the subject.DBl / 162944529.6 36Attorney Docket No.: AXF-004PC / 139523-5004-WQ

[0138] Clause 32. The system of any one of clauses 28, 30, and 31, wherein the communication effector comprises one or more sensors, monitors, speakers, and / or microphones.

[0139] Clause 33. The system of any one of clauses 28, and 30-32, wherein the communication effector is configured to output the decoded intent and / or brain state of the subject as a visual display and / or an audio output.

[0140] Clause 34. The system of any one of clauses 26, and 29-33, wherein the one or more external stimuli comprise an auditory stimulus, an auditory binary question, a non-visual stimulus, a tactile stimulus, a visual stimulus, an olfactory stimulus, a taste stimulus, a proprioceptive stimulus, or a nociceptive stimulus, or any combinations thereof.

[0141] Clause 35. The system of any one of clauses 26, and 29-34, wherein two or more external stimuli are provided to the subject concurrently.

[0142] Clause 36. The system of any one of clauses 26, and 29-35, wherein the one or more external stimuli are provided to the subject in a predefined order and / or at a predefined timing.

[0143]

[0144] Clause 37. The system of any one of clauses 26-36, wherein the subject is in a state of cognitive motor dissociation, a vegetative state, a minimally conscious state, or a coma, or any combinations thereof.

[0145] Clause 38. The system of any one of clauses 26-37, wherein the subject has a disorder of consciousness (DOC) and / or locked-in syndrome.

[0146] Clause 39. The system of any one of clauses 26-38, wherein the neural probe is configured to be implanted through a burr hole with a diameter of less than 2.4 mm.

[0147] Clause 40. The system of any one of clauses 26-39, wherein the neural probe comprises a shank configured to be inserted into the brain of the subject, optionally wherein the shank comprises a plurality of electrodes, optionally wherein the shank has a number of electrodes per cross-sectional area of the shank that is about 10'7electrodes / micron2to about 102electrodes / micron2, optionally wherein the electrodes are grouped substantially in two or more regions along the shank, optionally wherein the grouped electrodes are configured to record the bioelectrical signal from two or more brain regions of the subject, optionally wherein the two or more brain regions are selected from cerebral cortex, thalamus, cerebellum, hypothalamus, hippocampus, pituitary gland, brain stem, spinal cord, ventricles, putamen, midbrain, pons, medulla, striatum, amygdala, basal ganglia, pineal gland, frontal lobe, temporal lobe, occipital lobe, parietal lobe, motor strip, sensory strip, Broca’s area, Wernicke’s area, hippocampal formation, hippocampus proper, cingulate gyrus, limbic system, somatosensory cortex, motor cortex, sensory cortex, auditory cortex, prefrontal cortex, entorhinalDBl / 162944529.6 37Attorney Docket No.: AXF-004PC / 139523-5004-WQ cortex, visual cortex, perirhinal cortex, insular cortex, primary motor area, supplementary motor area, primary auditory area, auditory association area, primary visual area, and visual association area.

[0148] Clause 41. The system of clause 40, wherein the two or more brain regions comprise a cortex region and the thalamus of the subject.

[0149] Clause 42. The system of clause 40, wherein the electrodes are grouped substantially in a first region and a second region along the shank such that, after being inserted into the brain, the electrodes in the first region primarily record the bioelectrical signal from a region of the cerebral cortex of the subject and the electrodes in the second region primarily record the bioelectrical signal from an inner region of the subject’s brain, optionally wherein the inner region is selected from the thalamus, the hypothalamus, the ventricles, and the cerebellum, optionally wherein the first region and the second region are about 1 mm to about 120 mm apart along the shank.

[0150] Clause 43. The system of clause 40, wherein the computing device is configured to compare the bioelectrical signals recorded by the electrodes grouped substantially in two or more regions along the shank when mapping the bioelectrical signal to an expected response of the one or more external stimuli, mapping the bioelectrical signal to a brain state of the subject, and / or decoding the bioelectrical signal into an intent and / or brain state of the subject, optionally wherein the step of comparing comprises calculating delay timings, information flow, and / or cross complexity of the bioelectrical signals recorded by the electrodes grouped substantially in two or more regions along the shank.

