
==== Front
J Korean Neurosurg Soc
J Korean Neurosurg Soc
JKNS
Journal of Korean Neurosurgical Society
2005-3711
1598-7876
Korean Neurosurgical Society

38186369
10.3340/jkns.2023.0195
jkns-2023-0195
Review Article
Neurotrauma
Artificial Intelligence-Enhanced Neurocritical Care for Traumatic Brain Injury : Past, Present and Future
http://orcid.org/0009-0003-3472-9991
Kim Kyung Ah 1
http://orcid.org/0000-0001-5407-293X
Kim Hakseung 1
http://orcid.org/0000-0003-3278-0550
Ha Eun Jin 2
http://orcid.org/0000-0003-2929-5677
Yoon Byung C. 3
http://orcid.org/0000-0002-0988-2236
Kim Dong-Joo 1 4
1 Department of Brain and Cognitive Engineering, Korea University, Seoul, Korea
2 Department of Critical Care Medicine, Seoul National University Hospital, Seoul, Korea
3 Department of Radiology, Stanford University School of Medicine, VA Palo Alto Heath Care System, Palo Alto, CA, USA
4 Department of Neurology, Korea University College of Medicine, Seoul, Korea
Address for reprints : Dong-Joo Kim Department of Brain and Cognitive Engineering, Korea University, 145 Anam-ro, Seongbuk-gu, Seoul 02841, Korea Tel : +82-2-3290-5929, Fax : +82-2-3290-3970, E-mail : dongjookim@korea.ac.kr
9 2024
8 1 2024
67 5 493509
6 9 2023
18 12 2023
4 1 2024
Copyright © 2024 The Korean Neurosurgical Society
2024
https://creativecommons.org/licenses/by-nc/4.0/ This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
In neurointensive care units (NICUs), particularly in cases involving traumatic brain injury (TBI), swift and accurate decision-making is critical because of rapidly changing patient conditions and the risk of secondary brain injury. The use of artificial intelligence (AI) in NICU can enhance clinical decision support and provide valuable assistance in these complex scenarios. This article aims to provide a comprehensive review of the current status and future prospects of AI utilization in the NICU, along with the challenges that must be overcome to realize this. Presently, the primary application of AI in NICU is outcome prediction through the analysis of preadmission and high-resolution data during admission. Recent applications include augmented neuromonitoring via signal quality control and real-time event prediction. In addition, AI can integrate data gathered from various measures and support minimally invasive neuromonitoring to increase patient safety. However, despite the recent surge in AI adoption within the NICU, the majority of AI applications have been limited to simple classification tasks, thus leaving the true potential of AI largely untapped. Emerging AI technologies, such as generalist medical AI and digital twins, harbor immense potential for enhancing advanced neurocritical care through broader AI applications. If challenges such as acquiring high-quality data and ethical issues are overcome, these new AI technologies can be clinically utilized in the actual NICU environment. Emphasizing the need for continuous research and development to maximize the potential of AI in the NICU, we anticipate that this will further enhance the efficiency and accuracy of TBI treatment within the NICU.

Artificial intelligence
Clinical decision support
Critical care
Machine learning
Intensive care units
Traumatic brain injury
==== Body
pmcINTRODUCTION

Traumatic brain injury (TBI) is one of the leading causes of death and disability worldwide, with the estimated, global incidence of 27–69 million cases annually [8]. TBI can result in a range of progressive long-term physical and psychosocial impairments, including difficulties with cognitive abilities such as attention and memory, neurological symptoms such as headaches and dizziness, neuropsychiatric disorders, and an increased likelihood of neurodegenerative diseases later in life [120].

TBI can be classified into two types : primary and secondary. The former refers to the immediate consequences of the initial trauma, and as such, these effects are irreversible. Secondary brain injury refers to a series of events triggered by physiological responses following initial damage that may result in additional brain injury and worsen patient prognosis [81]. Therefore, the primary objective of TBI management in neurointensive care units (NICUs) is to prevent, evaluate, and treat these secondary insults. This involves implementing measures such as maintaining sufficient oxygenation and blood pressure, controlling intracranial pressure (ICP) and cerebral perfusion pressure (CPP), and providing appropriate nutritional support [57,121].

Despite major advances in our understanding of TBI, numerous challenges persist in managing this complex and heterogeneous condition. One of the primary challenges arises from the heterogeneity of injury mechanisms, patient characteristics, and injury severity. As a result, there is a pressing need for patient-specific treatments [10,74]. To determine the most suitable treatment strategies, it is essential to take into account the degree of TBI severity, specific areas of the brain affected, and unique physiological responses. Artificial intelligence (AI) has the potential to play a critical role in meeting these needs by providing tools for personalized treatment plans and decision support [97]. Particularly in the management of TBI, numerous studies are being conducted, exploring the diverse application possibilities of medical AI. It can perform tasks such as classification, prediction, and segmentation of TBI patients, thereby offering more precise diagnoses and efficient treatment directions. Table 1 lists some of these studies, demonstrating the specific roles AI can play in TBI management. This table summarizes various research examples on how AI can contribute in phases like initial management, NICU management, and neuroprognostication for TBI patients, thus providing insights into the clinical applicability of AI.

AI-enhanced neurocritical care has the potential to significantly enhance the treatment outcomes and quality of life for patients with TBI. Although AI techniques have rapidly gained acceptance within the neurosurgical community [87,114], the same level of acceptance does not extend to the utilization of AI for TBI management. This brief review provides the core concepts of machine learning (ML) and an overview of ML applications in NICUs with a focus on patients with TBI. It further explores the use of real-time AI in the NICU environment and concludes by investigating the prospective roles of generalist medical AI (GMAI) and digital twins (DT) in the NICU.

