Intraoperative neurophysiological monitoring (IONM) in neurosurgery: A critical appraisal of established practices, ongoing controversies, and future trajectories Authors: Yu’e Sun, Fei Xu, Nazneen Sudhan, Fuhai Ji, Ke Peng, Xiaqing Ma Published: 2026 ABSTRACT Intraoperative neurophysiological monitoring (IONM) has evolved from a novel technique into an evidence-based standard treatment method for high-risk neurosurgical and spinal surgeries. Its effectiveness is based on two interrelated pillars: optimized multimodal monitoring, mainly including motor and somatosensory evoked potentials (MEPs/SSEPs) as well as electromyography (EMG), and total intravenous anesthesia (TIVA) combined with a precise neuromuscular blockade-based anesthesia protocol. IONM significantly reduces neurological function damage during surgeries for spinal deformities, intramedullary tumors, acoustic neuromas, and gliomas in the brain functional area, redefining the standards of safe surgical practice. However, there are still certain challenges, including the difficulty in converting signal changes into clinical actions, controlling high false alarm rates, and overcoming technical/logistical obstacles in complex and lengthy surgeries. These objectively existing problems further highlight the importance of clinical judgment. Looking to the future, a key developmental direction involves transforming intraoperative neurophysiological monitoring (IONM) from a passive monitoring tool into a system capable of predictive guidance and comprehensive neuroprotection. Emerging models include artificial intelligence (AI) technologies for real-time analysis and technologies for fusing multimodal data into surgical “dashboards”, but they still face significant obstacles in data quality, clinical validation, and human-centered design. Closed-loop systems and the application of neurobiomarker recognition to achieve neuroprotection remain enduring research topics. In summary, the development of IONM technology towards a more mature direction requires a coordinated planning scheme: establishing evidence-based standards, promoting data-driven discoveries through large-scale collaborative research, and achieving deep multidisciplinary integration within the surgical team. The ultimate goal is to make IONM an intelligent guiding tool that not only monitors but also actively optimizes surgical strategies to ensure the preservation of neural function. Introduction Neurosurgical procedures aim to achieve the best therapeutic effect while protecting the delicate functions of the nervous system. The inherent risk of iatrogenic injury during neurosurgical procedures has driven the evolution of Intraoperative neurophysiological monitoring (IONM) from an auxiliary technique to the cornerstone of modern functional-oriented neurosurgery.1 The development of IONM has changed the surgical approach, shifting the focus from simple anatomical landmarks to real-time assessment of functional integrity.2, 3, 4, 5, 6 By integrating electrophysiological techniques such as somatosensory evoked potentials (SSEPs), motor evoked potentials (MEPs), electroencephalography (EEG), and electromyography (EMG), IONM provides surgeons with dynamic feedback, enabling them to actively adjust surgical strategies and minimize neurological dysfunction during the operation.7, 8, 9 The development history of IONM can be divided into different stages: the initial stage of monitoring sensory and motor pathways (such as SSEPs and MEPs), during which the principle of real-time functional feedback was established10, 11; the stage of localization and precision, represented by electrocorticography (ECoG), stereo-electroencephalography (SEEG), and direct cortical stimulation, during which surgery shifted from passive monitoring to active mapping of functional patterns 12, 13; and the emerging stage of data integration, during which artificial intelligence and multimodal fusion are expected to provide predictive rather than passive guidance.14, 15 The purpose of this review is not merely to list the relevant technologies. We will conduct an assessment and comprehensive analysis of the current IONM, aiming to achieve the following goals: to examine existing and emerging technologies, distinguishing reliable verification from promising speculation; to analyze persistent challenges, such as disputes over alarm thresholds, signal artifacts, and the gap in interdisciplinary collaboration - which fundamentally underlie the divergence between technical feasibility and continuous clinical practicality; to consider technological innovations in actual surgical procedures, evaluating how they can effectively expand the scope of safe resection; and to critically assess future development directions, including closed-loop systems and biomarker-based monitoring, clarifying whether their have the conditions for transformation, the obstacles to transformation, and the possibility that IONM can be transformed from a monitoring tool into an active auxiliary tool in the operating room. By focusing on the evidence gaps and transformation barriers in this discussion, this review aims to provide a roadmap for advancing IONM technology towards a more standardized, predictive, and deeply integrated field of precise neurosurgery. The established pillars of IONM: techniques, evidence, and standard of care IONM has evolved from an experimental tool into the cornerstone of modern neurosurgery. This transformation was achieved based on a series of reliable clinical evidence supporting the core technologies. This section will analyze and evaluate the data that established these methods as standard treatment approaches. Core monitoring methods supported by evidence Reliable IONM is based on a variety of validated technologies, each serving different functional pathways. The advantages of these technologies lie not only in their maturity but also in their correlation with clinically significant results in large-scale studies and meta-analyses. Somatosensory and motor evoked potentials (SSEPs & MEPs): guardians of long-tract integrity SSEPs can monitor the dorsal column-medial lemniscus pathway and provide real-time feedback on sensory pathway.16, 17 MEPs are elicited by transcranial electrical stimulation and used to assess the functional status of the corticospinal tract.18, 19, 20 In spinal and nerve vascular surgeries, the combination of these two techniques is one of the most reliable practices in IONM.11, 21, 22 A large amount of literature, including prospective cohort studies and meta-analyses, has