Neural monitoring data analysis device

Abstract

Aspects of the embodiments relate to a system configured to perform neuromonitoring data analysis by employing: receiving patient data including patient data including data describing at least one physical stimulus applied to a mammalian subject to responsively generate at least one signal in a plurality of neural structures of the subject's nervous system, and sensor data describing at least one neurophysiological response signal generated in response to the applied physical stimulus, The system is further configured to determine at least one characteristic related to at least one of the patient's plurality of neural structures based on the received patient data describing the at least one physical stimulus and the generated response signal.

Description

[Technical field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims international and / or domestic priority to Israeli patent application 284635, filed July 5, 2021, and to U.S. Provisional Application No. 63 / 218,673, filed July 6, 2021, both of which are incorporated by reference in their entireties. [Background technology]

[0002] The present disclosure generally relates to neuromonitoring, including, for example, intraoperative neuromonitoring. Intraoperative neuromonitoring is essential to avoid or reduce the risk of inadvertent damage to a patient's neural structures when performing surgery on tissue regions such as the spine or brain. During surgery, neurologists with expertise in intraoperative neurological monitoring (IONM) analyze recordings of neurological signals to provide early alerts to the surgeon to prevent or mitigate functional neurological deficits.

[0003] The above discussion is provided as a general overview of the related art in this field and is not to be construed as an admission that any of the information contained therein constitutes prior art to the present patent application.

[0004] The figures illustrate generally, by way of example, but not by way of limitation, various embodiments discussed in the present document.

[0005] For simplicity and clarity of illustration, elements shown in the figures are not necessarily drawn to scale. For example, dimensions of some of the elements may be exaggerated relative to other elements for clarity of presentation. Furthermore, reference numerals may be repeated among the figures to indicate corresponding or similar elements. Reference to previously presented elements is implied without necessarily further reference to the figure or description in which they appear. The figures are as follows: [Brief description of the drawings]

[0006] [Figure 1] FIG. 1 shows a schematic block diagram of a neuromonitoring data analysis device, according to some embodiments. [Figure 2A] FIG. 1 illustrates a general block diagram for creating a machine learning model implemented by a neuromonitoring data analysis device, according to one embodiment. [Figure 2B] FIG. 1 illustrates a more specific block diagram for creating a machine learning model implemented by a neuromonitoring data analysis device, according to one embodiment. [Diagram 3] FIG. 1 is a schematic block diagram of a neuro-monitoring labeling platform, according to some embodiments. [Figure 4] FIG. 1 is a schematic block diagram of a screenshot displaying a plot of a signal analyzed by a neuromonitoring data analysis device, according to some embodiments. [Diagram 5] FIG. 13 is a schematic block diagram of another screenshot displaying a plot of an analyzed signal displayed by the neuromonitoring data analysis device, in accordance with some embodiments. [Figure 6] FIG. 13 is a schematic block diagram of another screenshot displaying a plot of a signal analyzed by the neuromonitoring data analysis device, in accordance with some embodiments. [Figure 7] 1 is a flowchart of a method for monitoring a patient's nervous system, according to some embodiments. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0007] Aspects of the disclosed embodiments relate to automated or semi-automated neuromonitoring based on patient data on sensed and / or measured neurological signals received (directly and / or indirectly) and processed by a neuromonitoring data analysis device. In some embodiments, the device is operable and / or configured to provide an output related to and / or describing the patient's current functional neurological status, optionally together with a clinical interpretation thereof. In addition, the neuromonitoring data analysis device may be operable and / or configured to perform functional mapping of neural structures, neural regions, nerves and / or nerve roots, for example in spinal cord and brain regions. Furthermore, in some embodiments, the neuromonitoring data device may be configured to determine a functional state of a reflective reaction or reflex action in response to (neuro)stimulation. In some examples, output related to the data analysis performed by the apparatus may be provided (e.g., visually and / or audibly displayed) by an output device of a neuromonitoring system employed to monitor (e.g., intraoperatively) the nervous system of a patient undergoing a medical (e.g., surgical) procedure, and / or by an analysis apparatus that, among other things, analyzes data received from the intraoperative neuromonitoring system. Thus, while embodiments may, for example, describe the provision of nervous system monitoring information by a monitoring system and the provision of associated analysis output by an analysis apparatus, this should not be construed as limiting in any way. For example, analysis output (e.g., alerts, flags) generated by the apparatus may be overlaid with the nervous system monitoring information (e.g., signal plots) displayed by the monitoring system. In some examples, the analysis apparatus may simply perform the analysis and generate analysis results (e.g., anomaly classifications) rather than providing an output.

[0008] Some of the patient data may include patient data received pre-operatively and some of the patient data may include patient data received intra-operatively.

[0009] The patient data may relate to the patient's neurological function, for example, may describe at least one physical (e.g., neural) stimulus applied to a mammalian subject (e.g., a human subject) to responsively generate at least one signal in a plurality of neural structures of the patient's nervous system. The patient data may further describe sensor data describing at least one responsive signal generated in response to the applied one or more physical stimuli.

[0010] In some embodiments, the patient data may describe anatomical (e.g., structural) patient information and / or physiological patient information in addition to information of the patient's neurological function. Additional physiological patient information may include, for example, the patient's temperature, weight, blood oxygen saturation, electrocardiogram information, and / or the like.

[0011] The neuromonitoring data analysis device may be configured to determine at least one characteristic related to at least one of the plurality of neural structures based on the received patient data. The characteristic may relate to a neural functional state that may indicate, for example, an abnormality (e.g., a deficit) in the function of the patient's neural structure. In some examples, the output related to the characteristic of the functional state may include an alert, for example, if the deficit satisfies an alert output criterion, e.g., if one or more conditions for outputting an alert are met. In some examples, the alert output criterion may be met if the device identifies that the severity of the deficit exceeds a severity threshold. In some examples, the alert output criterion may be met if the device identifies that the surgically induced deficit is likely to cause a non-recoverable deficit (e.g., a deficit that permanently harms or damages a neural structure), or a non-recoverable deficit exceeding a certain severity, or a recoverable deficit exceeding a certain severity (e.g., a deficit that allows for recovery of a neural structure). The device is operable to distinguish between neural signal abnormalities that are clinically significant or clinically relevant to neural structure damage (e.g., indicative of damage caused to a neural structure) and neural signal abnormalities that are not associated with neural structure damage (e.g., classifying these abnormalities as not clinically relevant to neural structure damage), which may be caused by surgical accessories or instruments used to engage tissue adjacent to or adjacent to a neural structure. The alert may include information regarding the patient's specific clinical condition related to the neural parameter values ​​monitored during the procedure. For example, the device may be configured to provide a clinical interpretation of the patient data, including, for example, the type of neural pathway that was damaged (also injury), the severity of the damage (also injury) and / or resulting deficit, the origin and / or area of ​​the neural damage, and the origin and / or area of ​​the resulting functional deficit. In some embodiments, the device may be configured to identify misconfiguration of the neural monitoring system or setup and alert the user that the detected abnormality (also abnormal parameter value) is due to such misconfiguration.For example, the device may alert the user to incorrect or improper engagement of the sensing and / or stimulation electrodes with the patient.

[0012] It should be noted that in some examples, the terms "damage to neural structures," "nerve damage," and / or "functional deficit" (and grammatical variations thereof) used herein may relate to situations in which the integrity of a neural structure is adversely affected as a result of direct surgical engagement with the corresponding neural structure. In some examples, the device may be configured to distinguish between "spinal cord injury," "nerve damage," and optionally provide a corresponding output.

[0013] It is further noted that the expression "damage" with respect to neural structures due to surgical engagement may relate to non-permanent and / or permanent damage. A neural structure may at least partially or fully recover after injury, e.g., as a result of the data analyzer alerting a medical professional (e.g., a surgeon or neurologist), if, for example, a surgical accessory used by a surgeon is within an acceptable distance from the neural structure that allows for repair of the damage, and / or is within an unacceptable distance from the neural structure that may cause irreparable harm or damage to the neural structure.

[0014] The neuromonitoring system may be configured to monitor naturally occurring and non-naturally occurring neurophysiological signals. For example, the system may be configured to intraoperatively provide at least one physical (e.g., electrical, auditory, and / or visual) stimulus to a patient (also a target individual or mammalian subject) in association with at least one neural signal measurement modality to generate and / or monitor measurable neural activity signals in neural structures of the patient's nervous system.

[0015] The at least one neural measurement modality may relate to monitoring, for example, motor evoked potential (MEP) signals, somatosensory evoked potential (SSEP) signals, reflexes (e.g., H-reflex, bulbocavernosus reflex, blink reflex), autonomic nerve signals (ANS), electromyography (EMG) signals, electroencephalography (EEG) signals, and / or other neural signals.

[0016] For example, during brain surgery, the suction device applies electrical stimulation to determine how close it is to motor pathways including the corticobulbar tract by measuring the corresponding muscle response.

[0017] The device is further configured to analyze and / or process data relating to the neural signals obtained by at least two different neural signal measurement modalities to determine a (e.g., current) functional neural status of the patient, for example, by comparing a plurality of current neural signals to a plurality of normal baseline values ​​of the individual's neural functional signals (e.g., in a situation where the patient is not undergoing a surgical procedure). In some examples, the neural status may be represented by a functional staging parameter value.

[0018] In some embodiments, the device may be configured to process and analyze various neurological stimulation and response signals, and optionally process and analyze additional patient data received at the neuromonitoring data analyzer (e.g., physiological parameter values ​​received pre-operatively and intra-operatively), to provide an analysis result to be presented to a user. In some embodiments, the analysis output may indicate whether neurological function is normal or not. For example, the device may be configured to classify the functionality of a neural structure as normal or abnormal. In some embodiments, the device may be configured to determine whether an abnormal parameter value falls into one of the following two or more categories: a first abnormality indicative of a functional deficit and / or damage of a neural structure, and a second abnormality that is not indicative of a functional deficit and / or damage of a neural structure or is not associated with a functional deficit and / or damage of a neural structure.

[0019] In some embodiments, as briefly described above, the device may be configured to output a clinical interpretation of the monitored neurological signals based on the input data.

[0020] The device may provide and / or modify the current output displayed by the system based on the analysis performed. The analysis may include determining whether criteria are met to display or update a system output regarding the target individual's neurofunctional status and / or changes therein. In some embodiments, the analysis may be provided automatically without user intervention.