ML FOR NEUROCRITICAL CARE DATA ANALYSIS

Core concepts of ML

ML is a subfield of AI that enables computers to identify patterns and rules inherent in data without being explicitly programmed [87]. A key characteristic of ML models is their ability to continuously improve their performance through experience. Although more recent ML techniques, such as reinforcement learning and generative learning, have yielded promising results in certain areas (e.g., surgical planning [26] and generating neuroimages [29], etc.), the conventional tasks of medical AI have primarily focused on two main approaches : unsupervised learning for identifying undefined patterns or reducing data dimensionality and supervised learning for classification and regression tasks [111]. While both techniques can be used concurrently, the primary application of ML in medical fields has been centered around supervised classification and regression for image classification and segmentation, diagnosis, and prediction of outcome and complications [114].

While it is true that most of these ‘conventional’ tasks could theoretically be achieved through statistical methods, such statistical models rely on specific assumptions regarding data distribution, which may not be valid in numerous situations [135]. Specifically, if the data have high dimensionality (i.e., many variables) and/or if the relationship between data points is nonlinear, statistical assumptions are likely to fail. Nonlinearity and high dimensionality are common in medical data, making ML particularly advantageous. Logistic regression (LR), naïve bayes, decision trees, K-nearest neighbors, random forest (RF), and support vector machine (SVM) are commonly used classification algorithms in the medical field [6,11,33,123]. Recently, the utilization of ML has been increasing in the medical field, not only in clinical practice but also in personalized treatment [71,110]. This trend has been further accelerated by the emergence of various decision tree variants and deep learning technologies [9,45,64,69,113,142].

Role of ML in pre-admission data analysis

It is crucial for patients with TBI to predict the risk factors and perform early diagnosis during the prehospitalization process [34]. In addition, it is essential to optimize the triage process by appropriately classifying patients based on the severity of brain injury for treatment or transportation. In this regard, ML may enable a rapid and accurate diagnosis and treatment planning for patients with TBI.

The early recognition of TBI is particularly important because it effectively improves patient outcomes [35,134,137]. Choi et al. [19] used pre-hospital information from emergency medical service (EMS) providers, such as the EMS ambulance run sheet and EMS trauma in-depth registry, to identify the risk factors. Through the application of various ML techniques, including the elastic net algorithm, which exhibited the best performance, we identified that the three most significant predictors of TBI-related outcomes were loss of consciousness, Glasgow coma scale (GCS) score, and light reflex. In a study conducted by Hale et al. [41], an artificial neural network (ANN) using computed tomography (CT) data was employed to predict the likelihood of clinically relevant TBI in pediatric patients aged <18 years who experienced non-penetrating head trauma.

Accurately classifying patients with TBI according to their trauma severity before or immediately after transfer to the emergency room is crucial for efficient early treatment and has a significant impact on their prognosis [1]. Abe et al. [1] aimed to detect traumatic intracranial hemorrhage (tICH) and differentiate patients with tICH from those without tICH using information collected by on-site EMS personnel, such as systolic blood pressure, heart rate, body temperature, and respiratory rate. In the testing set, extreme gradient boosting (XGBoost), a variant of decision tree, showed the best predictive performance among the various ML algorithms with an area under an area under the receiver operating characteristic curve (AUC) of 0.80. Moyer et al. [85] developed a predictive model to identify patients requiring emergency neurosurgery among those diagnosed with moderate-to-severe TBI during the prehospital assessment process. Fifteen prehospital predictors, including the GCS, initial systolic blood pressure, initial diastolic blood pressure, and initial oxygen saturation, were used. The CatBoost model, another decision tree variant, outperformed the other models with an AUC of 0.81.

Optimizing surgical timing with ML

Once TBI patients are admitted to the hospital, it is important to quickly identify which patients require surgical intervention and to intervene at the right time. ML can contribute to this treatment intervention process, helping medical professionals make better decisions. A number of studies have been conducted to classify the severity of patients and identify patients who need surgical intervention. For instance, Güler et al. [39] developed a classification model to rapidly evaluate the severity of TBI in emergency department admitted patients. An ANN model was used to classify the severity of TBI using trauma scores, GCS, and electroencephalography (EEG) data. The model showed high agreement with the neurologists’ decisions, with an 87% concordance rate. Habibzadeh et al. [40] focused on those with positive initial cranial CT scan results for moderate TBI, developing a model to differentiate between patients needing neurosurgical intervention and those who do not, based on age, gender, GCS, Marshall score, presence of hematoma, and midline shift score. The SVM model performed best with an F1 score of 0.83 and an AUC of 0.93. Currently, TBI patients take follow-up CT scans after initial scans to assess intracranial damage progression, leading to concerns over unnecessary radiation exposure. In this context, this research shows potential as an effective decision-making tool for predicting intracranial damage and identifying patients in need of medical intervention without follow-up CT scans.