provided strong evidence for the role of multimodal monitoring (including SSEPs and MEPs) in standard treatment.23, 24 In spinal and spinal cord surgeries, such as deformity correction and intramedullary tumor resection, evidence indicates that this method can significantly reduce the risk of severe neurological injury.25 Similarly, in neurovascular surgeries such as aneurysm clipping, changes in somatosensory evoked potentials and motor evoked potentials can serve as early indicators of ischemia.26, 27, 28 There are cohort studies supporting this view, including a study by Wilent et al. (2021), involving over 2278 patients. The results indicate that timely response to the alerts and intervention can significantly reduced the occurrence of new motor dysfunction after surgery, highlighting the preventive effect of this method.29 Direct cortical and subcortical stimulation: the gold standard for functional mapping Although SSEPs and MEPs are mainstream monitoring tools, direct electrical stimulation remains an indispensable active localization technique. In tumor resection surgeries, especially for brain tumors located in functional areas, it is crucial for identifying and protecting the functional cortex (motor, language) as well as the subcortical white matter tracts. Studies have shown that direct stimulation is the indisputable gold standard for functional localization in intracranial tumor surgeries, enabling safe resection to the greatest extent while reducing complications.30, 31 Studies have also demonstrated a correlation between the positive mapping positions during the surgery and the postoperative functional outcomes, providing causal evidence for the practical feasibility of direct electrical stimulation.32, 33 Therefore, it has evolved from an auxiliary surgical method to an important component of the core workflow in tumors surgeries located in or adjacent to eloquent areas. Cranial nerve monitoring (free-running and triggered EMG): preserving critical peripheral pathways Continuous and triggered EMG is commonly used to monitor the functions of cranial motor nerves (e.g., facial, vagus, hypoglossal) during surgeries on the base of the skull and the brainstem. Its application in the surgery for acoustic neuroma is a perfect example of the successful integration of IONM.34 Researches have established dynamic, EMG-based risk stratification models that directly inform surgical strategy and predict postoperative facial nerve function.35, 36, 37 The technique’s ability to provide continuous, real-time feedback on nerve integrity has consequently made it a routine and expected component of these procedures, directly contributing to improved functional preservation rates.38 Synthesis: what constitutes a “pillar”? These monitoring methods share several common features: a clear mechanism, meaning that the biological principles being evaluated and the specific neural pathways have been fully understood; standardized procedures, including widely recognized guidelines for setup, stimulation parameters, and alarm standards; the relevance of results, meaning that strong and repeatable data demonstrate a correlation between specific signal changes and predictable postoperative neurological dysfunction; and the practicality of interventions, meaning that the feedback provided by them is timely and specific, sufficient to guide effective surgical or anesthetic intervention measures. These features make them the cornerstone of IONM. Therefore, the pillars of IONM are a set of evidence-based intervention measures. They form the indispensable foundation of modern functional neurosurgery. The following sections will explore the areas where this solid foundation remains controversial and those with evolving evidence. The strength of evidence, key outcomes, and scope for each modality are summarized in Table 1. This matrix supports the calibrated standard-of-care statements below.Table 1Evidence-level matrix for ionm modalities.ModalityEvidence strengthKey findingsScope of applicationMEPHighSignificantly reduces the risk of severe neurological injury in spinal surgeries; serves as an early indicator of ischemia in neurovascular surgeriesSpinal/spinal cord surgeries (deformity correction, tumor resection); Neurovascular surgeries (aneurysm clipping, carotid endarterectomy)SSEPHighAssesses the functional status of the corticospinal tract; timely intervention based on alerts significantly reduces new postoperative motor deficitsSpinal/spinal cord surgeries (deformity correction, intramedullary tumors); Neurovascular surgeries (aneurysm clipping); Surgeries near the motor cortexEMGHighProvides continuous, real-time feedback on nerve integrity; establishes risk stratification models to predict postoperative function (e.g., facial nerve)Skull base surgeries (acoustic neuroma); Brainstem surgeries; Procedures involving cranial motor nerves (facial, vagus, hypoglossal, etc.)BAEPModerate-HighMonitors auditory pathways and brainstem function; identifies risk of hearing impairment in cerebellopontine angle surgeriesPosterior fossa surgeries (acoustic neuroma, microvascular decompression); Procedures involving the brainstem or auditory nerveDCSHighIdentifies and preserves functional cortex and subcortical white matter tracts; directly correlates intraoperative mapping with postoperative functional outcomesResection of brain tumors in eloquent areas (motor, language); Epilepsy surgery (cortical functional mapping) Table 1Evidence-level matrix for ionm modalities.ModalityEvidence strengthKey findingsScope of applicationMEPHighSignificantly reduces the risk of severe neurological injury in spinal surgeries; serves as an early indicator of ischemia in neurovascular surgeriesSpinal/spinal cord surgeries (deformity correction, tumor resection); Neurovascular surgeries (aneurysm clipping, carotid endarterectomy)SSEPHighAssesses the functional status of the corticospinal tract; timely intervention based on alerts significantly reduces new postoperative motor deficitsSpinal/spinal cord surgeries (deformity correction, intramedullary tumors); Neurovascular surgeries (aneurysm clipping); Surgeries near the motor cortexEMGHighProvides continuous, real-time feedback on nerve integrity; establishes risk stratification models to predict postoperative function (e.g., facial nerve)Skull base surgeries (acoustic neuroma); Brainstem surgeries; Procedures involving cranial motor nerves (facial, vagus, hypoglossal, etc.)BAEPModerate-HighMonitors auditory pathways and brainstem function; identifies risk of