[0021] In some embodiments, the user may provide command inputs to the device to perform a selected analysis. In some embodiments, the neuromonitoring data analysis device may be configured to intraoperatively present the user with a list of possible command inputs related to intraoperative neuromonitoring and functional assessment. For example, the user may input (e.g., via the I / O of the system and / or device) the type of surgery, the patient's age, spine, brain, and / or result in different intraoperative neuromonitoring protocols. Additionally, the surgeon may provide inputs regarding different steps in the surgery, and based on this, the device may apply different protocols. For example, motor stimulation by transcranial motor evoked potentials should only be performed when the surgeon gives approval, for example.

[0022] In some embodiments, the device may be configured to automatically apply a series of checks. For example, the device may apply several tests to be performed by the system to check or analyze motor function by the device. If the results of the tests indicate a motor function deficit, the device may analyze the situation and provide one or more outputs indicating why the tests produced results indicative of a motor function deficit, and optionally provide recommendations regarding performing additional tests, including, for example, increasing stimulation. In some examples, the device may autonomously or semi-autonomously (e.g., subject to surgeon approval) cause the neuromonitoring system to apply additional stimulation(s).

[0023] In some embodiments, the device can automatically, autonomously, and continuously determine which analysis outputs should be displayed to the user (e.g., via the system's display) depending on the current situation, and provide recommendations regarding additional stimulations to be performed and their associated parameter values.

[0024] In some embodiments, the analysis may be performed by one or more machine learning models or algorithms, such as an artificial neural network (ANN), to provide an output indicative of the patient's functional neurological status. In some embodiments, the device may be configured to perform the signal analysis using a heuristic model. Furthermore, in some cases, the machine learning model and the heuristic model may be combined into a hybrid model for performing the neuromonitoring signal analysis.

[0025] In some embodiments, the machine learning model may be trained based on multiple analyses previously performed by experts in the field of neuromonitoring. In some examples, the device may be configured to receive data describing neuromonitoring analytics that may be used as input training data for the AI-based machine learning model. In some embodiments, the analytics may be provided intraoperatively.

[0026] A dataset for training a machine learning model can be segmented as follows: 60% training data (on which the ML model is trained), 20% test data (on which ML is not trained), and 20% validation data (this is used to validate parameter values, e.g., to prevent overfitting). The data sets may be randomly separated.

[0027] In some embodiments, the machine learning model may be updated intraoperatively by labeling data associated with intraoperative neuromonitoring under real-world conditions. In some embodiments, data associated with previously performed neuromonitoring analyses may be "mined" or otherwise processed to extract rules and observations that may be used in the process for future neuromonitoring analyses and / or used as training input data to train the neuromonitoring machine learning model. In some embodiments, the machine learning model may include a classifier. In some examples, the classifier may be a regression-based classifier based on an artificial neural network (ANN) or based on a gradient boosting model.

[0028] The gradient boosting algorithm generates a predictive model based on an ensemble of weak predictive models (e.g., decision trees). The model is designed to solve an optimization problem that seeks to minimize the difference between the model predictions on a test dataset and the actual labels in a dataset of labeled data. In some implementations, the actual labels may be provided on the fly by experts, e.g., online or in a real scenario, as outlined herein.

[0029] In some embodiments, the machine learning model may be tested by evaluating the labels generated by a test dataset. Validation measures may include, for example, accuracy, recall, and / or precision with respect to the actual labels in the dataset of labeled data.

[0030] In some embodiments, an output indicative of the target individual's neurological functional status and / or changes therein may be provided (e.g., displayed) in the form of, for example, categorical, ordinal, and / or numerical parameter values. The neurological functional status output may indicate whether the functional status is normal, deficient, impending deficient, improving, or impending improvement. The function of neural structures may be deficient or otherwise altered due to injury and / or due to systemic factors (including, for example, anesthetic agents, blood pressure, body temperature). For example, the system may be configured to identify and alert a user to a systemic physiological condition such as hypotension, and further configured to determine whether detected hypotension is the result of deep anesthesia, hypothermia, and / or technical issues.

[0031] In some embodiments, the neuromonitoring data analysis device may be configured to distinguish between changes indicative of neurological changes that are the result of nerve damage (e.g., indications associated with a neurological deficit), changes that are the result of systemic factors, and / or changes that are due to malfunction and / or misconfiguration of the neuromonitoring system.

[0032] In some examples, the neuromonitoring data analyzer may identify systemic-related signal contributions or variances and subtract the systemic-related signal contributions / variances from the total signal to analyze whether changes in neurofunction are also or exclusively due to systemic factors. In some examples, the neuromonitoring data analyzer may be configured to output warnings regarding signal artifacts introduced due to systemic physiological and / or environmental conditions (electrical artifacts and electromagnetic noise) that may compromise the reliability of neurophysiological monitoring.

[0033] Damage to neural structures can result in neural functional abnormalities, such as partial loss or complete functional nullification. Furthermore, damage can result in permanent or non-permanent functional deficits. Functionality of neural structures can be irreversible, partially reversible, or fully reversible.

[0034] The device may be configured to provide output in real time. It should be noted that the expression "real time" may also encompass the meaning of the term "quasi-real time". As used herein, the expression "real time" generally refers to updating of information essentially as fast as the data is received. In the context of the present disclosure, "real time" is intended to mean that the neurophysiological signals are recorded and processed (e.g., analyzed) at a data rate high enough and a time delay small enough that the output is displayed without any judder, latency or lag that may be noticeable to the user.

[0035] In some embodiments, for example, the device output of the neurological functional state may indicate a "normal" functional state, or may be abnormal, such as, for example, a "deterioration" or "decline" from a normal and / or current functional state, an impending "deterioration" or "decline" from a normal and / or current functional state to a poor functional state, and / or an immediate or impending "loss" of a normal or poor functional state as determined by comparison to multiple normal baseline values ​​of the neurological functional signal. Clearly, additional or alternative labels may be output by the device.

[0036] Neurological function abnormalities (e.g., decline from normal functional state, increased latency, etc.) may be manifested in several ways. For example, considering the evoked potential signal (MEP and / or SSEP) and the baseline response signal, the signal characteristics (e.g., in terms of amplitude, frequency and / or phase) of the response signal may be classified as follows: no response, remaining response, significant decline from baseline response, moderate decline from baseline response, normal, increased response relative to baseline response, increased latency, and / or the like. In some embodiments, detected abnormalities in the MEP and / or SSEP response signal may be identified by the device as a result of technical issues. For example, with respect to the SEP signal and / or EEG performed, bandpass filtering may be performed since relatively low frequency components introduce biases that may distort the analysis of the signal.

[0037] With respect to EEG signals, in some embodiments, the neuromonitoring data analysis device may be configured to analyze EEG signal characteristics in the spectral and time domains, for example, to perform signal power analysis. The signal characteristics may be used as indicators of cortical functional state and its arousal state. In some examples, the EEG analysis may detect systemic physiological conditions that are insufficient for reliable neuromonitoring. In some examples, the EEG analysis may help to assess the extent to which variability in monitored signals, for example, MEPs, SSEPs, and various reflex actions, is due to the influence of systemic factors.

[0038] In some embodiments, the neuromonitoring data analysis device may be configured to process the EMG signal. With respect to the EMG signal, anomalies (e.g., deviations from "quiet") may be characterized, for example, by trains and / or bursts. In some examples, anomalies detected in the EMG signal may be identified by the device as a result of technical problems or from low-level sensory loss.

[0039] In some embodiments, for example, device output of neurological functional status may indicate, for example, an "improvement" to an existing functional status, an imminent "improvement" from an existing functional status, and / or may indicate an immediate or imminent "relapse" or "recovery" of a previously resolved normal or hypo-functional status.

[0040] In some embodiments, the neurological functional state may indicate, for example, the severity of a current deficit in functional state, the likelihood or probability of an immediate condition causing a functional deficit in the nervous system of a particular severity, and / or the like, in a target individual within a particular time period, In some embodiments, the neurological functional state may indicate a degree of adequacy of the neurological functional state, the likelihood or probability of an immediate condition that increases the adequacy of the immediate neurological functional state, and / or the like.

[0041] In some embodiments, the device may provide an output indicating when a neurofunctional status is or is not impaired due to improper neuromonitoring system settings. For example, the device may be configured to distinguish between a baseline signal and a response signal, and further configured to indicate whether a discrepancy between the baseline signal and the response signal is the result of a technical issue, an incorrect neuromonitoring system setting, and / or is due to a surgery-induced physiological neurofunctional deficit.

[0042] In some embodiments, the neuromonitoring data analysis device may be configured to provide an output including a clinical interpretation of the received patient data. In some examples, to provide a clinical interpretation, the functional neurological state and related output may be time-stamped and recorded along with an indication of one or more actions performed on the target individual. For example, the actions of a neurophysiologist to perform intraoperative neuromonitoring (IONM) along with the actions of a surgeon and / or anesthesiologist performed on the patient and / or the position of the patient may be continuously monitored (e.g., sensed and recorded along with a time-stamp) and taken into account by the device to provide an analysis output. For example, the position and duration that the surgeon operatively engages with the target individual or patient may be continuously recorded, analyzed, and optionally output (e.g., displayed) by the device.

[0043] In some embodiments, the device may be configured to determine (e.g., classify) a nerve functional status (e.g., "normal" or "deficient") in relation to a type of nerve structure (e.g., pathway). For example, the device may be configured to identify and indicate that an abnormality is associated with a motor nerve pathway, a sensory nerve pathway, and / or an autonomic nerve pathway, along with an anatomical location or region.

[0044] In some embodiments, the neuromonitoring data analysis device is not only configured and operable to determine a functional neurological status, but may also perform nerve and nerve root mapping, hi some examples, nerve mapping may be associated with determining the functional neurological status.

[0045] For example, the anatomical location or region (and, e.g., associated somatosensory functional deficit) may be identified as, for example, left side, right side, bilateral, high cervical (e.g., cervical spine), low cervical (e.g., lumbar spine), thoracic, upper extremity (right and / or left) and / or lower extremity, along with an associated functional status (no deficit, partial deficit, complete deficit), and / or the like.

[0046] In some embodiments, the device may determine that a particular location may not be associated with a particular functional state, and, optionally, provide an indication to that effect.

[0047] In some embodiments, the device may be configured to map the determined nerve structures. Mapping of nerves and nerve roots may be performed by the device via evoked EMG. For example, in the evoked EMG, nerves and optionally their functional status may be associated with myotomes including, for example, L-deltoid, L-bicep, L-triceps, L-thenar, R-deltoid, R-bicep, R-triceps, R-thenar, or any combination of the above. Mapping related information may be presented to a user, for example, by an I / O device of the system and / or the device.