ML is not just limited to patient classification, but also plays an important role in determining the appropriate time for medical intervention according to the individual patient’s situation [47,56,77,124]. Some studies explore the general consequences of delayed neurosurgical intervention regardless of type or patient situation of the surgery, while others focus on detailing the appropriate time of intervention for specific surgical procedures and situations. Hanko et al. [42] predicted mortality and functional outcomes for TBI patients who received primary decompressive craniectomy (DC) treatment, considering a variety of variables including medical records, clinical symptoms, and pre- and post-operative conditions. The RF-based predictive model used in this study achieved an AUC of 0.811 for 6-month mortality and 0.873 for outcomes, with surgery timing being an important variable. It concluded that early implementation of primary DC is essential for effectively relieving ICP and ongoing brain hemorrhage. Similarly, a study by Seelig et al. [109] examined the outcomes of patients in a coma due to traumatic acute subdural hematoma, investigating various variables. It found that delay in surgery was a significant factor, particularly noting that patients operated within 4 hours of injury had a 30% mortality rate, while those operated after 4 hours had a mortality rate rising to 90%.

Outcome prediction via ML

Seemingly moderate or minor cases of TBI can rapidly escalate to severe TBI, necessitating close surveillance [119]. The generated variables and parameters offer valuable insights into prognostic assessments. Indeed, outcome prediction after TBI has long been a major topic in neurocritical care [72], and the increasing availability of ML techniques has stimulated their active incorporation into this topic. Tu et al. [128] developed a predictive model for mortality risk using the Taiwan triage and acuity scale and 11 characteristic variables, including age, sex, and obesity. The LR algorithm exhibited the best predictive performance with an AUC of 0.925. Another study by Tunthanathip and Oearsakul [129] applied ML to predict functional outcomes in pediatric patients with TBI. GCS, hypotension, pupillary light reflex, and subarachnoid hemorrhage were established as important factors for predicting functional outcomes. Among the classification algorithms, SVM showed the highest performance, with an accuracy of 0.94. Finally, Farzaneh et al. [32] proposed an interpretable ML-based framework for predicting long-term functional outcomes in patients with TBI. This study implemented the XGBoost classifier using 18 electronic health record (EHR) variables. The final model achieved an accuracy of 0.7488 on the test set. The advantage of this study is that it provides SHAP (SHapley Additive exPlanations) contribution, allowing for easier interpretation of the model in medical decision-making, particularly for predicting long-term functional outcomes in patients with TBI.

ML for high-resolution data analysis during intensive care

Management of severe TBI often necessitates continuous monitoring of various signals such as ICP, arterial pressure, and electrocardiography (ECG). The sheer scale, nonlinearity, and high dimensionality of the resulting data warrant analysis using ML. ML has undoubtedly demonstrated great potential for analyzing continuous neuromonitoring data from patients with TBI, allowing for the prediction of expected outcomes [116,119,125].

First, EEG is actively used to predict outcomes in patients with TBI [7,82]. For example, Haveman et al. [43] used quantitative EEG (qEEG) to predict the outcomes of patients with moderate-to-severe TBI 12 months after injury. This was accomplished by creating predictive models using an RF classifier based on qEEG features, age, and mean arterial blood pressure (MAP), which were collected over 7 days post-admission. The model showed an AUC of 0.94 on the training set and 0.81 on the validation set. Similarly, Noor et al. [88] used the absolute power spectral density (PSD) in each EEG frequency band to predict the outcomes of patients with moderate TBI 12 months after injury. In this study, Random under-sampling boosted trees were used. It was found that the absolute PSD in the delta and gamma bands was the best outcome predictor. They demonstrated an AUC of 0.97 and 0.95, respectively.

In addition, ICP signal is a valuable tool that can be used to predict patient prognosis. Raj et al. [96] developed a dynamic mortality prediction model for patients with TBI in the NICU based on ICP and MAP. Dynamic features related to ICP, MAP, CPP, motor response, and eye response were generated. These were then used in a LR model to predict the mortality rate at 30 days post-injury. The algorithm achieved a maximum AUC of 0.84, demonstrating superior performance compared to IMPACT-TBI [118], which is one of the most widely accepted static prediction models with an AUC of 0.78. Pimentel et al. [93] also attempted to predict the mortality rate of patients with TBI by extracting dynamic features from ICP and MAP signals. A Gaussian process (GP) framework was used to extract dynamic features from physiological signals. This method has the advantage of performing well under intermittently disrupted monitoring conditions, which are common in the NICU. By combining GP-based and pressure reactivity index (PRx) features, the ability to predict patient outcomes significantly improved, achieving an AUC of 0.76.

REAL-TIME AI APPLICATIONS IN THE NICU ENVIRONMENT

In complex and unpredictable environments such as the NICU, the speed and accuracy of medical decisions are critical [102,112]. The issue of rapidity and accuracy in the clinical decision-making process is further compounded by the sheer amount of data generated during neuromonitoring and altered levels of consciousness. The Clinical decision support system (CDSS) [86] is intended to improve healthcare delivery by enhancing medical decisions with targeted clinical knowledge, patient information, and other health information [90]. AI plays a variety of roles in CDSS, such as monitoring and analyzing a patient’s condition in real time and predicting future events. AI helps clinicians to quickly recognize situations and make appropriate medical decisions by analyzing large amounts of data in a complex environment.

Signal quality control

False alarms are prevalent in the NICU, adversely impacting staff alarm responsiveness, degrading the quality of care, and indirectly affecting patients by disrupting their sleep quality [23,25,61,126]. Because of the complexity of neurophysiological signals, simple filter-based artifact rejection techniques are often insufficient. Thus, several studies have proposed methods to minimize false alarms by identifying and removing artifacts that are the main cause of false alarms or by improving alarm accuracy [4,5,64,66,113,127,132] (Table 2). The primary benefit of artifact elimination would be improved signal quality and a subsequent reduction of false alarms, and signal quality control may lead to significant changes in the prognostic values of essential neurophysiological signals. For instance, Kim et al. [55] reported that the association between worse outcomes and the prevalence of hypotension significantly increased after the elimination of artifacts in arterial pressure.