hearing impairment in cerebellopontine angle surgeriesPosterior fossa surgeries (acoustic neuroma, microvascular decompression); Procedures involving the brainstem or auditory nerveDCSHighIdentifies and preserves functional cortex and subcortical white matter tracts; directly correlates intraoperative mapping with postoperative functional outcomesResection of brain tumors in eloquent areas (motor, language); Epilepsy surgery (cortical functional mapping) Evidence-level matrix for ionm modalities. Anesthetic protocols for optimized standard monitoring The effective implementation of IONM technology in clinical practice is closely related to the anesthesia plan. Anesthetics have profound and dose-dependent effects on the excitability of neurons and synaptic transmission, which can directly regulate or inhibit the electrophysiological signals being monitored. Therefore, developing and standardizing an anesthesia protocol compatible with IONM is not only auxiliary but also an essential component of the monitoring mode itself. The main goal is to achieve adequate anesthesia and analgesia to minimize interference with signal acquisition while ensuring the safety and immobility of the patient.39 The consensus protocol for optimal monitoring (especially for MEPs monitoring) mainly adopts total intravenous anesthesia (TIVA) technology.40, 41 TIVA is typically based on the infusion of propofol and opioids (such as remifentanil), and is preferred over volatile inhalation anesthetics.42, 43 Although both propofol and volatile drugs inhibit cortical activity, propofol has a milder effect on the excitability of spinal motor neurons and the inhibition of peripheral neuromuscular junctions. Volatile drugs, particularly at minimum alveolar concentration (MAC) levels, cause a dose-dependent reduction in the amplitude of MEPs, prolong the latency, and can significantly weaken or even completely eliminate the response.44, 45, 46 The use of these drugs is usually limited to very low concentrations (such as <0.5 MAC) as an auxiliary measure when absolutely necessary, but it should be noted that this may reduce the sensitivity of the monitoring. It is equally important to manage neuromuscular blocking agents (NMBAs) meticulously. Although complete neuromuscular blockade is incompatible with MEPs and free-running EMG monitoring, achieving a certain level of muscle relaxation is often necessary for surgical procedures and to prevent patient movement.47 The solution lies in achieving controlled partial blockade. This is most reliably achieved through continuous infusion of short-acting neuromuscular blocking agents (such as rocuronium), with the dose adjusted based on train of four(TOF) monitoring to maintain a stable, low-level response (e.g., 1–2 contractions). This method not only provides excellent surgical conditions for surgeons, but also maintains the ability to trigger interpretable motor evoked potentials and electromyographic signals.48, 49 In addition, attention should also be paid to the depth of anesthesia. The bispectral index (BIS) can assist in adjusting the dosage of anesthetic drugs, ensuring that patients receive adequate anesthesia effects while avoiding the use of excessive anesthetic drugs that may lead to signal suppression. Stable hemodynamics is also very important, as hypotension may cause ischemia in the spinal cord or brain, manifesting as neuroimaging changes related to actual injuries rather than artifacts caused by anesthesia.50, 51, 52 In summary, the anesthesia plan for IONM is a carefully designed and dynamically balanced strategy, centered on TIVA, precise control of neuromuscular blockade, and maintenance of stable physiology. This optimized environment is the prerequisite for obtaining reliable and interpretable signals, thereby ensuring IONM’s evidentiary value in clinical practice. The pivotal role of TIVA and the detrimental effects of volatile anesthetics and deep paralysis on signal quality are effectively contrasted in Fig. 1.Fig. 1Anesthetic protocols for optimized standard monitoring. A consensus TIVA protocol, carefully titrated and combined with partial short-acting NMBA blockade (adjusted by TOF), provides optimal conditions for reliable IONM signals (MEPs/EMG) by minimizing interference at both the neuronal and neuromuscular levels. BIS monitoring enables precise anesthesia depth control, further safeguarding signal quality. In contrast, volatile anesthetics and excessive neuromuscular blockade are visually shown to disrupt or abolish monitoring signals, highlighting their unsuitability for IONM-guided procedures. Anesthetic protocols for optimized standard monitoring. A consensus TIVA protocol, carefully titrated and combined with partial short-acting NMBA blockade (adjusted by TOF), provides optimal conditions for reliable IONM signals (MEPs/EMG) by minimizing interference at both the neuronal and neuromuscular levels. BIS monitoring enables precise anesthesia depth control, further safeguarding signal quality. In contrast, volatile anesthetics and excessive neuromuscular blockade are visually shown to disrupt or abolish monitoring signals, highlighting their unsuitability for IONM-guided procedures. To operationalize the above, Table 2 outlines recommended intraoperative targets for MEPs, SSEPs, and EMG-dominant cases.Table 2Practical monitoring parameters.ModalityBISMAP (mmHg)Temp (°C)Hb (g/dL)PaCO₂ (mmHg)NMBA StrategyMEP40–60; avoid volatiles < 0.5 MAC; prefer propofol TIVA≥ 65 (spine)/≥ 70 (brain)≥ 36≥ 935–45Induction: none/minimal; Maintenance: rocuronium TOF 1–2; Reverse before closureSSEP40–60; sevo ≤ 0.3 MAC optional≥ 60≥ 35.5≥ 835–40As MEP, TOF may be 1–3EMG50–70; avoid propofol during mapping; dexmedetomidine + remifentanil≥ 65≥ 36≥ 935–45Usually none; if needed, short-acting agent + TOF Table 2Practical monitoring parameters.ModalityBISMAP (mmHg)Temp (°C)Hb (g/dL)PaCO₂ (mmHg)NMBA StrategyMEP40–60; avoid volatiles < 0.5 MAC; prefer propofol TIVA≥ 65 (spine)/≥ 70 (brain)≥ 36≥ 935–45Induction: none/minimal; Maintenance: rocuronium TOF 1–2; Reverse before closureSSEP40–60; sevo ≤ 0.3 MAC optional≥ 60≥ 35.5≥ 835–40As MEP, TOF may be 1–3EMG50–70; avoid propofol during mapping; dexmedetomidine + remifentanil≥ 65≥ 36≥ 935–45Usually none; if needed, short-acting agent + TOF Practical monitoring parameters. Anesthetic management for awake craniotomy and intracranial mapping The above TIVA-based anesthesia protocol is the foundation for most neurosurgical procedures performed under general anesthesia. However, for awake craniotomy surgeries (especially