[0048] In some embodiments, the neuromonitoring data analysis device may be operable to map brain regions, which may be performed by the neuromonitoring data analysis device via direct brain stimulation. The neuromonitoring data analysis device may further provide an output indicative of a specific neurofunctional state of a localized nerve or nerve root.

[0049] Both evoked EMG and direct brain stimulation are performed utilizing electrical stimuli with different or distinct characteristics delivered through a probe or surgical tool.

[0050] In one exemplary scenario, brain regions that are surgically engaged by the surgeon may be automatically mapped (e.g., by the system and / or device) to specific nerve roots and presented to the user along with corresponding functional labels and functional status.

[0051] For example, nerve root mapping in infratentorial brain pathways such as the cauda equina during tumor removal surgery may be performed using probe stimulation and evoked EMG recordings. A neuromonitoring data analyzer analyzes the stimulation response from the corresponding muscle and the stimulation threshold that elicits that response, alerting to the proximity of the brain nerve root.

[0052] In a further example, continuous motor mapping of the cerebral cortex and subcutaneous white matter (supratentorial brain pathways) may be performed by probe stimulation or by continuous stimulation delivered through the slice. The device analyzes the stimulation response and the stimulation threshold that elicits that response, and warns of proximity to motor areas within the surgical field.

[0053] Similarly, with respect to spinal surgery, the device may be configured to automatically identify the type of nerve pathway (e.g., somatosensory, motor sensory or autonomic signals, or any combination of the foregoing) that the surgeon operatively engages during surgery, along with an output indicating the location (e.g., relative to the vertebral position) at which the surgeon engages the spinal nerve.

[0054] In some embodiments, the neuromonitoring data analyzer may be configured to determine the functional status of a reflex arc by sensing a characteristic associated with a reflex response or reflex in response to a stimulus. Exemplary reflexes may include, for example, the H-reflex during thoracic spine surgery, the bulbocavernosus reflex, the blink reflex, and / or the like.

[0055] In some embodiments, the device may be configured to provide an output that indicates executable instructions on how to take possible steps to maintain a current (e.g., normal) neurological functional state, reverse deterioration of neurological function, or prevent further deterioration.

[0056] The executable instructions may include, for example, one or more follow-up tests and / or adaptation of parameter values ​​to improve or optimize the patient's physiological overall condition for neurophysiological monitoring.

[0057] In some embodiments, the output may relate to instructions for validating the system output.

[0058] In some embodiments, the neuromonitoring data analysis device may be configured to provide decision support when an abnormality is detected and identified (e.g., classified). For example, the device may provide an output to a user indicating actionable suggestions or instructions to the user on how to verify, overcome, or correct the abnormality to return to a normal functional state. For example, the neuromonitoring device output may include instructions or suggestions to elicit additional motor evoked potentials, instructions or suggestions to elicit additional somatosensory evoked potentials, instructions or suggestions to modify stimulation-related properties (e.g., intensity, polarity) of the additional motor evoked potentials, instructions or suggestions to modify stimulation intensity of the additional somatosensory evoked potentials, reflex stimulation, and peripheral nerve stimulation such as train of four (TOF), and / or the like.

[0059] In some embodiments, the output may include instructions or suggestions for updating and / or modifying recording parameters, instructions or suggestions related to malfunction identification checks and / or configuration checks of the neuromonitoring system, ways to overcome malfunctions and / or misconfigurations of the system, including instructions or suggestions for automated, semi-automated and / or manual troubleshooting of neuromonitoring system misconfigurations. The output may include, for example, (e.g., troubleshooting) instructions or suggestions such as updating recording parameters, system malfunction checks, impedance checks, anesthesia parameter checks, electrode contact checks, instructions to check physiological patient parameters, instructions or suggestions to check patient position, instructions or suggestions to put the procedure on hold, or any combination of the foregoing. In some scenarios, the output provided by the device may indicate that the user does not undertake any additional steps.

[0060] A user of the neuromonitoring data analysis device may provide command inputs to the device and / or neuromonitoring system, for example, according to received system outputs and, as described above, in a real-life scenario environment, according to information regarding the surgical stage.

[0061] In some embodiments, output related to (e.g., describing) neurofunctional status may be provided in conjunction with output describing structural status, e.g., via medical images obtained by one or more imaging modalities including, for example, X-ray (e.g., including computed tomography) based imaging techniques, nuclear imaging techniques, MRI imaging techniques, and / or ultrasound imaging techniques.

[0062] In some embodiments, the neuromonitoring data analyzer is configured (e.g., providing a human-machine interface) to train the machine learning mode through offline labeling of neural data recorded during previously performed and completed surgeries, e.g., retrospectively, under conditions that seek to mimic actual clinical conditions.

[0063] In some embodiments, a labeling platform is provided that is configured to enable offline (e.g., retrospective) labeling of various neurophysiological data, and / or online or “real-time” labeling during simulated or actual surgery, for the purposes of training machine learning models.

[0064] In some embodiments, the labeling platform may be configured as an add-on module of a neuromonitoring system used to receive and record patient data. In some examples, the neuromonitoring data analyzer may be included in the neuromonitoring system. In some embodiments, a subsystem may include both the neuromonitoring data analyzer and the labeling platform. In some examples, the neuromonitoring data analyzer may include the labeling platform. In some further examples, the labeling platform may include the neuromonitoring data analyzer. In this specification, the neuromonitoring data analyzer and the labeling platform are considered as separate entities, without being limited thereto, merely for the sake of simplifying the following discussion.

[0065] The online or "real-time" labeling of various neurophysiological data related to clinical events in a real scenario environment allows investigating the intraoperative situation in real time (e.g., by considering environmental parameters such as noise or sounds made by instruments, questions to the surgeon, questions to the anesthesiologist, giving instructions to the surgeon and / or the anesthesiologist, etc.) to arrive at the corresponding labeling of the neurophysiological data received by the device. Thus, the labeling platform may enable "open" machine learning, such that the machine learning model can be adapted continuously. In contrast, in a closed machine learning process, the machine learning model is fixed and does not change during the use of the neuromonitoring data analysis device. In some embodiments, the machine learning model of the neuromonitoring data analysis device may be updated "on the fly" by using the online labeling platform disclosed herein.

[0066] In some embodiments, the label provided online may be compared to an output label generated by the device in a retrospective environment or offline scenario.

[0067] Training machine learning models in an online environment (e.g., intraoperatively) may result in different and more realistic labeling of neurophysiological data compared to offline labeling environments, as exemplified herein, and may result in better model training.

[0068] Example 1: Online vs. Offline Labeling In one exemplary scenario, a surgeon positions the patient several minutes after anesthesia is induced and the neurophysiological setup for decompressive cervical surgery is completed. At these moments, a significant deterioration of the neurophysiological signal is recorded.

[0069] A neurophysiologist (or "expert") may be asked to distinguish between two events that may be responsible for this deterioration: Labeling Option 1A: Benign systemic effects as a result of the anesthesiologist changing the anesthetic dose to establish adequate depth of anesthesia. Labeling Option 1B: Progressive spinal cord injury as a result of increased pressure on the spinal cord in the new head position.

[0070] To reach a decision as to which of these two events applies, in an online labeling scenario, the neurophysiologist may question the anesthesiologist and ask what he or she has done in the past few minutes to determine whether or not a neurophysiological deterioration has occurred due to deepening anesthesia. In the latter case, the increase in neurological deficit is more likely due to pressure on the spinal cord than due to an increase in the depth of anesthesia.

[0071] Obviously, in an offline labeling environment, it is not possible to question the anesthesiologist, and the labeling associated with the observed deterioration of the neurophysiological signal is done by "guessing" the most likely scenario, for example by moving back and forth in time along the length of previously recorded neurophysiological data for corresponding or similar surgeries, to mimic the real scenario. However, such "guessed" or retrospective labeling may not, of course, reflect the real clinical situation that existed when the surgery was performed, which may lead to erroneous labeling and thus suboptimal or erroneous training of the machine learning model. In contrast, an online labeling environment more accurately reflects a given real clinical situation. Therefore, online labeling may be less prone to bias and error.

[0072] Example 2: Online vs. Offline Labeling In a further embodiment, online labeling allows for follow-up testing and modification of test parameters to confirm or reject a hypothesis In contrast, when performing offline labeling, testing of hypotheses is not possible.

[0073] In the context of Example 1, the neurophysiologist can recommend to the operating room staff to reduce the depth of anesthesia and to reposition the patient if neurological improvement is not evident, thus making it possible to reliably distinguish between two (e.g., possible or probable) causes of decreased neurophysiological function.

[0074] Additional steps that the neurophysiologist performs as part of the situation include: Option 2A: Technical testing of the neuromonitoring system and setup to rule out technical failures, Option 2B: Increasing the intensity of stimulation or changing the polarity or location of stimulation; Option 2C: Increase the frequency of testing and evaluate trends over several minutes.

[0075] All the above options allow the neurophysiologist to arrive at a more accurate diagnosis, thus resulting in both better treatment and more accurate labelling.

[0076] Example 3 - Improving or optimizing test sets for validation of algorithms / machine learning models To objectively examine the performance of the neuromonitoring data analyzer, it may be necessary for a neurophysiologist to monitor the surgery (real or simulated) using the labeling platform to label the monitoring data under real conditions, in real time or substantially real time, during the surgery, so that the labeling input provided closely reflects the work of the neurophysiologist when he has all the information available in the operating room at his disposal.

[0077] At the same time, the neuromonitoring data analysis device receives patient data from the same surgery and outputs labels according to the algorithms and / or trained machine learning models, independent of the labeling provided to the labeling platform. The labels output by the neuromonitoring data analysis system can be compared with the labels provided to the labeling platform by the physician, which allows the physician's labels to be compared with the labels output by the neuromonitoring data analysis device. This comparison allows the performance of the neuromonitoring data analysis device to be analyzed in a reliable manner.

[0078] In some embodiments, the neuromonitoring data analyzer may be configured as an add-on (e.g., software and / or hardware implemented) module of a neuromonitoring system used to receive and record patient data. In some examples, the neuromonitoring data analyzer may be included in the neuromonitoring system. In some further examples, the neuromonitoring data analyzer may be external to the (intra-operative) neuromonitoring system. In some examples, all or some of the portions, components, and / or modules of the analyzer may be implemented by the monitoring system. In some examples, none of the portions, components, and / or modules of the analyzer may be implemented by the monitoring system.