[0151] Clause 44. The system of any one of clauses 26-43, wherein two or more probes are implanted into the brain of the subject, and wherein the computing device is configured to compare the bioelectrical signals recorded by electrodes on the two or more probes when mapping the bioelectrical signal to an expected response of the one or more external stimuli, mapping the bioelectrical signal to a brain state of the subject, and / or decoding the bioelectrical signal into an intent and / or brain state of the subject, optionally wherein the step of comparing comprises calculating delay timings, information flow, and / or cross complexity of the bioelectrical signals recorded by the electrodes on the two or more probes.

[0152] Clause 45. The system of any one of clauses 27-44, wherein the brain state of the subject is an internal event within the brain of the subject, optionally wherein the brain state of the subject is selected from a seizure, a sub-clinical seizure, a pre-seizure activity, a cortical spreading depression, a P300 wave, a somatosensory evoked potential, and other sensory evoked potential.

[0153] Clause 46. The system of any one of clauses 26-45, wherein the neural probe comprises a fluorinated elastomer, optionally wherein the neural probe comprises a perfluorinated elastomer.DB1 / 162944529.6 38Attorney Docket No.: AXF-004PC / 139523-5004-WQ

[0154] Clause 47. The system of any one of clauses 26-46, wherein the one or more external stimuli is provided to the subject by one of: the computing device, an artificial intelligence, and a human being.

[0155] Clause 48. The system of any one of clauses 26-47, wherein the bioelectrical signal comprises neural spikes and / or local field potentials.

[0156] Clause 49. The system of any one of clauses 26, 27, 29, and 34-48, wherein the training dataset further comprises location information of the neural probe, images of the brain of the subject, video and / or audio recordings of the subject, and / or physiological signals of the subject.

[0157] Clause 50. The system of clause 49, wherein the location information of the neural probe is determined based on characteristics of the bioelectrical signal and / or by brain imaging, optionally wherein the characteristics of the bioelectrical signal are selected from current source density, bursting patterns, neural spike waveforms, and local field potential waveforms.

[0158] Clause 51. The system of any one of clauses 26-50, wherein the subject is a human, a rodent, a pig, a sheep, or a non-human primate.

[0159] Clause 52. Use of the method of any one of clauses 1-25 for the diagnosis, prognosis, and / or treatment of a patient.

[0160] Clause 53. Use of the system of any one of clauses 26-51 for the diagnosis, prognosis, and / or treatment of a patient.EXAMPLES

[0161] Example 1. Training and Using A Neural Decoder with Rats

[0162] In this example, a neural decoding system was trained with and used on rats. Fig. 14A shows the dataset used to conduct a supervised training of a neural decoder mapping bioelectrical signals (i.e., neural spikes and local field potentials) to four previously observed brain states (or diagnoses) and the brain states (or diagnoses) output by the trained neural decoder. A behavioral score was assigned to each of the four brain states. The output brain state was the maximum of four similarity scores where the neural decoder mapped the bioelectrical signal to match one of the four brain states and at each timestep output a similarity score to each of the four brain states, with the state getting the maximum similarity score being declared the predicted state by the neural decoder. The overall accuracy of the trained neural decoding system was above 94%. The neural decoder may also be trained in an unsupervised manner, as shown in Fig. 14B. Tn the unsupervised training, the neural decoder output at each time step a sample within an n-dimensional latent space (e.g., a 2-dimensional latent space). The neural decoder successfully separated in the latent space the time samples associatedDBl / 162944529.6 39Attorney Docket No.: AXF-004PC / 139523-5004-WQ with a behavioral state score from 1-10. Each future time sample can then be mapped by the neural decoder to a decoded brain state by comparison of the new time sample to the existing classification of the latent space. This decoded brain state may be used by an external agent to assist diagnosis of, or directly as a diagnosis of the brain state of the patient. This decoded brain state may also be used for prognosis or assisting in prognosis of a patient.