those involving direct cortical or subcortical stimulation for functional localization in tumor or epilepsy surgeries), a unique and highly specialized anesthesia method is required, namely awake anesthesia. Awake anesthesia has three significant characteristics: ensuring patient comfort and cooperation during the testing phase, maintaining a safe airway and physiological stability, and most importantly, minimizing the interference of drugs with spontaneous ECoG activity and the induced reactions during stimulation localization.53 The most typical awake anesthesia protocol follows the “anesthesia-awake-anesthesia” (AAA) or MAC model.54 In the initial “anesthesia” stage (used for craniotomy), a propofol and remifentanil-based TIVA is typically used for induction and maintenance, similar to the standard protocol, but long-acting sedatives or NMBAs are absolutely avoided. A laryngeal mask is often used during this stage to assist ventilation and also ensures a rapid and smooth awakening of the patient.55 The crucial step of transitioning from the “anesthesia state” to the “awake” stage for cortical layer imaging requires meticulous preparation. Many hospitals use a small dose of dexmedetomidine infusion for sedation due to its excellent sedative effect and minimal respiratory depression, and also combine a small dose of remifentanil infusion for analgesia. During this stage, propofol should be avoided as it may significantly inhibit the excitability of the cortical layer and the epileptiform activity on the EEG, potentially leading to false-negative imaging results. At this stage, the anesthesiologist assumes more of the role of a “neuro-psychological monitor”, continuously assessing the patient’s level of consciousness, pain level, anxiety status, and the ability to complete tasks (such as counting, object naming, limb movement).56 Airway management strategies are of utmost importance. Once the patient regains consciousness, shows cooperation, and is able to breathe independently, the supraglottic airway should be removed. Anesthesiologists should prepare in advance for possible complications (such as convulsions, nausea, or breathing difficulties), while also ensuring that the sterile area and the patient’s position remain unaffected. Finally, after the positioning and tumor resection are completed, the patient will undergo another sedation (the second “sleeping” stage) to facilitate wound suturing, and usually a laryngeal mask airway will be reinserted. This tailored and dynamic anesthesia strategy not only provides support but is also an intrinsic and positive component of the IONM process during awake surgery. It directly achieves high-precision functional localization, thereby determining the maximum safe resection range. Scalp block and emergency airway plan Prior to dural opening, a scalp block with 0.5% ropivacaine + 1:200,000 epinephrine reduces incisional pain and sedative requirements. In case of emergence agitation, airway obstruction, or emesis during awake mapping, a preplanned GA conversion protocol is initiated: rapid-sequence induction with succinylcholine 1–1.5 mg/kg or rocuronium 1.2 mg/kg, insertion of supraglottic airway or endotracheal tube, and resumption of TIVA with signal monitoring. Management of intraoperative seizures Seizures induced by cortical stimulation or ECoG are managed by immediate cessation of stimulation, administration of lorazepam 2 mg IV or midazolam 1–2 mg IV, and deepening anesthesia (BIS reduction). Duration, morphology, and timing are documented for postoperative correlation. Clinical impact: defining the standard of care In the specific field of neurosurgery, IONM has fundamentally altered the balance between risks and benefits, and has thus become an indispensable part of the definition of safe surgical practices. This is due to the combination of powerful monitoring technology and optimized anesthesia protocols, which has transformed IONM from a mere auxiliary tool into an essential component of surgical safety. In complex spinal surgeries, especially in cases of spinal deformity correction and intramedullary tumor resection, multimodal neuroimaging navigation (SSEPs andMEPs) has become necessary.57 High-level evidence indicates that IONM significantly reduces the incidence of severe nerve damage, which has led to its inclusion in professional association guidelines.23 Its role extends beyond preventing spinal cord injuries; it also enables more thorough removal of intramedullary lesions and gives surgeons greater confidence in taking corrective measures when dealing with high-risk spinal deformity cases, as they know that the functional integrity of the spinal cord is being monitored in real time.58 Similarly, in the surgery for acoustic neuromas, continuous facial nerve EMG monitoring has significantly improved the surgical outcomes.59 The ability to dynamically assess the integrity of the nerve and employ triggered stimulation for localization has transformed postoperative facial paralysis from a common complication to a controllable risk. This has raised patients’ expectations and surgical goals, emphasized the preservation of function, and made IONM an indispensable part of such surgeries.38 In these cases, the omission of IONM would deviate from contemporary best practices. When the tumor is located within or close to the functional area, the advantages of IONM will be better demonstrated.33, 60 For instance, in the surgery of supratentorial gliomas, direct cortical and subcortical stimulation is used for localization, which has evolved from a research technique to a key element of “maximum functional resection”. It directly supports the surgical concept of maximizing tumor resection while minimizing complications, which is also the core goal of modern neuro-oncology.61 Its application is closely related to prolonging progression-free survival and maintaining quality of life. This not only consolidates its status as a monitoring tool but also makes it a key therapeutic tool for defining the standard surgical approach for such lesions.62, 63 In summary, the clinical value of these established IONM techniques lies in their ability to actively prevent neural injury and enable more precise resection, thereby redefining the standard of care for complex neurosurgical procedures and significantly enhancing surgical feasibility (Fig. 2).Fig. 2Clinical impact: defining the standard of care. IONM has redefined safe neurosurgical practice by transitioning from passive monitoring to active neural protection. It enables radical