[0079] Referring to FIG. 1 , the neuromonitoring data analysis device 1000 may include an I / O device 1100, a processor 1200, and a memory 1300.

[0080] In some example implementations, the neuromonitoring data analysis apparatus may provide output to its user via I / O device 1100 with one or more output devices. The one or more output devices may include, for example, devices configured to convert electrical signals into outputs that a human can sense as output, such as sound, light, and / or touch. The output devices may include a display screen and / or audio output device(s), such as, for example, speaker(s) and / or earphones.

[0081] The I / O device 1100 may further include one or more input devices configured to receive any type of data and / or information, for example, by converting machine-generated signals and / or human-generated signals, such as physical movements, physical contact or physical pressure, and / or the like, into electrical signals as input data to the computing system. Examples of such input devices include touch screens, microphones, hand gesture tracking devices, handheld pointing devices (e.g., computer mice, styluses), and / or the like.

[0082] The I / O device 1100 may be used to access data and / or information generated by the neuromonitoring data analysis apparatus 1000 and / or to provide inputs including, for example, control commands, operating parameters, queries, and / or the like. For example, the I / O device 1100 may enable a user of the neuromonitoring system to apply one or more physical stimuli to a patient, for example, via the output electrodes 1102. The I / O device 1100 may further be configured to sense signals generated in response to applying non-electrical stimuli to the patient, including, for example, sounds for auditory brainstem evoked potentials, light flashes for visual evoked potentials, thermal stimuli, and / or tactile stimuli, via the sensors 1104. The sensed response signals may include, for example, SSEPs and / or MEPs, etc. The output electrodes 1102 may include one or more electrodes. The sensors 1104 may include one or more sensors including, for example, electrodes.

[0083] The neuromonitoring data analysis device 1000 may further include a processor 1200 and a memory 1300 configured to store data 1310 (e.g., patient data) and algorithmic code and / or machine learning (ML) models 1320. The processor 1200 may be configured to execute the algorithmic code and / or apply the machine learning (ML) models 1320 to the processing of the data 1310, thereby implementing an intraoperative neuromonitoring data analysis (INDA) engine 1400. The INDA engine 1400 may be configured to characterize a patient's neurofunctional status and / or enable online labeling, for example, as outlined herein.

[0084] The term "processor" as used herein may additionally or alternatively refer to a controller. The processor 1200 may be implemented by various types of processor devices and / or processor architectures, including, for example, embedded processors, communications processors, graphics processing unit (GPU) accelerated computing, soft-core processors, quantum-based processors, and / or general-purpose processors.

[0085] The memory 1300 may be implemented by various types of memory including transactional memory and / or long-term memory functions, and may function as file storage, document storage, program storage, or as working memory. The latter may be in the form of, for example, static random access memory (SRAM), dynamic random access memory (DRAM), read-only memory (ROM), cache, and / or flash memory. As working memory, the memory 1300 may include, for example, time-based instructions and / or non-time-based instructions. As long-term memory, the memory 1300 may include, for example, volatile or non-volatile computer storage media, hard disk drives, solid-state drives, magnetic storage media, flash memory, and / or other storage facilities. The hardware memory facilities may store fixed sets of information (e.g., software code), including, for example, but not limited to, files, programs, applications, source code, object code, data, and / or the like.

[0086] The neuromonitoring data analysis device 1000 may further comprise at least one communication module 1500 configured to enable wired and / or wireless communication between various components and / or modules of the device, which may communicate with each other via one or more communication buses (not shown), signal lines (not shown) and / or network infrastructures. The communication module 1500 may be configured to enable communication using one or more communication formats, protocols and / or technologies, such as, for example, Internet communication, optical or RF communication, telephone-based communication technologies, and / or the like. In some embodiments, the communication module 1500 may include I / O device drivers (not shown) and network interface drivers (not shown) to enable transmission and / or reception of data over a network. The device drivers may, for example, interface with a keypad or to a USB port. The network interface driver may, for example, implement protocols for the Internet, or an intranet, a wide area network (WAN), a local area network (LAN), using, for example, a wireless local area network (WLAN), a metropolitan area network (MAN), a personal area network (PAN), an extranet, 2G, 3G, 3.5G, 4G, 5G, 6G mobile networks, 3GPP, LTE, LTE Advanced, Bluetooth® (e.g., Bluetooth Smart), ZigBee™, Near Field Communication (NFC), and / or any other current or future communications network, standard, and / or system.

[0087] The neuromonitoring data analysis device 1000 may further include a power module 1600 for providing power to various components and / or modules and / or subsystems of the device. The power module 1600 may include an internal power source (e.g., a rechargeable battery) and / or an interface to allow connection to an external power source.

[0088] It will be understood that separate hardware components, such as a processor and / or memory, may be assigned to each component and / or module of the neuro-monitoring data analysis device 1000. However, for simplicity and without being limiting, the description and claims may refer to a single module and / or component. For example, the processor 1200 may be implemented by several processors, but the following description refers to the processor 1200 as the component that performs all necessary processing functions of the neuro-monitoring data analysis device 1000.

[0089] The functionality of the neuromonitoring data analysis device 1000 may be implemented, fully or partially, by a multi-function mobile communication device, also known as a "smart phone," a mobile or portable device, a non-mobile or non-portable device, a digital video camera, a personal computer, a laptop computer, a tablet computer, a server (which may be associated with one or more servers or storage systems and / or services associated with a business or corporate entity, including, for example, a file hosting service, a cloud storage service, an online file storage provider, a peer-to-peer file storage or hosting service, and / or a cyberlocker), a personal digital assistant, a workstation, a wearable device, a handheld computer, a notebook computer, a vehicular device, a non-vehicular device, and / or a stationary device. For example, some of the INDA engine 1400 functionality may be implemented by devices, apparatus, and / or systems located on-premise (e.g., in a hospital or other clinical facility) and some off-premise (e.g., in the "cloud"). Alternative configurations may also be envisioned.

[0090] Various approaches may be employed to implement the machine learning models used by the neuromonitoring data analysis device 1000 of the neuromonitoring system. For example, a machine learning (ML) model may include multiple ML sub-models. In an embodiment, the ML sub-model may be associated with a particular level of the ML model. In some cases, multiple ML models may be organized in a hierarchical structure, for example, as described herein in conjunction with FIG. 2A and FIG. 2B. In some embodiments, the ML sub-models may be feature-based, optionally requiring signal pre-processing to perform feature extraction. In some other embodiments, the ML sub-models may be based on artificial neural networks (ANNs).

[0091] For example, as shown generally in FIG. 2A, at least one first ML sub-model (e.g., channel model 3010) may be created and applied to analyze selected ones of the input signals 2000 of a signal modality.

[0092] A signal modality may relate to a signal category including, for example, one of MEP, SSEP, EMG, EEG, and / or the like, and different types of input signals may relate to corresponding different channels of the same signal modality. For example, different types or channels of MEP modality signals may include, for example, deltoid left, deltoid right, biceps left / right, triceps left / right, thenar left, thenar right, tibialis anterior left / right, quadriceps-tibialis anterior left / right.

[0093] To distinguish between different channel sub-models 2010 used to analyze the signals of each channel of a signal modality, a capital alphabetic letter is added after the number, e.g., first channel sub-model 2010A, second channel sub-model 2010B, third channel sub-model 2010C, etc. However, when there is no need to specifically distinguish between each channel model, they are collectively referred to simply as channel sub-models 2010.

[0094] In some embodiments, at least one first ML sub-model (or “at least one first channel sub-model 2010A”) may be applied to analyze a first MEP input signal, at least one second channel sub-model 2010B may be applied to analyze a second MEP input signal, etc.

[0095] As a second level, at least one modality sub-model 2020 may be used to analyze signals of a particular modality (e.g., MEP modality, SSEP modality) as a whole. For example, the MEP modality sub-model may be used to analyze multiple signals of a selected modality to provide a modality prediction for the MEP modality. The at least one modality sub-model 2020 may receive predictions (e.g., labels) generated by at least one first ML sub-model or channel sub-model 2010.

[0096] As a third level, at least one clinical diagnostic sub-model 2500 may be used. The at least one clinical diagnostic sub-model 2500 may receive predictions generated by at least one modality sub-model 3030 of the second level and provide a clinical output 2700 based thereon.

[0097] In some embodiments, at least one modality sub-model 2020 and / or clinical diagnosis sub-model 2500 may be supplemented with clinical and / or surgical data 2600 to analyze modality predictions and generate clinical output 2700.

[0098] 2B, at a first level, a particular signal of a selected modality (e.g., MEP signal 2100, SSEP signal 2200, EEG signal 2300, and / or EMG signal 2400) may be received by a corresponding channel sub-model. For example, the motor channel sub-model 2110 may receive the MEP signal 2100, and the sensor channel sub-model 2220 may receive the SSEP signal 2200.

[0099] In some embodiments, the input signals may be pre-processed for the purpose of feature extraction. For example, the MEP signal 2100 may be pre-processed to perform feature extraction on at least one MEP signal (block 2105), and the SSEP signal 2200 may be pre-processed to perform feature extraction on at least one SSEP signal (block 2205). As mentioned above, ML models such as the motion channel sub-models do not necessarily have to be feature extraction based, but may be based on deep learning ML models such as artificial neural networks.

[0100] Further, at the second level, the analysis output generated by the motion channel submodel 2110 is provided to the motion modality submodel 2140 for further analysis thereby, and the analysis output generated by the sensor channel submodel 2210 is provided to the sensor modality submodel 2240 for further analysis thereby. Optionally, the input to the ML submodel of the second level is the output of the ML model of the first level. Based on the input of the first ML model, the ML submodel of the second level determines, for example, whether there is a significant change in the signal that requires expert attention.

[0101] The analysis outputs generated by the motor modality sub-model 2140 and the sensor modality sub-model 2240 are provided to the clinical diagnosis sub-model 2500 for further analysis, resulting in a clinical output representing a clinical interpretation of the data provided to the various ML sub-models.

[0102] In some embodiments, clinical and / or surgical data may be provided to the modality sub-model (e.g., MEP and / or SSEP) and / or clinical diagnosis sub-model 2500 to generate clinical output 2700. In some embodiments, data regarding EEG signal 2300 and / or EMG signal 2400 may be provided to clinical diagnosis model 2500. In some embodiments, EEG signal 2300 and EMG signal 2400 may be pre-processed for the purpose of feature extraction (blocks 2305 and 2405, respectively).

[0103] The following three examples relate to scenarios in which the device correctly identifies an abnormality in the signal as being clinically significant, ie, as damage to neural structures.