Event prediction

Intracranial hypertension (IH) is common during the acute phase of TBI and can worsen secondary insult [95]. The prediction of IH could allow rapid intervention and, thus, result in a better outcome; hence, it has long been considered an important topic in the neurocritical care fields [12,63]. Schweingruber et al. [108] reported the performance of long short-term memory (LSTM) models in predicting significant events at various time intervals ranging from 1 to 24 hours. These models were trained using demographic variables and were continuously monitored for arterial blood pressure (ABP) and ICP. The model showed robust performance even at a distance of 24 hours, with an AUC of 0.826. However, the most clinically relevant prediction was made 2 hours before the significant event, achieving an AUC of 0.953. Additionally, this study defined two phases of IH : ‘long phase IH,’ lasting more than 2 hours with IH above 22 mmHg, and ‘short phase IH,’ lasting less than 2 hours. The LSTM model predicted the occurrence of both situations at a distance of 2 hours, resulting in AUC of 0.953 and 0.728 for the long and short phases, respectively. Similarly, Güiza et al. [38] developed a model that could predict increased ICP episodes lasting >10 minutes at 30 mmHg, 30 minutes in advance, using only the dynamic characteristics of continuous ICP and MAP. The prediction model was built using multivariate LR and a GP, achieving a classification accuracy of 77%, sensitivity of 82%, and specificity of 75% on the validation set. In addition to predicting simple IH events, Lee et al. [63] focused on predicting ‘life-threatening’ intracranial hypertension (LTH) specifically. In this study, episodes of IH were classified as LTH if they were characterized by a PRx exceeding 0 and a pressure-time dose greater than 5. This study aimed to predict whether ongoing IH conditions would escalate to LTH within 5 minutes. CatBoost was used to predict using ABP and ICP-related parameters. This model achieved an AUC of 0.73 for predicting LTH on the test dataset. Furthermore, numerous studies have developed models using various variables and algorithms to predict ICP episodes early [16,52,75,107,136] (Fig. 1).

Minimally invasive neuromonitoring

Simple clinical assessments may not detect subtle changes in brain physiology or alterations in the neurological status that could appear at a later stage [115]. This is particularly true for TBI, where the patient’s condition may rapidly deteriorate, necessitating the integration of clinical assessments with other neuromonitoring techniques [112]. Even ICP (and CPP to some extent), which has conventionally served as a core measure in TBI, has yielded conflicting results concerning its prognostic value [18,138]. Over the past decade, multimodal neuromonitoring has garnered increasing attention. For example, studies have shown that integrating brain tissue oxygen tension-based treatments with conventional ICP/CPP-oriented protocols leads to a better neurological prognosis and lower death rates [58,99]. Cerebral microdialysis can also provide a better understanding of hypoperfusion and metabolism when complemented by other modalities [74]. Thus, multimodal neuromonitoring provides a more accurate understanding of each patient’s individual circumstances and conditions, paving the way for the delivery of individualized therapy [74]. However, most multimodal monitoring techniques are invasive, resulting in a range of problems. For example, invasive ICP monitoring using a ventricular catheter carries the risk of numerous complications including infection and intracranial hemorrhage [31,112]. Cerebral microdialysis is not an exception to these concerns as it has additional problems such as tissue inflammatory response and biofouling [115]. Therefore, many previous studies have investigated neuromonitoring using minimally invasive or non-invasive methods [3,106,131]. In addition, with the development of AI, it has become possible to non-invasively acquire and analyze accurate signals.

Various technologies spanning fluid dynamics, ophthalmic, otic, and electrophysiologic fields have been proposed for non-invasive ICP monitoring, including magnetic resonance imaging (MRI), transcranial Doppler ultrasound imaging, optic nerve sheath diameter assessment, and others. One of the most traditional methods for non-invasive ICP estimation is the use of cerebral blood flow velocity (CBFV) and ABP [48,51,53]. According to Imaduddin et al. [48], a Pseudo-Bayesian model was used to estimate ICP using non-invasive measurements such as ABP and CBFV. This model showed a mean absolute error (MAE) of 0.6 mmHg for mean ICP estimation, compared to measurements taken invasively. Similarly, Kashif et al. [53] developed a patient-specific model-based algorithm to estimate ICP using routinely collected ABP and CBFV signals, eliminating the need for calibration. Using 35 hours of data from 37 TBI patients, the model achieved ICP estimates with a MAE of 1.6 mmHg and standard deviation of 7.6 mmHg. The best performance of the model was a sensitivity of 90%, a specificity of 80%, and an AUC of 0.88, indicating improved accuracy and reduced variability compared to previous studies [89]. Recently, attempts have been made to estimate the ICP using ECG. In 2021, Sadrawi et al. [105] conducted a study using ECG and a U-Net-based deep convolutional autoencoder system to noninvasively estimate the mean ICP. The mean ICP evaluation had an MAE of 2.404±0.043 mmHg, and Pearson’s linear correlation coefficient was 0.89. Although the results were inferior to those of previous studies that estimated ICP using CBFV and ABP, they are still valuable because they use completely non-invasive ECG signals.