[0163] Example 2. Training and Using A Neural Decoder with Human Subjects

[0164] In this example, a trained neural decoding system was used to decode the brain state of human patients with an auditory stimulus applied to the patients. The local and global responses (e.g., the recorded bioelectrical signals) to the auditory stimulus were used to decode the brain state of the patients, as depicted in Fig. 15 A. The decoded brain states of the patient were consistent with the clinical observations of the patients (Fig. 15B). A neural decoder trained with data from rats in an unsupervised manner (e.g., as described in Example 1) also accurately decoded the brain states of the human patients, as demonstrated in Figs. 15C and 15D. These decoded brain states may be used by an external agent to assist diagnosis of, or directly as a diagnosis of brain stated of patients. These decoded brain states may also be used for prognosis or assisting in prognosis of patients. These results demonstrate, inter alia, the efficacy, reliability, adaptability of the methods and systems disclosed herein.

[0165] The term “about” or “approximately” is used herein to provide literal support for the exact number that it precedes, as well as a number that is near to or approximately the number that the term precedes. In determining whether a number is near to or approximately a specifically recited number, the near or approximating unrecited number may be a number, which, in the context in which it is presented, provides the substantial equivalent of the specifically recited number. It should be appreciated that all numerical values and ranges disclosed herein are approximate values and ranges, whether or not “about” is used in conjunction therewith. It should also be appreciated that the term “about,” as used herein, in conjunction with a numeral refers to a value that may be ±0.01% (inclusive) of that numeral, ±0.1% (inclusive) of that numeral, ±0.5% (inclusive) of that numeral, ±1% (inclusive) of that numeral, ±2% (inclusive) of that numeral, ±3% (inclusive) of that numeral, ±5% (inclusive) of that numeral, ±10% (inclusive) of that numeral, or ±15% (inclusive) of that numeral. It should further be appreciated that when a numerical range is disclosed herein, any numerical value falling within the range is also specifically disclosed.

[0166] It will be appreciated by those skilled in the art that changes could be made to the exemplary embodiments shown and described above without departing from the broad inventive concepts thereof. It is to be understood that the embodiments and claims disclosed herein are not DBl / 162944529.6 40Attorney Docket No.: AXF-004PC / 139523-5004-WQ limited in their application to the details of construction and arrangement of the components set forth in the description and illustrated in the drawings. Rather, the description and the drawings provide examples of the embodiments envisioned. The embodiments and claims disclosed herein are further capable of other embodiments and of being practiced and carried out in various ways.

[0167] Specific features of the exemplary embodiments may or may not be part of the claimed invention and various features of the disclosed embodiments may be combined. Unless specifically set forth herein, the terms “a”, “an” and “the” are not limited to one element but instead should be read as meaning “at least one”. Finally, unless specifically set forth herein, a disclosed or claimed method should not be limited to the performance of their steps in the order written, and one skilled in the art can readily appreciate that the steps may be performed in any practical order.DBl / 162944529.6 41

Claims

Attorney Docket No.: AXF-004PC / 139523-5004-WQCLAIMSWhat is claimed is:

1. A method of training a neural decoder, the method comprising: implanting a neural probe into the brain of a subject, the neural probe configured to record and transmit bioelectrical signals; providing one or more external stimuli to the subject, the one or more external stimuli having an expected response; receiving, at a computing device in communication with the neural probe, and in response to the one or more external stimuli, a recorded bioelectrical signal from the neural probe; mapping, at the computing device, the bioelectrical signal to the expected response; repeating the providing, receiving, and mapping steps a plurality of times to generate a training dataset of mapped bioelectrical signals; and at the computing device, training a neural decoder algorithm based on the training dataset of mapped bioelectrical signals and without external subject feedback.