[0104] In an exemplary scenario, the device identifies impairment of corticospinal tract function during cervical laminectomy, also known as central spinal cord syndrome. The device may identify a significant decrease in motor response to the point of loss of response from the right hand and leg muscles, and optionally classify the abnormality as central spinal cord syndrome.

[0105] In an exemplary scenario, impairment of corticospinal tract function during cervical laminectomy, also known as Brown-Séquard syndrome, is identified by the device. A significant decrease in motor response from the hand muscles to the point of bilateral loss of response is identified by the device and, optionally, classified as Brown-Séquard syndrome.

[0106] In an exemplary scenario, the device identifies impairment of C5 / C6 nerve function during a cervical laminectomy. The data describes the signal displayed by the monitoring system, indicating a significant decrease in motor response from the right biceps brachii. A significant baseline response and a "current" response may be displayed by the monitoring system using different colors.

[0107] The following two examples concern scenarios in which the abnormalities in the signal relate to a reduction in motor response caused by a systemic effect rather than to neural structural damage, and therefore the detected abnormalities are not clinically significant with respect to neural structural damage.

[0108] An exemplary scenario concerns impaired corticospinal tract function demonstrated from baseline testing of the right leg: A marked reduction in motor responses in the hand muscles and moderate reductions in the arm and leg muscles are displayed, which are due to anesthetic drugs and changes in blood pressure.

[0109] Another exemplary scenario concerns an impairment of corticospinal tract function demonstrated from a baseline test of the right leg. A significant decrease in motor response in the left hand and left leg muscles appears due to anesthetic drugs and blood pressure changes. The right leg channel displays no baseline and "current" response, while the left channel displays a significant decrease in the "current" response. Thus, the decrease in motor response represented by the signal displayed by the monitoring system is due to systemic effects and pre-operative dysfunction.

[0110] The following four examples relate to data describing signals indicative of neural structure damage with associated false positive alerts provided by prior art neuromonitoring data analyzers, for example with respect to cervical spine immobilization.

[0111] These four examples below demonstrate that prior art devices often misidentify or misclassify clinically insignificant abnormalities in a signal (abnormalities not associated with damage to neural structures) as clinically significant abnormalities (e.g., abnormalities associated with damage to neural structures). This is because, for example, prior art devices may use rule-based criteria such as "peak-to-peak" of individual signals and rely solely on comparison of selected "current" signals to corresponding baseline signals, area under the curve models, or the like. In contrast, embodiments of the analysis device presented herein consider (e.g., collectively analyze) data describing multiple different or separate signals sensed by a monitoring system, optionally in combination with rule-based approaches, for example, by trained machine learning models, to accurately identify damage.

[0112] In one example scenario, the prior art device displays the baseline motor response in one color and displays the "current" motor response using a second color different from the first color. The prior art device outputs "alerts" or "flags" (e.g., "red flags" at the corresponding timestamp or portion of the signal plot displayed by the neuromonitoring system) of insignificant changes that are erroneously identified by the prior art device as clinically significant changes in the signal related to neural structure damage. Additionally, the prior art analysis device erroneously identifies artifacts caused by technical stimuli as responses.

[0113] In a further example scenario, the baseline (first color) and "current" (second color) motor responses may be displayed. Prior art analyzers output alerts ("red flags") in relation to insignificant changes (e.g., clinically insignificant changes with respect to neural structure damage) presumably due to changes in general anesthesia. Prior art analyzers may also erroneously identify artifacts caused by technical stimuli as responses.

[0114] In a further example scenario, in prior art analyzers, an artifact caused by a technical stimulus at baseline is erroneously identified or measured as a response, and its disappearance in the "current" response results in a false positive alert.

[0115] In another example scenario, prior art analyzers flag or generate alerts for insignificant changes in somatosensory responses (eg, changes that are not clinically significant with respect to neural structure damage).

[0116] Further, referring now to FIG. 3, the neuromonitoring labeling platform 1002 may be configured to have a labeling interface 1700 that enables online or on-the-fly (e.g., real or simulated intraoperative) labeling capabilities.

[0117] For example, I / O device 1100 may also include one or more device interfaces (e.g., one or more touch screens) that allow a practitioner to perform "on-the-fly" or real-time labeling of neuromonitoring data in an actual surgical environment. For example, I / O device 1100 may display user selectable fields that allow a practitioner to perform online labeling of displayed signals and / or data, e.g., as shown generally with respect to Figures 4-6, which generally illustrate various fields that may be displayed by I / O device 1100.

[0118] 4 shows diagrammatically various fields that may be displayed by the I / O device 1100 if the professional selects a particular signal or channel representing, for example, "Left-DI signal and baseline." Note that in some embodiments, the device may not include a display, but may be configured for the output of a system configured to generate various signals. The device may be in operative communication with the system to perform analysis of the data describing the signals that are displayed as graphs by the system.

[0119] The various channel fields show corresponding signal plots such as the measured signal and associated baseline. However, solely for ease of explanation, the plots have been removed and replaced with corresponding descriptions of the plots. For example, field 4010 is annotated as "Left-DI signal and baseline."

[0120] Depending on the data or signal field selected by the expert, different selectable labeling options are automatically displayed by the I / O device 1100 .

[0121] For example, as shown diagrammatically in Fig. 4, once a particular channel, such as Left-Di signal and baseline 4010, is selected by the expert, the I / O device 1100 displays corresponding channel labeling options 4150A, which are expert-selectable fields annotated with various labels associated with different channel-related scenarios, such as "high response", "normal response", "moderate decline", "significant decline", "residual response", and "no response". The selection made is indicated diagrammatically by a bold rectangle 4001A. Thus, based on the analysis of the plurality of signals, the analyzer may be configured to classify at least one of the plurality of signals into a corresponding category and, optionally, provide a corresponding output.

[0122] In another embodiment, as shown diagrammatically in Fig. 5, once the expert selects all signals of a particular modality (e.g., MEP signals 4010-4110), the I / O device 1100 displays the corresponding modality labeling options 4150B annotated with different labels associated with different modality-related scenarios, which may for example be one of "normal" and "severely degraded". The selection made is indicated diagrammatically by a bold rectangle 4001B. Thus, based on the analysis of the plurality of signals, the analyzer may be configured to classify at least one of the plurality of signals into a corresponding category and, optionally, provide a corresponding output.

[0123] In yet further examples, as shown diagrammatically in FIG. 6, once the expert selects all signals of all modalities, the I / O device 1100 displays corresponding clinical interpretation labeling options 4150C annotated with different labels associated with different clinical scenarios, which may be, for example, one of “normal”, “spinal cord injury”, “nerve injury”, and “systemic”. Selection of all signals is diagrammatically indicated by a bold rectangle 4001C. Correspondingly, in some examples, the analyzer may classify the status of at least one signal of the plurality of signals into one of the following categories, namely “normal”, “spinal cord injury”, “nerve injury”, and “systemic”, based on a collective analysis of the plurality of signals associated with at least two neural structures, and optionally provide a corresponding output.

[0124] In some embodiments, the I / O device 1100 may display a checkbox annotated with “Technical Issues” to indicate whether one of the selected scenarios is related to a technical issue.

[0125] Additional fields that may be displayed by the I / O device 1100 may include, for example, a title and time field 4105 displaying information about the current time and the displayed modality, an EEG signal field 4120, an EEG spectrogram field 4130, a selected label display field 4140, a general comments field 4170, a page properties configuration field 4180, and a notes field 4190 for entering notes about the selected label.

[0126] With further reference to Figure 7, a method for analyzing data describing signals in the nervous system of a patient undergoing a medical procedure may, in some embodiments, include receiving (block 7100) data describing a plurality of neurophysiological response signals generated in at least two (e.g., separate or different) neural structures (e.g., associated with different pathways) of the patient as a result of administering a plurality of physical stimuli to the patient.

[0127] In some embodiments, the method may further include identifying, based on the received data, an abnormal event indicative of damage to at least one of the at least two neural structures (block 7200).

[0128] In some embodiments, the method may include providing a first output related to the identified event (block 7300).

[0129] Additional Examples: Example 1 relates to a neuromonitoring data analysis device configured to analyze neuromonitoring signals of a subject's nervous system, the device comprising at least one processor and at least one memory configured to store data and software code portions executable by the at least one processor to perform the following operations: receive patient data including data describing at least one physical stimulus applied to a mammalian subject to responsively generate at least one signal in a plurality of neural structures of the subject's nervous system, and sensor data describing at least one neurophysiological response signal generated in response to the applied physical stimulus; and determine at least one characteristic of at least one of the patient's plurality of neural structures based on the received patient data describing the at least one physical stimulus and the generated response signal. Optionally, a machine learning model including multiple levels of hierarchically arranged machine learning sub-models is used to analyze the received patient data.

[0130] Example 2 includes the subject matter of example 1, optionally wherein at least one characteristic is related to a functional state of the neural structure.

[0131] Example 3 includes the subject matter of example 1 and / or example 2, optionally wherein the neural structures of the plurality of neural structures include neural pathways and / or receptors.

[0132] Example 4 includes the subject matter of any one or more of examples 1-3, optionally including at least one classifier for determining the neurological functional state.

[0133] Example 5 includes the subject matter of example 4, optionally including employing a machine learning model including, for example, gradient boosting and / or an artificial neural network.

[0134] Example 6 includes the subject matter of Example 5, optionally including an interface that enables an expert, during surgery, to associate the received patient data describing the at least one physical stimulus and the generated response signal with one or more labels for training a machine learning model, the one or more labels relating to characteristics of the patient's nervous system.

[0135] Example 7 includes the subject matter of Example 4 and / or Example 5, and is optionally configured to classify the neurological functional status into one of the following: "normal", "decreased", or "absent".

[0136] Example 8 includes the subject matter of any one or more of Examples 1-7, and optionally, the at least one physical stimulus is associated with one of the following: evoked potentials, reflexes, spontaneous potentials, or any combination of the above.

[0137] Example 9 includes the subject matter of any one or more of Examples 1-8, and optionally, the memory is configured to receive patient data further including one of the following: clinical patient data, demographic patient data, anesthesia data, physiological patient data, surgical data, baseline motor evoked potential signal data, baseline EMG signals, baseline EEG signals, baseline somatosensory evoked potential signal data, reflex actions, or any combination thereof.

[0138] Example 10 includes the subject matter of any one or more of Examples 1-9, and optionally, the information further includes clinical information describing one of the following: a systemic physiological condition, a system malfunction, or both.