Unfortunately, there are a number of challenges to applying those noninvasive methods in clinical practice. Almost all review articles on this topic have concluded that it is impossible to estimate ICP noninvasively in all situations, for all patients, and over a long period of time without calibration [31,59,94,100,103,141]. For instance, CBFV not only exhibits high intra- and inter-user variability but is also significantly affected by other physiological changes such as medications, and autoregulation. These factors compromise the reliability of CBFV as a source signal, furthermore, making it challenging for long-term continuous use. Additionally, other non-invasive ICP monitoring measures also have distinct limitations such as insufficient validation, inability for continuous monitoring, poor signal quality, variations due to different operator expertise, and inter-observer variability. For example, neurological pupillary index measurement using pupillometry has currently attracted much attention in research and clinical practice [17,49,91], but there is still controversy about its suitability for continuous monitoring, accuracy in reflecting changes in ICP, and precision in estimating ICP [70,101,117].

Despite these limitations, noninvasive methods of measuring ICP can be a viable and attractive alternative for patients who are not amenable to invasive monitoring. Continuous improvements and validations are being made in non-invasive methods, and particularly, the development of multimodal non-invasive neuromonitoring technologies that integrate various methods is expected to significantly reduce existing limitations. The advancement of such technologies will mark an important turning point in non-invasive ICP measurement.

AI-ENHANCED NEUROCRITICAL CARE IN THE NEAR FUTURE

In neurocritical care, clinical decisions require careful consideration of the vast amount of information generated by neuromonitoring and patient management. In this endeavor, several decision support AI applications have been proposed; at present, AI applications have been designed and utilized for relatively limited tasks such as patient monitoring, event prediction, and prognosis forecasting, that is, simple classification tasks [130]. Furthermore, most, if not all, existing AI applications are designed to analyze signals, images, or tabular data separately without integration. However, with rapid advancements in AI technology, newer and more advanced approaches have emerged. GMAI and DT technology are two such examples that hold promise for revolutionizing the landscape of neurocritical care. Although somewhat conceptual, these technologies have the potential to significantly enhance the AI capabilities in this field.

Concepts of DT and GMAI

A DT is a virtual replica of a physical system or object and is a technological concept for efficiently testing and simulating products or processes in a digital environment [62,122]. In the medical field, DT can be used as real-time digital replicates of patients to develop effective treatment strategies. Additionally, it could be useful for optimizing medical processes and hospital management strategies [21].

The first application of DT in the medical field was the Archimedes project in 2003, where a DT approach was used and validated to model the complex management of diabetes [27]. Later, Lal et al. [60] developed a DT model of intensive care unit (ICU) patients to predict the therapeutic response during the first 24 hours of sepsis and conducted pilot testing in an actual ICU. In this study, various technologies such as agent-based modeling, discrete event simulation, and Bayesian networks were used to visualize major organs, including the nervous and cardiovascular systems, and simulate treatment effects. The pilot testing of the DT showed a moderate level of agreement (fair level; Cohen’s kappa, 0.41) for the first response and a significantly high level of agreement (good level; Cohen’s kappa, 0.65) for the second response. In recent years, DT models have been developed in other medical fields, such as cardiology and endocrinology, to provide tailored treatment for patients [20,24,30].

The concept of the GMAI stems from the shared limitations of existing medical AI models, which are designed to perform specific tasks and do not generalize to new tasks or datasets [37,44,92]. GMAI aims to address the limitations of traditional medical AI models and relies on a foundation model pretrained on large amounts of data to flexibly perform various tasks [84]. Conceptually, the GMAI can conduct diverse medical tasks without altering the trained network, thereby ensuring robustness and versatility [67]. According to Moor et al. [84], the tremendous potential of the GMAI can support clinical decision-making by drafting radiology reports considering patient history, summarizing real-time patient data, and complex electronic health records to predict future patient status or compare treatment options. In addition, interactive notetaking would be possible by utilizing the development of voice-to-text models to draft reports based on conversations with clinicians; this advantage of GMAI could also be exploited as chatbots for individual patients.

Promises of DT and GMAI in TBI

With DT and GMAI, clinicians may virtually navigate treatment options, crafting personalized medical plans and rehabilitation programs. Furthermore, patient-friendly chatbots and personalized reports can help patients intuitively understand their own condition, enabling them to participate more actively in their healthcare, treatment and recovery processes. The vast spectrum of symptoms and outcomes in TBI patients, driven by the injury’s cause, location, and severity, demands personalized treatment approaches – a long-standing challenge in this field [73,140]. This heterogeneity in TBI has long been recognized as a significant challenge. By leveraging data on a patient’s demographic characteristics, risk factors, and injury patterns, a DT model can be built to aid in the establishment of personalized treatment plans. McIver [79] proposed a model that creates patient-specific finite element meshes based on patients’ MRI data. The model, by using DT technology, reflects the structure and function of an individual’s brain, and can simulate the effects of various types of head trauma. In addition, it enables proactive identification and prediction of patients who may suffer severe long-term damage from repeated head impacts, contributing to prevention and effective management strategies for TBI. Advances in GMAI may potentially allow for the expansion of AI’s application beyond tasks such as prediction and classification [13,36,84,98]. For instance, by integrating medical imaging with clinical information, GMAI can create integrated radiological reports that include not only predictive analysis but also recommendations for personalized treatment and surgical plans for clinicians. Based on this information, chatbot services can be provided to patients, offering easy-to-understand consultations and personalized health education materials.