yet safe resections in spinal, cranial base, and eloquent area tumors, making functional preservation a central, achievable goal. Clinical impact: defining the standard of care. IONM has redefined safe neurosurgical practice by transitioning from passive monitoring to active neural protection. It enables radical yet safe resections in spinal, cranial base, and eloquent area tumors, making functional preservation a central, achievable goal. Navigating the gray zones: controversies, challenges, and precision gaps The interpretation dilemma: from signal to action The core value of IONM lies not in the signals themselves, but in the correct interpretation of these signals and the subsequent acquisition of guiding information for clinical actions. The transition process from electrophysiological data to surgical decisions is regarded as the most significant and persistent challenge in this field, reflecting a fundamental gap in accuracy. The alarm threshold dilemma The debate over the optimal alarm threshold, especially for MEPs, is a typical example of this challenge. Although an amplitude reduction of more than 50% is a widely adopted warning standard, some advocate raising the threshold to above 80% to improve specificity and reduce false alarms.64 This debate highlights a key trade-off: a lower threshold increases sensitivity but may lead to unnecessary surgical pauses or interventions for transient, non-injurious changes (such as those caused by anesthesia fluctuations or mild hypotension). Conversely, a higher threshold improves specificity but may miss true injuries developing before potential irreversible damage occurs. The optimal threshold may also vary depending on the surgical type (such as spinal deformity surgery and intramedullary tumor surgery) and the suspected mechanism of injury (ischemic injury versus mechanical injury).65 The burden of false positives and false negatives Both of these two types of errors will reduce the accuracy of clinical judgment. False positive alerts (signal changes occurring in the absence of postoperative dysfunction) may lead to “alert fatigue”, resulting in unnecessary and potentially risky surgical interventions and prolonging the operation time. The reasons are diverse, including technical factors (such as electrical interference, electrode displacement), anesthetic effects, and physiological changes (such as hypotension, hypothermia), all of which can simulate injury phenomena.66 False negative results (although new functional disorders occur after surgery but the signal does not change) are relatively rare, but they are more harmful, potentially causing a false sense of security. This situation may occur on small and unmonitored pathways (such as the dorsal root entry zone, discrete nerve root injuries) or in cases where damage occurs after monitoring, such as postoperative hematoma.67, 68, 69 The challenge of the abnormal baseline For those patients who already have neurological functional impairments (such as due to tumor compression, spinal stenosis, or previous stroke), interpreting their abnormal signals is even more challenging. These patients typically have weaker or unstable baseline responses, making it difficult to determine significant changes and reducing the predictive value of standard alerting criteria. At the same time, this group constitutes a significant evidence gap, as most protocols are developed and optimized based on patients who were functionally normal before surgery.70 Distinguishing signal from noise The surgical environment is filled with electrical field interference. To distinguish between genuine neurophysiological changes and interference (such as electrocautery, irrigation, patient movement, or equipment failure), one must remain vigilant and possess professional judgment. This interpretation task falls on clinical neurophysiologists, whose experience becomes a crucial but unquantifiable factor in the monitoring process.18 In conclusion, “the interpretation challenge” further indicates that IONM is a complex clinical tool rather than an automated diagnostic test. The process of converting signals into effective actions requires not only standardized procedures but also complex clinical judgment, a deep understanding of interfering factors, and clear team communication to deal with inherent uncertainties. It is precisely this challenge that makes intraoperative magnetic resonance imaging, despite its many proven advantages, both an art and a science. Technical and logistical hurdles in complex scenarios Although the principles of IONM have been largely established, there are still technical and logistical challenges in reliably applying these principles in complex or prolonged surgeries. These obstacles often test the limits of equipment, personnel and procedures, and may even undermine all the benefits that IONM is intended to provide. The burden of long-duration surgeries Long-term surgeries, such as those for complex spinal deformities, large intramedullary tumors, or extensive cranial base lesions, impose a significant burden on the entire monitoring system. The electrode impedance may drift, necessitating time-consuming adjustments. At the same time, both patients and monitoring personnel will become fatigued, thereby reducing signal quality and lowering the alertness for interpretation. Strictly adhering to anesthesia protocols over such a long period also becomes a challenging task, thereby increasing the risk of signal fluctuations due to pharmacological drift. The “cable confusion” of multimodal monitoring To comprehensively monitor complex cases, it is usually necessary to simultaneously record SSEPs, MEPs, multiple cranial nerve EMG, and sometimes even EEG or brainstem auditory evoked potentials (BAEPs). However, the large number of electrodes, stimulators and cables generated in such situations impose a significant burden on logistics. It also increases setup time, the risk of displacement or interference, and may hinder the surgical process during actual operations. To solve this “cable chaos” problem, detailed planning and coordination among the surgical team, anesthesia team, and neurophysiology team are required.71 Resource intensity and team dynamics To fully utilize the effectiveness of neural monitoring, experienced clinical neurophysiologists are required to be specifically in charge and to explain the process. However, in resource-poor environments or during complex surgeries, this specialized human resource may be overused, resulting in delayed interpretation or