[0139] Example 11 includes the subject matter of any one or more of Examples 1-10, and optionally further configured to characterize the pathway injury as associated with one of the following: motor injury including a motor injury impact location, somatosensory injury including a sensory impact location, spinal cord pathway injury severity including complete or incomplete injury, or any combination thereof.

[0140] Example 12 includes the subject matter of any one or more of Examples 1-11, optionally further configured to characterize the damage to at least one neural structure as associated with one of the following: myotome damage, neural damage severity, including complete damage, or incomplete damage.

[0141] Example 13 includes the subject matter of any one or more of examples 1-12, and optionally, at least one characteristic is provided to the user in real-time or near real-time.

[0142] Example 14 includes the subject matter of any one of Examples 1 to 13, and is optionally further configured to provide a surgical recommendation output including one of the following: elicit additional motor evoked potentials, elicit additional somatosensory evoked potentials, change the stimulation intensity of the additional motor evoked potentials, change the stimulation intensity of the additional somatosensory evoked potentials, update recording parameters, check for malfunction of the neuromonitoring system, check impedance, check anesthesia parameters, perform peripheral nerve stimulation, check physiological parameters, check patient position, maintain surgery, change and / or check electrode positioning, position the patient, or any combination of the foregoing.

[0143] Example 15 includes the subject matter of any one or more of examples 1-14, and optionally, the device is configured for intraoperative use.

[0144] Example 16 includes the subject matter of any one or more of examples 4-15, and optionally, the machine learning model includes a plurality of machine learning sub-models.

[0145] Example 17 includes the subject matter of Example 7, and optionally, a first level machine learning sub-model of the hierarchically arranged machine learning sub-models is configured to analyze a plurality of channels of signals associated with a signal modality to generate an analysis output for each of a plurality of channels.

[0146] Example 18 includes the subject matter of Example 17, wherein a second level machine learning submodel of the hierarchically arranged machine learning submodels is configured to receive a plurality of per-channel analysis outputs, and a second level machine learning submodule is configured to analyze the plurality of received per-channel analysis outputs of a signal modality to provide an analysis output describing the modality.

[0147] Example 19 includes the subject matter of Example 18, and optionally, a machine learning submodule of a third level of the hierarchically arranged submodules receives an analysis output describing the modality, and the third level machine learning submodule is configured to analyze the output received from the second submodule to generate an output describing a clinical interpretation of the received patient data.

[0148] Example 20 includes a method for performing neuromonitoring data analysis, the method comprising:

[0149] receiving patient data,

[0150] receiving patient data, the patient data including data describing at least one physical stimulus applied to a mammalian subject to responsively generate at least one signal in a plurality of neural structures of the subject's nervous system, and sensor data describing at least one neurophysiological response signal generated in response to the applied physical stimulus;

[0151] and determining at least one characteristic related to at least one of the patient's multiple neural structures based on the received patient data describing the at least one physical stimulus and the generated response signal. Optionally, the determining includes employing a machine learning model having multiple levels of hierarchically arranged machine learning sub-models to analyze the received patient data.

[0152] Example 21 includes the subject matter of Example 20, optionally wherein at least one characteristic is related to a functional state of a neural structure.

[0153] Example 22 includes the subject matter of Example 20 and / or Example 21, optionally, where determining includes classifying the neurological functional state.

[0154] Example 23 includes the subject matter of any one or more of examples 21-23, and optionally, determining includes employing a machine learning model to analyze the patient data.

[0155] Example 24 includes the subject matter of any one or more of Examples 22-23, and optionally classifying the neurological functional status into one of the following: "normal," "decreased," or "absent."

[0156] Example 25 includes any one or more of the subject matter of Examples 20 to 24, and optionally further includes the following:

[0157] Eliciting additional motor evoked potentials,

[0158] Eliciting additional somatosensory evoked potentials,

[0159] Varying the stimulation intensity for additional motor evoked potentials,

[0160] Varying the stimulation intensity for additional somatosensory evoked potentials,

[0161] Update recording parameters;

[0162] Check for malfunctions in the neuromonitoring system,

[0163] Check the impedance,

[0164] Checking anesthesia parameters,

[0165] Peripheral nerve stimulation,

[0166] Checking physiological parameters,

[0167] Check the patient position,

[0168] Maintaining surgery

[0169] Electrode positioning,

[0170] Positioning the patient, or

[0171] and providing a surgical recommendation output comprising one of: any combination of the foregoing.

[0172] Example 26 includes the subject matter of any one or more of Examples 20-25, optionally including performing data analysis of the patient data intraoperatively.

[0173] Example 27 includes the subject matter of any one or more of examples 23-26, optionally, the machine learning model includes a plurality of machine learning sub-models.

[0174] Example 28 includes the subject matter of Example 27, and optionally analyzes a plurality of channels of signals associated with a signal modality to generate analysis outputs for each of a plurality of channels by a first level machine learning sub-model of the hierarchically arranged sub-models.

[0175] Example 29 includes the subject matter of example 28, optionally further comprising receiving analysis outputs for each of the plurality of channels by a machine learning submodule of a second level of the hierarchically arranged submodels;

[0176] Analyzing, by a second level machine learning module, the analysis output for each of the plurality of received channels of a signal modality to provide an analysis output describing the modality.

[0177] Example 30 relates to the subject matter of example 29 and, optionally, to a machine learning submodule at a third level of the hierarchically arranged submodules, receiving an analysis output describing the modality;

[0178] and analyzing, by a third level machine learning sub-module, the output received from the second sub-module to generate an output describing a clinical interpretation of the received patient data.

[0179] Example 31 includes a method that includes receiving patient data at a neuromonitoring labeling platform associated with the neuromonitoring system during execution of the neuromonitoring system, displaying information related to the received patient data on a display of an input / output (I / O) device of the labeling platform, and labeling the patient data during execution using the I / O device, where one or more labels are associated with characteristics of the patient's nervous system.

[0180] Example 32 includes the subject matter of Example 31, and optionally, the labeling is performed intraoperatively for labeling patient data received from the neuromonitoring system to train a machine learning model of the neuromonitoring data analysis device.

[0181] Example 33 includes the subject matter of example 31 and / or example 32, and optionally, the labeling includes selecting a portion of the displayed information by dragging one or more windows on a display of the I / O device.

[0182] Example 34 includes the subject matter of any one or more of Examples 31-33, and optionally, the labeling includes selecting a portion of the displayed information by touch pressing a display of the I / O device.

[0183] Example 35 includes the subject matter of any one or more of Examples 31-34, and optionally, the labeling includes selecting a portion of the displayed information by providing voice input to the I / O device.

[0184] Example 36 includes the subject matter of any one or more of Examples 31-35, and optionally, the labeling includes selecting a portion of the displayed information by providing gaze-based input to the I / O device.

[0185] Example 37 includes the subject matter of any one or more of Examples 31-36, and optionally, the labeling includes selecting a portion of the displayed information based on an association between the selected information and a corresponding hierarchical level of the selected information in a machine learning model that includes multiple levels of hierarchically arranged machine learning sub-models.

[0186] In some examples, the neuromonitoring data analysis device is configured to perform the following steps: administering to the mammal a plurality of neurophysical stimuli at a plurality of neural structures, e.g., of or associated with different or separate pathways; analyzing data describing a plurality of neurophysiological response signals generated associated with the plurality of neural structures of the mammal in response to administering one or more neural stimuli to the mammal; determining whether a detected abnormality is indicative of damage to at least one of the at least two neural structures; and, optionally, providing an output if the detected abnormality is indicative of damage to the at least one neural structure. The different pathways may, for example, be associated with different motor and / or sensory and / or parasympathetic and / or sympathetic pathways. In some examples, the "different pathways" may be associated with different pathways of the same category (different motor pathways) or different categories (motor and sensory pathways).

[0187] In some embodiments, a neuromonitoring data analysis device configured to monitor a nervous system of a subject comprises at least one processor and at least one memory configured to store data and software code portions executable by the at least one processor to receive patient data and sensor data, identify abnormal events indicative of damage to at least one of the at least two neural structures based on the received patient data, and provide a first output related to the identified events.

[0188] In some embodiments, the patient data describes a plurality of physical stimuli applied to at least two neural structures of a nervous system of a mammalian subject, and in some embodiments, the sensor data describes a plurality of neurophysiological response signals generated at the at least two neural structures in response to the application of the plurality of physical stimuli.

[0189] In some examples, the identifying step is further based on sensor data describing multiple physical stimuli, optionally simultaneously applied to multiple neural structures.

[0190] In some examples, the identifying step includes distinguishing between 1) a first abnormal event associated with damage to at least one of the at least two neural structures and 2) a second abnormal event not associated with damage to at least one of the at least two neural structures.

[0191] In an embodiment, by considering multiple neurophysiological response signals generated in at least two neural structures and, optionally, data describing the applied physical stimulus, the false positive rate of abnormal events identified as associated with damage to the neural structures is reduced compared to the false positive rate obtained if each response signal is analyzed individually.

[0192] In some examples, by considering multiple neurophysiological response signals generated by and / or in at least two neural structures, and optionally data describing the applied physical stimulus, the false negative rate of abnormal events identified as associated with damage to the neural structures is reduced compared to the false negative rate obtained if each response signal is analyzed individually.

[0193] In some embodiments, identifying includes classifying the abnormal event into one of the following categories: a first abnormal event associated with damage and / or injury to at least one of the at least two neural structures, and a second abnormal event not associated with damage to at least one of the at least two neural structures.

[0194] In some embodiments, the first output includes information regarding a probability that the detected abnormal event is associated with damage to at least one neural structure.

[0195] In some embodiments, the apparatus is configured to provide a second output describing a second abnormal event.

[0196] In some examples, the output describing the second anomalous event includes information regarding whether the second anomalous event is related to one of the following: a sensor misconfiguration; a signal artifact, a systemic physiological factor, a system malfunction, an environmental factor, or any combination of the foregoing.

[0197] In some examples, the systemic factors include one or more of the following: patient anesthesia, blood pressure, patient position, patient posture, or any combination of the foregoing.

[0198] In some embodiments, the first abnormal event is related to a functional state of a neural structure.

[0199] In some embodiments, the neural structures include neural pathways and / or receptors.

[0200] In some examples, the first abnormal event is associated with a neurological functional state that is classified into one of the following categories: "deteriorated" or "absent."

[0201] In some embodiments, the second abnormal event includes classifying the functional status of at least one neural structure as "normal."