DT and GMAI can also be used to monitor rehabilitation status and design personalized rehabilitation programs by comprehensively considering a patient’s medical history, cognitive status, and motor skills. For example, Huanxia [46] presented a method to monitor the gait of patients with gait disorders in real time using a powered exoskeleton and DT technology. The powered exoskeleton supports the patient’s movement, while DT analyzes gait patterns by digitally replicating the patient’s actual physical state. This system is capable of providing fall alerts and evaluating the effectiveness of rehabilitation, thereby aiding clinicians in implementing appropriate interventions based on the patient’s condition.

Pitfalls and remaining challenges

The combination of DT and GMAI could offer significant benefits to both clinicians and patients by supporting medical decision-making at every step and directing personalized care effectively. However, despite their tremendous potential, several challenges must be addressed to implement and utilize these technologies in practice.

Data scale

To develop models that can be utilized for various purposes in the medical field, a wide range of diverse and huge datasets are needed, as well as standardized specifications for signals, devices, and other components to ensure data uniformity. Thus, collaboration between different centers, guidelines, and unified storage methods for data standardization is necessary. Additionally, even when data are available, there are concerns about the substantial costs of model training as the size of the foundation model for DT and GMAI increases, as well as the potential environmental concerns associated with it.

Privacy and security

Medical AI handles sensitive personal information related to patient health. Specifically, the DT and GMAI models require a substantial amount of personal information to achieve high performance. DT represents a virtual replica of an individual, and security becomes an even more critical issue. Hence, relevant data must be collected, stored, and processed in a secure and privacy-compliant manner.

Validation

Because of the complexity of the DT and GMAI models, validation poses significant challenges. Unlike the current task-specific AI models, these models perform diverse tasks, making it highly challenging to anticipate all possible scenarios. In addition, given that the DT and GMAI models directly affect patient health when used in the medical field, the validation process is especially important. Thus, diverse and precise validation procedures are required.

Application

Integration of AI into real medical practice, especially in the NICU for TBI patients, presents significant challenges. First, one of the major challenges for AI applications in clinical settings is the issue of compatibility between software and hardware. An algorithm optimized for a specific system may not function properly in another system. For example, an algorithm trained on a particular MRI scanner device may not work well on a different brand or model of scanner. Second, medical AI is still too immature to function effectively especially, in complex medical situations such as TBI. The inherent learning curve of AI demands a significant amount of data and information for actual clinical application; thus, it would take some time for AI to gain a sufficient level of clinical proficiency. While it is crucial to evaluate and validate the performance of algorithms, such as accuracy, it is more important to validate the compatibility, effectiveness, stability, and other related aspects of algorithms when applied in actual clinical settings. Such validation necessitates close collaboration between doctors, computer scientists, and data managers. This will give us a deep understanding of the clinical value of medical AI technology. Finally, until all of these medical AI validation processes are completed, clinicians should avoid over-reliance on AI decisions and always keep in mind the possibility of errors or misdiagnosis.

CONCLUSION

The application of AI, particularly ML, and concepts such as DT and GMAI hold substantial promise for improving neurocritical care for patients with TBI. It has potential across a wide range of applications from prognostication, decision support, patient specific treatment planning, and even rehabilitation. However, although these opportunities are significant, challenges remain, including the need for high-quality data, privacy concerns, ethical considerations, and further validation of these technologies. Despite these hurdles, the future of AI in neurocritical care seems bright, and continuous research and development in this field is necessary to unlock its full potential.

This work was supported by a National Research Foundation of Korea (NRF) Grant funded by the Korean government (Ministry of Science and ICT, MSIT) (No. 2022R1A2C1013205).

Fig. 1. A visual representation of the entire process of the ICP related event prediction model. A : Signal data such as ABP, ICP, and ECG, and EHR data such as personal information, medical history, and radiology reports are collected from patients during their hospitalization. B : Before using the collected data as model input, preprocessing is required. Signal data is filtered according to predefined ranges and segmented into windows of specific sizes. Additionally, considering the quality of signal, noise is removed. Non-signal data from patients undergoes processes such as encoding, normalization, and imputation of missing data, depending on the type of data. C : An appropriate model is selected and trained using the processed ICP-related signal parameters and patient health-related parameters. D : This model, by utilizing signal data from specific window and preprocessed EHR data, can predict life-threatening clinical conditions, such as IH, within a defined time interval. E : In a clinical setting, such predictions can assist clinicians in making decisions for early interventions like drug treatment, surgery, and supportive care. ABP : arterial blood pressure, ICP : intracranial pressure, ECG : electrocardiogram, EHRs : electronic health records, BMI : body mass index, CT : computed tomography, MRI : magnetic resonance imaging, LightGBM : light gradient boosting machine, LSTM : long short term memory, IH : intracranial hypertension, EVD : external ventricular drainage, DC : decompressive craniectomy.