distraction. Moreover, the success of invasive neural monitoring also depends on seamless and real-time communication within the surgical team. Mistakes in this communication - such as alarms not being clearly heard, understood, or executed - represent serious logistical problems and may lead to the failure of technical monitoring.72, 73 Anatomical and pathological limitations In some more complex scenarios, the monitoring effectiveness of IONM can be severely limited. For instance, in surgeries targeting lesions in the cerebellum or brainstem, the tight interconnection of neural pathways makes it difficult to distinguish specific functions. Therefore, damage to small and crucial nerve nuclei may not be detectable through standard cortical-spinal or somatosensory pathway monitoring.74, 75 Similarly, in surgeries for vascular malformations (such as arteriovenous malformations (AVMs)), the pathophysiological process involves hemodynamic changes and compensatory phenomena, and the interpretation of changes in MEPs/SSEPs may become ambiguous because they may reflect reversible vascular movements rather than irreversible ischemia.76, 77, 78 In conclusion, the technical and logistical obstacles encountered in complex scenarios indicate that the reliability of IONM is not solely guaranteed by technology. It can be said that IONM is a resource-intensive task that requires continuous technical stability, expert human supervision, and comprehensive team logistical support. Solving these problems is crucial for converting the theoretical advantages of IONM into stable and high-quality treatment effects applicable to all complex situations in modern neurosurgery. The next frontier: from monitoring to predictive guidance—a critical look at emerging paradigms Artificial intelligence and machine learning Artificial intelligence (AI) and machine learning (ML) are among the most promising and controversial cutting-edge fields in IONM.79 The basic concept is to deploy algorithms that can analyze the massive and multi-variable intraoperative data streams (such as electrophysiological signals, anesthesia parameters, vital signs, and key surgical moments) and extract details that observers cannot perceive. The goal is to achieve a higher level of integration, thereby enabling true prediction and precise guidance. However, the path to becoming a reliable clinical tool is fraught with obstacles.80, 81 The promise: from descriptive analytics to predictive power Current IONM practice are mainly descriptive and diagnostic, while AI/ML aims to make it predictive and prescriptive. The potential directions for efforts include: 1) real-time signal interpretation and artifact elimination: the algorithms trained on massive datasets could quickly distinguish between genuine neurophysiological changes from artifacst caused by electrocautery or anesthetic drift, thereby reducing false alarms and alleviating the cognitive burden on readers; 2) multimodal data fusion and risk stratification: by continuously integrating MEPs/SSEPs trends with hemodynamic data, surgical stages, and patient-specific factors, the ML model can generate dynamic and personalized risk scores for neural injury, enabling the team to detect an increase in risk levels before individual parameters exceed traditional thresholds; 3) outcome prediction: analyzing the entire intraoperative electrophysiological “features” can not only predict whether there are defects, but also predict the possible severity and nature of the postoperative defects, thereby providing guidance for the immediate rehabilitation plan.82 The reality: the gap between algorithm and application While AI/ML holds the promise of transforming IONM from a descriptive to a predictive and prescriptive guidance system, this transition is fraught with significant hurdles (Fig. 3). The primary challenges can be summarized as follows:1) Data aspect: Effective machine learning requires a large amount of high-quality, carefully labeled data. However, most IONM data currently exist in isolated and non-standardized forms. Therefore, creating the necessary “data sets”, that is, linking intraoperative signals with corresponding surgical events and validated postoperative results, is a necessary but arduous and resource-intensive task. 2) Algorithmic aspect: powerful ML models, especially deep learning networks, have an opaque decision – making process. In high - risk surgical environments, alerts issued by algorithms that are difficult to understand may not prompt people to take prompt and decisive actions. Explainable AI (XAI) is a crucial subfield that must be continuously improved as predictive models evolve.83 3) Universality: models trained based on data from a certain institution, specific protocols and equipment may suffer catastrophic failures in different environments. Ensuring their stability in different surgical populations, pathological conditions and technical settings is a significant challenge.84 4) Clinical applicability: Transitioning from proof-of-concept studies to prospective clinical trials that demonstrate improved patient treatment outcomes is a crucial step. Additionally, integrating real-time artificial intelligence systems into existing surgical room workflows and achieving seamless human-machine interaction also present their own challenges in terms of usability and ergonomics.Fig. 3Artificial intelligence and machine learning in IONM. AI/ML holds the potential to transform IONM from a descriptive tool into a system for predictive guidance, yet it remains an auxiliary tool, not a replacement for expert judgment. Significant hurdles in data quality, model interpretability, and clinical integration must be overcome for its full realization. Artificial intelligence and machine learning in IONM. AI/ML holds the potential to transform IONM from a descriptive tool into a system for predictive guidance, yet it remains an auxiliary tool, not a replacement for expert judgment. Significant hurdles in data quality, model interpretability, and clinical integration must be overcome for its full realization. In conclusion, although artificial intelligence/machine learning has the potential to fundamentally transform IONM technology into an intelligent guidance system, it is still in the early stages of development. In the foreseeable future, the role of artificial intelligence is likely to be that of providing a powerful auxiliary tool for clinical neurophysiologists, rather than an alternative to expert judgment. Multimodal integration and data fusion: building the surgical “dashboard” The research of IONM is moving towards a fully integrated intraoperative monitoring and guidance system, which is conceptualized as a surgical “dashboard” or “digital cockpit”, aiming to go beyond the current practice of displaying different data streams on separate screens. Instead, it aims to integrate these elements into a real-time, context-aware model of the patient’s functional anatomy. The goal is to provide a more comprehensive and accurate overall view, in order to enhance the surgical team’ spatial understanding, clarify the potential causes of electrophysiological changes, and provide support for more informed and precise decision-making.85, 86 This dashboard will seamlessly integrate core neurophysiological trends (such as SSEPs, MEPs, EMG), preoperative and intraoperative imaging structures and navigation data, key vascular information from auxiliary means (such as indocyanine green video angiography), and carefully organized systemic physiological parameters.87 Crucially, it should also include a timeline of the surgical stage. The power of this integration lies in its interrelation; for instance, the gradually weakening motor evoked potentials can be immediately matched on the dashboard with the position of the nearby motor nerve bundles on the navigation instrument and the simultaneous drop in blood pressure. This instantaneous comprehensive analysis helps to distinguish mechanical stimulation from systemic ischemia, transforming interpretation from speculation to guided insight.26 However, the current stage of this field is still at the conceptual stage. This is because numerous technical barriers have restricted its implementation, including the lack of universal data standards, the challenges of real-time data flow architecture, and the absence of multi-dimensional data fusion platforms. Therefore, successfully building this dashboard is more of a complex system engineering challenge rather than a pure data science issue. Furthermore, the human-machine interface remains a challenge. The design of the dashboard must be carefully considered to enhance the cognitive and decision-making capabilities of the surgical team. Excessive integration of information may lead to cognitive overload. How to design this interface intuitively, adapt it to individual surgical stages, and highlight the most critical comprehensive alerts are all aspects that need to be taken into account. Its ultimate value does not depend on the amount of data displayed, but on its comprehensibility and operability in a high-risk surgical environment. In summary, multimodal fusion is a reasonable development of IONM towards a coherent situational awareness system. Building an effective surgical dashboard requires close interdisciplinary collaboration among neurosurgeons, neurophysiologists, biomedical engineers, and interface designers to ensure that the integrated data flow truly illuminates the surgical path ahead and transforms raw information into actionable intelligence. Towards closed-loop systems and novel biomarkers The goal of predictive analysis is to develop an intelligent closed-loop system that can interpret data and implement calibration intervention measures to maintain the integrity of the nervous system.88 There is also an increasing interest in seeking new biomarkers beyond traditional IONM, aiming to monitor the brain and spinal cord at the metabolic, hemodynamic and molecular levels.89, 90 In IONM, the concept of a closed-loop system represents a paradigm shift from “monitoring and alerting” to “perception and response”. It would continuously analyze and integrate data streams and automatically initiate corrective measures once patterns indicative of damage are detected. Theoretically, this could involve adjusting the infusion rate of vasoconstrictors during spinal deformity correction surgery, or adjusting the depth of anesthesia based on signs of cortical stimulation. The potential lies in eliminating human response time and interpretation differences, thereby enabling precise and timely responses to constantly changing situations. However, achieving a clinically feasible and reliable closed-loop system presents numerous formidable challenges. This requires not only the predictive AImodels mentioned earlier, but also their seamless integration with therapeutic actuators (such as infusion pumps) into a secure and reliable control architecture. Medical legal and ethical considerations are of great significance, and the algorithmic decision-making process must be absolutely deterministic—a standard far beyond the current capabilities. In the foreseeable future, the most likely application is in decision support systems, providing intervention suggestions to the clinical team while retaining the role of human supervision and judgment, and combining it with computing speed and pattern recognition. While pursuing automated control, people are also exploring new biomarkers to fill the blind spots of traditional IONM. The standard methods mainly assess the integrity of large and rapid conducting pathways, but are insensitive to damage responses affecting small fibers, local metabolism, or the neurovascular unit. Emerging technologies aim to fill these gaps. Intraoperative imaging biomarkers, such as real-time ultrasound or laser speckle contrast imaging, can directly present blood flow and tissue perfusion conditions. Optical techniques such as near-infrared spectroscopy (NIRS) and diffuse correlation spectroscopy (DCS) can non-invasively measure the oxygenation and blood flow of the brain, potentially issuing ischemia warnings earlier than electrophysiological signals.91, 92 Perhaps the most forward-looking and transformative area is the exploration of molecular and metabolic biomarkers. Research is investigating the possibility of using microdialysis technology to detect changes in the neurochemical environment within neural tissues (such as glutamate excitotoxicity, lactate/pyruvate ratio), or by applying advanced biosensors.93 Although these methods are currently limited to the research environment, they are expected to detect damage at the cellular level before electrophysiological failure or structural damage occurs, thus opening a true window of preventive opportunity. In conclusion, the research on closed-loop systems and new biomarkers represents the long-term vision of neuro-monitoring technology. Despite significant obstacles, the direction is clear: the future of intraoperative neuro-protection lies in creating a more integrated, intelligent, and biologically