[0202] In some examples, the plurality of physical stimuli are associated with one of the following: evoked potentials, reflexes, spontaneous potentials, or any combination of the above.

[0203] In some examples, the neuromonitoring data analysis device is configured to receive patient data including data describing a plurality of physical stimuli applied to at least two neural structures of a nervous system of a mammalian subject, and sensor data describing a plurality of neurophysiological response signals generated in the at least two neural structures in response to applying the plurality of physical stimuli. The device is further configured to (e.g., collectively) analyze the plurality of neurophysiological response signals, identify an abnormal event indicative of damage and at least one of the at least two neural structures based on the (e.g., collectively) analysis, and optionally provide a first output related to the identified event. In some examples, the phrase "collectively analyzing" refers to analyzing a greater amount of data describing various signals to detect and classify an abnormality in one or more of the analyzed signals. By considering the contribution of a relatively large number of signals, the false positive rate (e.g., classification of an abnormality as a neural structure damage) is reduced compared to approaches in which a relatively small number of signals are analyzed for anomaly detection and classification.

[0204] In some examples, the analyzing includes analyzing (e.g., collectively) data describing the plurality of response signals and describing physical stimuli applied to the at least two neural structures to generate the plurality of response signals.

[0205] In some examples, by considering (e.g., collectively analyzing) a plurality of neurophysiological response signals generated by at least two neural structures and, optionally, data describing the applied physical stimulus, the false positive rate of anomalous events identified as being associated with damage to the neural structures is reduced compared to the false positive rate that would be obtained if each response signal were analyzed individually. In some examples, by considering a plurality of neurophysiological response signals generated by at least two neural structures and, optionally, data describing the applied physical stimulus, the false negative rate of anomalous events identified as being associated with damage to the neural structures is reduced compared to the false negative rate that would be obtained if each response signal were analyzed individually.

[0206] In some examples, the neuromonitoring data analysis device is configured such that if the detected abnormality is determined not to be associated with damage to the at least one neural structure, a second output including information regarding the detected abnormality is provided.

[0207] In some embodiments, the first output indicates a probability that the detected abnormality indicates damage and / or injury to at least one neural structure.

[0208] In some examples, damage to at least one neural structure can cause a functional deficit and is the result of physical engagement (surgical engagement) with a tissue region that includes at least one neural structure.

[0209] In some embodiments, the device is configured to detect anomalies by comparing the generated response signals with corresponding baseline signals.

[0210] In some embodiments, the device includes a classifier for classifying a neurofunctional status of at least one neural structure into one of the following categories: "normal", "degraded", or "absent".

[0211] In some examples, a neuromonitoring data analysis device is configured to receive patient data including data describing a plurality of physical stimuli applied to at least two neural structures of a nervous system of a mammalian subject and sensor data describing a plurality of neurophysiological response signals generated in the at least two neural structures in response to the plurality of applied physical stimuli; determine whether a detected abnormality in the data describing the response signals is indicative of damage to at least one of the at least two neural structures; and, if the detected abnormality is determined by the device to be associated with damage to the at least one neural structure, the device is optionally configured to provide a first output including information regarding the damage to the at least one neural structure.

[0212] In some examples, the patient data describes one of the following: clinical patient data, demographic patient data, anesthesia data, physiological patient data, surgical data, baseline motor evoked potential signal data, baseline EMG signals, baseline EEG signals, baseline somatosensory evoked potential signal data, reflex actions, or any combination of the foregoing.

[0213] In some examples, the first output includes information characterizing the pathway injury as associated with one of the following, e.g., a motor injury including a motor injury impact location, a somatosensory injury including a sensory impact location, a spinal cord pathway injury severity including complete or incomplete injury, or any combination of the foregoing. In some examples, the device is configured to characterize injury to at least one neural structure as associated with one of the following: a myotome injury, and / or a nerve injury severity including complete or incomplete injury.

[0214] In some embodiments, the second output of the abnormal event includes clinical information describing one: the general physiological condition, the neuromonitoring system malfunction, or both.

[0215] In some embodiments, the device is configured to provide a surgical recommendation output that includes one of the following: elicit additional motor evoked potentials, elicit additional somatosensory evoked potentials, change stimulation intensity of additional motor evoked potentials, change stimulation intensity of additional somatosensory evoked potentials, update recording parameters, check for malfunction of neuromonitoring system, check impedance, check anesthesia parameters, perform peripheral nerve stimulation, check physiological parameters, check and / or change patient position and / or posture (method), maintain surgery, position electrodes, or any combination of the foregoing. In some embodiments, the surgical recommendation output depends on the identified abnormality.

[0216] In some examples, the apparatus includes a plurality of hierarchically arranged machine learning submodels. A first level machine learning submodel of the hierarchically arranged machine learning submodels may be configured to analyze a plurality of channels of signals associated with a signal modality to generate a plurality of respective channel-by-channel analysis outputs. A second level machine learning submodule of the hierarchically arranged machine learning submodels may be configured to receive the plurality of channel-by-channel analysis outputs. The second level machine learning submodule may be configured to analyze the plurality of received channel-by-channel analysis outputs of the signal modality to provide an analysis output describing the modality. A third level machine learning submodule of the hierarchically arranged machine learning submodels may receive the analysis output describing the modality. The third level machine learning submodule is configured to analyze the output received from the second submodule to generate an output describing a clinical interpretation of the received patient data.

[0217] In some examples, a neuromonitoring data analysis device is configured to receive patient data including data describing a plurality of physical stimuli applied to a mammalian subject to responsively generate a plurality of corresponding stimulation signals in at least two neural structures of the subject's nervous system and sensor data describing a plurality of neurophysiological response signals generated in response to the plurality of applied physical stimuli, determine whether the response signals exhibit abnormal characteristics with respect to at least one of the at least two neural structures, distinguish between the at least one abnormal characteristic indicative of damage to at least one of the at least two neural structures and the at least one abnormal characteristic not indicative of damage and / or injury to the at least one neural structure, and optionally provide an output if the detected abnormal characteristic is indicative of damage to the at least one neural structure.

[0218] In some embodiments, a neuromonitoring data analyzer is configured to receive patient data including data describing a plurality of physical stimuli applied to a mammalian subject to responsively generate a plurality of corresponding stimulation signals in at least two neural structures of the subject's nervous system, and sensor data describing a plurality of neurophysiological response signals generated in response to the plurality of applied physical stimuli, classifying an abnormality detected in the response signals into one of the following categories: a first abnormality indicative of damage to at least one of the at least two neural structures, and a second abnormality not associated with damage to the at least one neural structure, and optionally providing an output if the abnormality is classified as the first abnormality. In some embodiments, the classifying also takes into account a baseline signal associated with each of the at least two neural structures.

[0219] In some embodiments, the classifying is performed after the steps of analyzing the received data of the applied physical stimulus and the generated response signals, and detecting an anomaly in at least one of the generated response signals based on the analyzing.

[0220] In some embodiments, the analyzing includes collectively analyzing the generated response signal and, optionally, the received data of the applied physical stimulus. In some embodiments, the analyzing includes collectively analyzing the received data of the applied physical stimulus, the generated response signal, and the corresponding baseline signal.

[0221] In some examples, a neuromonitoring data analysis method includes applying neurostimulation to a mammal, analyzing data describing signals generated in response to the applied neurostimulation, detecting an anomaly associated with the generated response signal, determining whether the detected anomaly is indicative of damage to at least one of the at least two neural structures, and, optionally, providing an output if the detected anomaly is indicative of damage to the at least one neural structure. In some examples, the data also describes physical stimuli applied to the mammalian subject to responsively generate a plurality of response signals in the nervous system of the mammal. In some examples, the data (collectively) analyzed also describes a corresponding baseline response signal. In some examples, the detected anomaly indicative of damage to the at least one neural structure is associated with a functional state of the neural structure. In some examples, the determining includes classifying the functional state of the neural structure. In some examples, the method includes classifying the neural functional state into one of the following: "normal," "degraded," or "absent."

[0222] In some examples, the method includes analyzing (e.g., collectively) a plurality of channels of signals associated with a signal modality to generate analysis outputs for each of the plurality of channels by a first level machine learning sub-model of a hierarchically arranged learning sub-model.

[0223] In some examples, the method includes receiving, by a machine learning submodule at a second level of the hierarchically arranged submodules, a plurality of analysis outputs for each channel, and analyzing, by the second level of the machine learning submodules, the plurality of received analysis outputs for each channel of the signal modality to provide an analysis output describing the modality. In some examples, the method includes receiving, at a machine learning submodule at a third level of the hierarchically arranged machine learning submodules, the analysis output describing the modality, and analyzing, by the third level of the machine learning submodules, the output received from the second level of the submodule to generate an output describing a clinical interpretation of the received patient data.

[0224] In some examples, by considering multiple neurophysiological response signals generated in at least two neural structures and, optionally, data describing the applied physical stimulus, the false positive rate of an abnormal event identified as being associated with damage to a neural structure is reduced compared to the false positive rate obtained if each response signal is analyzed individually.

[0225] In some examples, by considering multiple neurophysiological response signals generated in at least two neural structures and, optionally, data describing the applied physical stimulus, the false negative rate of abnormal events identified as associated with damage to the neural structures is reduced compared to the false negative rate obtained if each response signal is analyzed individually.

[0226] The methods described herein and illustrated in the accompanying figures are not to be construed as limiting. For example, the methods described herein may include additional processes or operations or may include fewer processes or operations as compared to those described herein and / or illustrated in the figures. In addition, the method steps are not necessarily limited to the chronological order as illustrated and described herein.

[0227] Any digital computer system, apparatus, unit, device, module and / or engine illustrated herein can be configured or otherwise programmed to perform the methods disclosed herein, and as long as the system, apparatus, module and / or engine is configured to perform such methods, it is within the scope and spirit of the present disclosure. When the system, apparatus, module and / or engine is programmed to perform a particular function according to computer readable and executable instructions from the program software implementing the methods disclosed herein, it actually becomes a special purpose computer dedicated to the embodiment of the methods disclosed herein. The methods and / or processes disclosed herein can be implemented as a computer program product that can be tangibly embodied in an information carrier, including, for example, in a non-transitory tangible computer readable and / or non-transitory tangible machine readable storage device. The computer program product can be directly loaded into the internal memory of a digital computer that includes the software code portions for performing the methods and / or processes disclosed herein.