Table 1. Application of AI in phase-specific TBI management

Phases of TBI management	Study	Task	Indicators/strategies	Description	Method	
Initial management	Hanko et al. [42], Yu et al. [139], Cui et al. [22]	Triage	Prehospital variables	Utilize ML algorithms such as XGBoost and LR to help triage patients in the pre-hospital setting by classifying their severity or predicting their risk of premature death	ML	
Ellethy et al. [28], Brossard et al. [14], Jain et al. [50], Keshavamurthy et al. [54]	Diagnosis	CT	RF, SVM, etc. and computer vision are used to rapidly auto-detect lesions in CT scans, and effectively classify patients when integrated with clinical data	ML, DL	
Intensive care unit management	Ye et al. [136]	Multimodality monitoring	ICP	Continuous and early prediction of ICP in using only ICP data with LSTM model	DL	
Schweingruber et al. [108], Jung et al. [52], Lee et al. [63]	Multimodality monitoring	ICP, ABP	For predicting the critical phases of ICP at various levels of accurac y, depending on the time between the prediction moment and the actual risk phase, models such as LSTM and boosting algorithms were utilized	ML, DL	
Miyagawa et al. [83]	Multimodality monitoring	ONSD	Using multivariate analyses and LR, the study found a statistically significant correlation between parameters such as ONSD and ventricular width extracted from CT images, which could be used	SA, ML	
Melinosky et al. [80]	Multimodality monitoring	ECG, PPG	To predict early neurological damage after TBI, this study analyzed ECG and PPG data in the first hour post-resuscitation using a multivariable logistic model, yielding higher predictive accuracy than clinical data alone	ML	
Marshall et al. [76], Ryu et al. [104]	Medical management	BCT	In patients undergoing BCT, improved survival and neurological outcomes were noted, emphasizing uncontrolled RICH as a key mortality predictor. Studies highlight the importance of BCT, especially when alternatives like DC are not feasible	SA	
McIntyre et al. [78], Wang et al. [133]	Medical management	Hypothermia	Studies identified statistically benefits of therapeutic hypothermia in reducing mortality, enhancing brain function recovery, and decreasing stress responses in TBI patients. It also emphasized that these effects vary depending on factors like the depth and duration of hypothermia	SA	
Burgess et al. [15]	Medical management	Hyperosmolar therapy	The study statistically compared the effects of HTS and Mannitol on all-cause mortality, ICP changes, and serious adverse events. It was found that both treatments were similar in terms of average ICP reduction. Notably, HTS demonstrated a lower risk of ICP treatment failure	SA	
Li et al. [65]	Medical management	Hyperosmolar therapy	This study explored using brain EIT to evaluate mannitol treatments for IH, applying clustering methods and SVMs for analysis. The findings showed that EIT could effectively categorize patient responses to mannitol, thus promising a more individualized and effective treatment approach	ML	
Hanko et al. [42]	Medical management	DC	ML models were utilized to predict postoperative mortality and unfavorable outcomes in TBI patients who underwent DC. The best performance was achieved by the RF model for predicting 6-month outcomes	ML	
Neuro prognostication	Tunthanathip and Oearsakul [129], Farzaneh et al. [32], Haveman et al. [43], Raj et al. [96], Pimentel et al. [93]	Outcome prognostication	Clinical scales, vital signs, neuroimaging	Utilizing XGBoost, RF, LR, ANN, and DT, research predicts outcomes like mortality and brain function decline in TBI patients. Predictions are based on in-hospital indicators and target various time frames, including 30 days and 6 months post-injury	ML, DL	
Abujaber et al. [2], Lin et al. [68]	Weaning process	Ventilation	LR, ANN, SVM, RF, DT, and boosting algorithms are employed to predict successful weaning in patients requiring prolonged mechanical ventilation	ML	
AI : artificial intelligence, TBI : traumatic brain injury, ML : machine learning, XGBoost : extreme gradient boosting, LR : logistic regression, ML : machine learning, CT : computed tomography, RF : random forest, SVM : support vector machine, DL : deep learning, ICP : intracranial pressure, LSTM : long short term memory, ABP : arterial blood pressure, ONSD : optic nerve sheath diameter, SA : statistical analyses, ECG : electrocardiogram, PPG : photoplethysmography, BCT : barbiturate coma therapy, RICH : refractory intracranial hypertension, DC : decompressive craniectomy, HTS : hypertonic saline, EIT : electrical impedance tomography, IH : intracranial hypertension, ANN : artificial neural network, DT : decision tree

Table 2. Approaches to reduce false alarms in intensive care units

Scope	Study	Monitoring parameter	Method	Performance	
Enhancing alarm accuracy	Au-Yeung et al. [5]	ECG, PPG, BP	RF	Score*, 83.03	
Wang et al. [132]	ABP, ECG	SVM	Accuracy, 90.39%; FA suppression rate, 83.99%; TA suppression rate, 5.25%	
Li and Clifford [66]	ABP, PPG, ECG	RVM	FA suppression rate	
Asystole, 86.4%; EB, 100%; ET, 27.8%; VT, 30.5%	
Artifact removal	Athaya and Choi [4]	PPG	RF	Accuracy, 84.00%±2.89%	
Tsien et al. [127]	ECG, BP, CO2, O2	DT	Accuracy of eliminating artifacts	
ECG, 98.5%; BP, 98.9%; CO2, 98.3%; O2, 99.8%	
Son et al. [113]	ABP	DBN	Accuracy of eliminating artifacts	
ABP, 95.9%	
Lee et al. [64]	ABP, ICP	STAE-CNN	Accuracy of eliminating artifacts	
ABP, 97%; ICP, 94%	
* Score = (TP + TN) / (TP + TN + FP + 5 × FN) × 100.

ECG : electrocardiogram, PPG : photoplethysmography, BP : blood pressure, RF : random forest, ABP : arterial blood pressure, SVM : support vector machine, FA : false alarm, TA : true alarm, RVM : relevance vector machine, EB : extreme bradycardia, ET : extreme tachycardia, VT : ventricular tachycardia, CO2 : carbon dioxide concentration, O2 : oxygen concentration, DT : decision tree, DBN : deep belief network, ICP : intracranial pressure, STAE-CNN : stacked autoencoder-convolutional neural network, TP : true positive, TN : true negative, FP : false positive, FN : false nagative

Conflicts of interest

No potential conflict of interest relevant to this article was reported.