2. A method of training a neural decoder, the method comprising: implanting a neural probe into the brain of a subject, the neural probe configured to record and transmit bioelectrical signals; recording a brain state of the subject; receiving, at a computing device in communication with the neural probe, a recorded bioelectrical signal from the neural probe while the brain state of the subject is being recorded; mapping, at the computing device, the bioelectrical signal to the recorded brain state of the subject; repeating the recording, receiving, and mapping steps a plurality of times to generate a training dataset of mapped bioelectrical signals; and at the computing device, training a neural decoder algorithm based on the training dataset of mapped bioelectrical signals and without external subject feedback.DBl / 162944529.6 42Attorney Docket No.: AXF-004PC / 139523-5004-WQ3. The method of claim 2, further comprising providing one or more external stimuli to the subject, the one or more external stimuli having an expected response.

4. The method of claim 1 or 3, wherein the one or more external stimuli comprise an auditory stimulus, an auditory binary question, a non-visual stimulus, a tactile stimulus, a visual stimulus, an olfactory stimulus, a taste stimulus, a proprioceptive stimulus, or a nociceptive stimulus, or any combinations thereof, wherein the one or more external stimuli are provided to the subject by one or more of: the computing device, an artificial intelligence, and a human being, or any combinations thereof, and / or wherein the one or more external stimuli are provided to the subject in a predefined order and / or at a predefined timing.

5. The method of claim 1 or 3, wherein two or more external stimuli are provided to the subject concurrently.

6. The method of claim 1 or 2, wherein the subject is in a state of cognitive motor dissociation, a vegetative state, a minimally conscious state, or a coma, or any combinations thereof.

7. The method of claim 1 or 2, wherein the subject has a disorder of consciousness (DOC) and / or locked-in syndrome.

8. The method of claim 1 or 2, wherein the neural probe comprises a fluorinated elastomer, wherein the neural probe comprises a perfluorinated elastomer, and / or wherein the neural probe is implanted through a burr hole with a diameter of less than 2.4 mm.

9. The method of claim 1 or 2, wherein the neural probe comprises a shank configured to be inserted into the brain of the subject, and wherein the shank comprises a plurality of electrodes, optionally wherein the shank has a number of electrodes per cross-sectional area of the shank that is about 10'7electrodes / micron2to about 102electrodes / micron2, optionally wherein the electrodes are grouped substantially in two or more regions along the shank, optionally wherein the groupedDBl / 162944529.6 43Attorney Docket No.: AXF-004PC / 139523-5004-WQ electrodes are configured to record the bioelectrical signal from two or more brain regions of the subject, optionally wherein the two or more brain regions are selected from the cerebral cortex, thalamus, cerebellum, hypothalamus, hippocampus, pituitary gland, brain stem, spinal cord, ventricles, putamen, midbrain, pons, medulla, striatum, amygdala, basal ganglia, pineal gland, frontal lobe, temporal lobe, occipital lobe, parietal lobe, motor strip, sensory strip, Broca’s area, Wernicke’s area, hippocampal formation, hippocampus proper, cingulate gyrus, limbic system, somatosensory cortex, motor cortex, sensory cortex, auditory cortex, prefrontal cortex, entorhinal cortex, visual cortex, perirhinal cortex, insular cortex, primary motor area, supplementary motor area, primary auditory area, auditory association area, primary visual area, and visual association area.

10. The method of claim 9, wherein the two or more brain regions comprise a cortex region and the thalamus of the subject.

11. The method of claim 9, wherein the electrodes are grouped substantially in a first region and a second region along the shank such that, after being inserted into the brain, the electrodes in the first region primarily record the bioelectrical signal from a region of the cerebral cortex of the subject and the electrodes in the second region primarily record the bioelectrical signal from an inner region of the subject’s brain, optionally wherein the inner region is selected from the thalamus, the hypothalamus, the ventricles, and the cerebellum, optionally wherein the first region and the second region are about 1 mm to about 120 mm apart along the shank.