[0228] The methods and / or processes disclosed herein may be implemented as a computer program that may be embodied intangibly by a computer-readable signal medium. A computer-readable signal medium may include a propagated data signal in which computer-readable program code is embodied, for example, in baseband or as part of a carrier wave. Such a propagated signal may take any of a variety of forms, including, but not limited to, electromagnetic, optical, or any combination thereof. A computer-readable signal medium may not be a non-transitory computer or machine-readable storage device, but may be any computer-readable medium that can communicate, propagate, or transmit a program used by or in connection with the devices, systems, platforms, methods, operations, and / or processes discussed herein.

[0229] The terms "non-transitory computer-readable storage" and "non-transitory machine-readable storage" encompass distribution media, intermediate storage media, computer execution memory, and any other medium or device capable of being stored for subsequent reading by a computer program that implements an embodiment of the methodology disclosed herein. A computer program product can be deployed to be executed on one computer, or on multiple computers at one site, or can be distributed across multiple sites and interconnected by one or more communications networks.

[0230] These computer readable and computer executable instructions may be provided to a processor of a general purpose computer, a special purpose computer, or other programmable data processing apparatus to manufacture a machine, whereby the instructions executing via the processor of the computer or other programmable data processing apparatus create means for performing the functions / acts specified in the flowcharts and / or block diagrams of the block or blocks. These computer readable and computer executable program instructions may also be stored on a computer readable storage medium capable of instructing a computer, programmable data processing apparatus, and / or other device to function in a particular manner, whereby a computer readable storage medium having instructions stored thereon includes an article of manufacture including instructions implementing aspects of the functions / acts specified in the flowcharts and / or block diagrams of the block or blocks.

[0231] Furthermore, computer-readable and computer-executable instructions can be loaded into a computer, other programmable data processing apparatus, or other device to execute a series of operational steps on the computer, other programmable data processing apparatus, or other device to generate a computer-implemented process such that the instructions executing on the computer, other programmable data processing apparatus, or other device perform the functions / acts specified in one or more blocks of the flowcharts and / or block diagrams.

[0232] The term "engine" may include one or more computer modules, where a module may be a self-contained hardware and / or software component that interfaces with a larger system. A module may comprise one or more machine executable instructions. A module may be embodied by a circuit or controller programmed to cause a system, apparatus, and / or platform to perform the methods, processes, and / or operations disclosed herein. For example, a module may be implemented as a hardware circuit including custom VLSI circuits or gate arrays, application specific integrated circuits (ASICs), off-the-shelf semiconductors such as logic chips, transistors, and / or other discrete components, and the like. A module may also be implemented in a programmable hardware device, such as a field programmable gate array, programmable array logic, programmable logic device, and / or the like.

[0233] In the discussion, unless otherwise indicated, adjectives such as "substantially" and "about" modifying a condition or relationship characteristic of a feature or characteristic of an embodiment of the invention should be understood to mean that the condition or characteristic is defined within an acceptable range permitted for the operation of the embodiment for its intended use.

[0234] Unless otherwise specified, the terms "substantially," "about," and / or "close to" in relation to a size or value may mean within an inclusive range of -10% to +10% of the respective size or value.

[0235] "Combined with" can mean either indirectly "combined with" or directly "combined with".

[0236] It is important to note that what may be included in the method is not limited to these figures or the corresponding description. For example, the method may include additional or fewer processes or operations as compared to those described in the figures. Furthermore, the method embodiments are not necessarily limited to the chronological order as shown and described herein.

[0237] As used herein, discussions using terms such as "processing," "computing," "calculating," "determining," "analyzing," "checking," "estimating," "deriving," "selecting," "inferring," and the like, may refer to an operation and / or process(es) of a computer, computing platform, computing system, or other electronic computing device that manipulates and / or transforms data represented as physical (e.g., electronic) quantities in the computer's registers and / or memory into other data that is similarly represented as physical quantities in the computer's registers and / or memory, or other information storage medium that may store instructions for performing the operation and / or process. The term determining may also include the meaning "determining heuristically" in some cases.

[0238] It should be noted that if an embodiment refers to a condition "above threshold", this should not be interpreted as excluding embodiments referring to a condition "equal to or above threshold". Similarly, if an embodiment refers to a condition "below threshold", this should not be interpreted as excluding embodiments referring to a condition "equal to or below threshold". If a condition is interpreted as being fulfilled if the value of a given parameter is above a threshold, it is clear that the same condition is considered not to be fulfilled if the value of the given parameter is below the given threshold. Conversely, a condition should be interpreted as being fulfilled if the value of the given parameter is equal to or above the threshold, and the same condition is considered not to be fulfilled if the value of the given parameter is below (and not below) the given threshold.

[0239] When a claim or specification refers to "a" or "an" element and / or feature, it is to be understood that such a reference is not to be construed as referring to only one of that element. Thus, for example, a reference to "an element" or "at least one element" can also encompass "one or more elements."

[0240] Terms used in the singular form shall also include the plural unless expressly stated otherwise or the context otherwise requires.

[0241] In this specification and claims, the verbs "comprise," "include," and "have" and their conjugations are used to indicate that the data portion or portions of the verb are not necessarily an exhaustive list of components, elements, or portions of the subject or subjects of the verb.

[0242] Unless otherwise noted, the use of the term "and / or" between the last two members of a list of alternatives for selection indicates that one or more of the listed alternatives are appropriate and can be selected. Furthermore, the term "and / or" can be used interchangeably with the terms "at least one of," "any one of," or "one or more of," followed by a list of various alternatives.

[0243] As used herein, the phrase "A, B, C, or any combination of the foregoing" should be interpreted to mean all of the following: (i) A or B or C, or any combination of A, B, and C; (ii) at least one of A, B, and C; (iii) A, and / or B, and / or C; and (iv) A, B, and / or C. Where appropriate, the phrases A, B, and / or C can be interpreted to mean A, B, or C. The phrases A, B, or C should be interpreted to mean "selected from the group consisting of A, B, and C." This concept is illustrated for three elements (i.e., A, B, C), but extends to fewer and greater numbers of elements (e.g., A, B, C, D, etc.).

[0244] It will be understood that certain features of the invention that are, for clarity, described in the context of separate embodiments or examples, may also be provided in combination in a single embodiment. Conversely, various features of the invention that are, for brevity, described in the context of a single embodiment, example, and / or option, may also be provided separately or in any suitable subcombination or as appropriate in other described embodiments, examples, or options of the invention. Certain features described in the context of various embodiments, examples, and / or optional implementations are not considered essential features of those embodiments, examples, and / or optional implementations, unless the embodiment, example, and / or optional implementation is inoperable without those elements.

[0245] It should be noted that the terms "in some embodiments," "according to some embodiments," "for example," "eg," "for instance," and "optionally" can be used interchangeably herein.

[0246] The number of elements shown in the figures should not be construed as limiting in any way but for illustrative purposes only.

[0247] It should be noted that the term "operable to" can encompass the meaning of the term "modified or configured to." In other words, a machine "capable of performing" a task can, in some embodiments, encompass the mere ability to perform that function (e.g., "modified"), and in other embodiments, encompass a machine that is actually made (e.g., "configured") to perform that function.

[0248] Throughout this application, various embodiments may be presented in and / or in connection with a range format. It should be understood that the description in range format is merely for convenience and brevity and should not be construed as an inflexible limitation on the scope of the embodiments. Thus, the description of a range should be considered to have specifically disclosed all possible subranges as well as individual numerical values ​​within that range. For example, description of a range such as 1 to 6 should be considered to have specifically disclosed subranges such as 1 to 3, 1 to 4, 1 to 5, 2 to 4, 2 to 6, 3 to 6, etc., as well as individual numerical values ​​within that range, e.g., 1, 2, 3, 4, 5, and 6. This applies regardless of the breadth of the range.

[0249] The phrases "ranging / ranges between" a first and a second designation number, and "ranging / ranges from" a first designation number "to" a second designation number, are used interchangeably herein and are meant to include the first and second designation numbers and all fractional and integer numbers therebetween.

[0250] While the invention has been described with respect to a limited number of embodiments, these should not be construed as limiting the scope of the invention, but rather as illustrative of some of the embodiments.

Claims

1. A nerve monitoring data analysis device configured to monitor the nervous system of a subject, (a) collectively analyzing a plurality of neurophysiological signals generated by at least two channels associated with at least one nerve structure in response to the application of a plurality of physical stimuli to the subject, thereby extracting features from the plurality of neurophysiological signals; (b) classifying the features using an ML model and identifying at least one neurophysiological response signal among the plurality of neurophysiological response signals indicating damage in the at least one nerve structure; A nerve monitoring data analysis device comprising a processor configured to perform the above.

2. The device according to claim 1, wherein the at least two channels are associated with at least two nerve structures.

3. The device according to claim 1, wherein the plurality of neurophysiological response signals are generated in response to simultaneous application of physical stimuli.

4. The device according to claim 1, wherein (b) includes distinguishing a first abnormal event associated with damage to the at least one nerve structure from a second abnormal event not associated with damage to the at least one nerve structure.

5. The device according to claim 4, wherein the second abnormal event is associated with systemic factors.

6. The device according to claim 5, wherein the systemic factors include at least one selected from the group consisting of patient anesthesia, blood pressure, patient position, and patient posture.

7. The device according to claim 1, wherein (b) includes comparing the plurality of neurophysiological response signals with corresponding baseline signals.

8. The device according to claim 1, wherein the processor is further configured to analyze patient data including at least one selected from the group consisting of demographic data, anesthesia data, physiological data, surgical data, baseline motor evoked potential signal data, baseline EMG signal data, baseline EEG signal data, baseline somatosensory evoked potential signal data, and reflex data.

9. The apparatus according to claim 1, wherein the processor executes a machine learning model having a plurality of levels of hierarchically arranged machine learning sub-models.

10. The apparatus according to claim 9, wherein the first level of the machine learning model is configured to generate a plurality of signal modality outputs that associate each of the plurality of neurophysiological response signals with a specific signal modality and distinguish normal activity from abnormal activity.

11. The apparatus according to claim 10, wherein the second level of the machine learning model is configured to analyze each signal modality output to provide an analysis output that describes the signal modality.

12. The apparatus according to claim 11, wherein the second level is further configured to detect the occurrence of a pattern of change in the aggregated signal of each signal modality.

13. The apparatus according to claim 11, wherein the third level of the machine learning model interprets the analysis output as a clinical interpretation of each of the plurality of neurophysiological response signals.

14. The apparatus according to claim 10, wherein each of the plurality of neurophysiological response signals is analyzed for the feature extraction based on the specific signal modality.

15. The apparatus according to claim 14, wherein the features are features of MEP signals or features of SSEP signals.

Images