comprehensive monitoring environment. Synthesis and future roadmap: toward a mature discipline The IONM process documented in this study reveals a field that is maturing yet not fully mature. It has evolved from a research-based technology to an evidence-based standard of care for numerous high-risk neurosurgical and spinal surgeries. The established pillars—multimodal monitoring combining MEPs, SSEPs, and EMG, supported by an optimized anesthesia plan—have significantly transformed surgical practice, enabling more aggressive resections while reducing neurological complications. Simultaneously, the exploration of gray areas and emerging paradigms highlights the frontiers the discipline must advance to fully realize its potential. Looking to the future, the maturation of IONM into a robust, predictive, and comprehensive clinical science hinges on addressing three interdependent imperatives: standardization, data-driven discovery, and expanded collaboration, as outlined in the proposed roadmap (Fig. 4).Fig. 4Synthesis and Future Roadmap. IONM has evolved from a research tool into an evidence-based standard of care, enabling aggressive resections while reducing complications. To mature into a predictive, comprehensive science, it must solidify its standards, leverage data through AI, and achieve deep multidisciplinary integration within the surgical team. Synthesis and Future Roadmap. IONM has evolved from a research tool into an evidence-based standard of care, enabling aggressive resections while reducing complications. To mature into a predictive, comprehensive science, it must solidify its standards, leverage data through AI, and achieve deep multidisciplinary integration within the surgical team. The current and primary task is to establish and widely adopt strict, evidence-based standards. This not only includes the general acceptance of monitoring for certain surgical procedures, but also the nuances of practice. The field needs to reach consensus on standardized alert standards, which must be validated and applicable to specific surgical environments and patient populations, including those with abnormal underlying conditions. It also needs to establish standardized training and qualification certification pathways for clinical neurophysiologists to ensure the professionalism of their interpretation capabilities. Additionally, standardized data collection, storage, and reporting protocols are not only crucial for clinical consistency but also the foundational cornerstone of the next stage of research. Without such a degree of standardization, the discipline may still remain at the stage of relying on expert practice and cannot develop into a repeatable and verifiable science. The second key requirement is to adopt a data-driven, transformation research-oriented model. The controversies in interpretation and the potential of artificial intelligence reveal a crucial fact: the vast amount of data generated in each monitoring case is still largely underutilized. The future roadmap must include the creation of large, multi-center, and meticulously organized databases, linking high-fidelity surgical process signals with detailed surgical procedures, anesthesia records, and meticulous postoperative results. This infrastructure will provide impetus for the development of MLmodels, enabling us to shift from passive threshold responses to personalized and predictive risk analysis. It will also make it possible for the rigorous clinical validation of new biomarkers and integrated dashboard systems, transforming their evaluation from proof of concept to proven clinical utility. Research must also focus on cost-effectiveness and implementation science to ensure that these advancements are accessible and sustainable. Finally, to reach a mature state, a profound transformation is required, moving towards deep, structured multidisciplinary integration. IONM cannot be an isolated service; it must become an intrinsic component of the surgical team’s cognitive workflow. This includes jointly developing closed-loop decision support systems centered on human factors, creating intuitive interfaces, integrating data into actionable insights while avoiding information overload. This requires the collaboration of neurosurgeons, neurophysiologists, anesthesiologists, engineers, and data scientists from the very beginning to jointly design the next generation of tools and solutions. The ultimate goal is to eliminate disciplinary boundaries and establish a common mindset, in which the preservation of function becomes a unified, real-time objective, achieved through the utilization of all available data and technologies. In summary, the development roadmap of IONM foresees a future scenario: it will no longer merely be a functional monitoring tool, but will become an intelligent guide for protecting functions. By establishing its standards, unleashing the power of data, and fully integrating with the structure of the surgical team, IONM can surpass its current role. It has the potential to evolve into a mature discipline, not only protecting the nervous system during surgery, but also actively shaping and optimizing surgical strategies, ensuring that the pursuit of curative treatment always remains consistent with the necessity of protecting the essence of the patient. CRediT authorship contribution statement Ke Peng: Writing – review & editing, Investigation, Data curation, Conceptualization. Fuhai Ji: Writing – review & editing, Supervision. Nazneen Sudhan: Writing – review & editing, Supervision, Conceptualization. Fei Xu: Writing – original draft. Yu’e Sun: Writing – original draft. Xiaqing Ma: Supervision, Funding acquisition, Conceptualization. All authors have read and agreed to the published version of the manuscript. Consent for publication Not applicable. Ethical statement Not applicable. Declaration of generative AI and AI-assisted technologies in the writing process During the preparation of this work the authors used jova.ai in order to present the Figures in a more detailed and precise manner. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article. Funding This work was supported by the General Program of the National Natural Science Foundation of China to X.M. (82201366); Jiangsu Provincial Research Hospital: YJXYY202204-YSB33 to X.M. And Jiangsu Health International Exchange Program to X.M. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Fuhai Ji is an Associate Editor for this journal and was not involved in the editorial review or the decision to publish this article.