12. The method of claim 9, wherein the step of mapping comprises comparing the bioelectrical signals recorded by the electrodes grouped substantially in two or more regions along the shank, optionally wherein the step of comparing comprises calculating delay timings, information flow, and / or cross-complexity of the bioelectrical signals recorded by the electrodes grouped substantially in two or more regions along the shank.

13. The method of claim 2, wherein the brain state of the subject is an internal event within the brain of the subject, optionally wherein the brain state of the subject is selected from a seizure, a sub-DBl / 162944529.6 44Attorney Docket No.: AXF-004PC / 139523-5004-WQ clinical seizure, a pre-seizure activity, a cortical spreading depression, a P300 wave, a somatosensory evoked potential, and other sensory evoked potential.

14. The method of claim 1 or 2, wherein the bioelectrical signal comprises single-unit and / or multi-unit neural spikes and / or local field potentials.

15. The method of claim 1 or 2, wherein the training dataset further comprises location information of the neural probe, images of the brain of the subject, video and / or audio recordings of the subject, and / or physiological signals of the subject, optionally wherein the location information of the neural probe is determined based on characteristics of the bioelectrical signal and / or by brain imaging, optionally wherein the characteristics of the bioelectrical signal are selected from current source density, bursting patterns, neural spike waveforms, and local field potential waveforms.

16. The method of claim 1 or 2, wherein the subject is a human, a rodent, a pig, a sheep, or a nonhuman primate.

17. A system for implementing the method of any one of claims 1, and 4-16, the system comprising: a neural probe configured to be implanted into the brain of a subject, the neural probe configured to record and transmit bioelectrical signals; and a computing device in communication with the neural probe, the computing device configured to: in response to one or more external stimuli provided to the subject, receive a recorded bioelectrical signal from the neural probe; map the bioelectrical signal to an expected response of the external stimulus; generate a training dataset of mapped bioelectrical signals by repeatedly receiving and mapping bioelectrical signals; and based on the training dataset of mapped bioelectrical signals, train a neural decoder algorithm without external subject feedback.DBl / 162944529.6 45Attorney Docket No.: AXF-004PC / 139523-5004-WQ18. A system for implementing the method of any one of claims 2-16, the system comprising: a neural probe configured to be implanted into the brain of a subject, the neural probe configured to record and transmit bioelectrical signals; and a computing device in communication with the neural probe, the computing device configured to: receive a recorded bioelectrical signal from the neural probe; map the bioelectrical signal to a brain state of the subject; generate a training dataset of mapped bioelectrical signals by repeatedly receiving and mapping bioelectrical signals; and based on the training dataset of mapped bioelectrical signals, train a neural decoder algorithm without external subject feedback.

19. A method of interpreting neural activity of a subject using a neural decoder trained by the method of any one of claims 1-16, the method comprising: receiving, at a computing device a recorded bioelectrical signal from a neural probe that is implanted into the brain of the subject; decoding, at the computing device via the neural decoder trained by the method of any one of claims 1-16, the bioelectrical signal into an intent and / or brain state of the subject; and transmitting the decoded intent and / or brain state to a communication effector in communication with the computing device.

20. A system for implementing the method of claim 19, the system comprising: a neural probe configured to be implanted into the brain of a subject, the neural probe configured to record and transmit bioelectrical signals; a computing device in communication with the neural probe, the computing device configured to: receive a recorded bioelectrical signal from the neural probe; andDBl / 162944529.6 46Attorney Docket No.: AXF-004PC / 139523-5004-WQ decode, via a neural decoder trained by the method of any one of claims 1-16, the bioelectrical signal into an intent and / or brain state of the subject; and a communication effector in communication with the computing device, the communication effector configured to output the decoded intent and / or brain state of the subject.

21. Use of the method of claim 19 for the diagnosis, prognosis, and / or treatment of a patient.

22. Use of the system of claim 20 for the diagnosis, prognosis, and / or treatment of a patient.DBl / 162944529.6 47

Images