Informed consent

This type of study does not require informed consent.

Author contributions

Conceptualization : KAK, HK, DJK; Funding acquisition : DJK; Methodology : EJH, BCY; Project administration : HK, DJK; Visualization : KAK; Writing - original draft : KAK; Writing - review & editing : KAK, HK, EJH, BCY, DJK

Data sharing

None

Preprint

None
==== Refs
References

1 Abe D Inaji M Hase T Takahashi S Sakai R Ayabe F A prehospital triage system to detect traumatic intracranial hemorrhage using machine learning algorithms JAMA Netw Open 5 e2216393 2022 35687335
2 Abujaber A Fadlalla A Gammoh D Abdelrahman H Mollazehi M ElMenyar A Using trauma registry data to predict prolonged mechanical ventilation in patients with traumatic brain injury: machine learning approach PLoS one 15 e0235231 2020 32639971
3 Al-Mufti F Smith B Lander M Damodara N Nuoman R El-Ghanem M Novel minimally invasive multi-modality monitoring modalities in neurocritical care J Neurol Sci 390 184 192 2018 29801883
4 Athaya T, Choi S : Evaluation of different machine learning models for photoplethysmogram signal artifact detection. 2020 International conference on information and communication technology convergence (ICTC); 2020 Oct 21-23; Jeju, Korea. New York : IEEE, c2020, pp1206-1208
5 Au-Yeung WM Sahani AK Isselbacher EM Armoundas AA Reduction of false alarms in the intensive care unit using an optimized machine learning based approach NPJ Digit Med 2 86 2019 31508497
6 Awaysheh A Wilcke J Elvinger F Rees L Fan W Zimmerman KL Review of medical decision support and machine-learning methods Vet Pathol 56 512 525 2019 30866728
7 Azabou E Navarro V Kubis N Gavaret M Heming N Cariou A Value and mechanisms of EEG reactivity in the prognosis of patients with impaired consciousness: a systematic review Crit Care 22 184 2018 30071861
8 Backhaus S : Traumatic Brain Injury (TBI) in Kreutzer JS, DeLuca J, Caplan B (eds) : Encyclopedia of Clinical Neuropsychology. New York : Springer New York, 2011, pp2550-2554
9 Bakator M Radosav D Deep learning and medical diagnosis: a review of literature Multimodal Technol Interact 2 47 2018
10 Beqiri E Smielewski P Robba C Czosnyka M Cabeleira MT Tas J Feasibility of individualised severe traumatic brain injury management using an automated assessment of optimal cerebral perfusion pressure: the COGiTATE phase II study protocol BMJ Open 9 e030727 2019
11 Bhavsar KA Singla J Al-Otaibi YD Song OY Zikria YB Bashir AK Medical diagnosis using machine learning: a statistical review Comput Mater Contin 67 107 125 2021
12 Bonds BW Yang S Hu PF Kalpakis K Stansbury LG Scalea TM Predicting secondary insults after severe traumatic brain injury J Trauma Acute Care Surg 79 85 90 2015 26091319
13 Briganti G : A clinician’s guide to large language models. Future Medicine AI 1 : FMAI1, 2023
14 Brossard C Lemasson B Attyé A De Busschère JA Payen JF Barbier EL Contribution of CT-scan analysis by artificial intelligence to the clinical care of TBI patients Front Neurol 12 666875 2021 34177773
15 Burgess S Abu-Laban RB Slavik RS Vu EN Zed PJ A systematic review of randomized controlled trials comparing hypertonic sodium solutions and mannitol for traumatic brain injury: implications for emergency department management Ann Pharmacother 50 291 300 2016 26825644
16 Carra G Güiza F Depreitere B Meyfroidt G; CENTER-TBI High-Resolution ICU (HR ICU) Sub-Study Participants Prediction model for intracranial hypertension demonstrates robust performance during external validation on the CENTER-TBI dataset Intensive Care Med 47 124 126 2021 33001234
17 Chen JW Gombart ZJ Rogers S Gardiner SK Cecil S Bullock RM Pupillary reactivity as an early indicator of increased intracranial pressure: the introduction of the Neurological Pupil index Surg Neurol Int 2 82 2011 21748035
18 Chesnut RM Temkin N Carney N Dikmen S Rondina C Videtta W A trial of intracranial-pressure monitoring in traumatic brain injury N Engl J Med 367 2471 2481 2012 23234472
19 Choi Y Park JH Hong KJ Ro YS Song KJ Shin SD Development and validation of a prehospital-stage prediction tool for traumatic brain injury: a multicentre retrospective cohort study in Korea BMJ Open 12 e055918 2022
20 Corral-Acero J Margara F Marciniak M Rodero C Loncaric F Feng Y The ‘Digital Twin’ to enable the vision of precision cardiology Eur Heart J 41 4556 4564 2020 32128588
21 Croatti A Gabellini M Montagna S Ricci A On the integration of agents and digital twins in healthcare J Med Syst 44 161 2020 32748066
22 Cui W Ge S Shi Y Wu X Luo J Lui H Death after discharge: prognostic model of 1-year mortality in traumatic brain injury patients undergoing decompressive craniectomy Chin Neurosurg J 7 24 2021 33879254
23 Cvach M Monitor alarm fatigue: an integrative review Biomed Instrum Technol 46 268 277 2012 22839984
24 DeJournett L DeJournett J In silico testing of an artificial-intelligencebased artificial pancreas designed for use in the intensive care unit setting J Diabetes Sci Technol 10 1360 1371 2016 27301982
25 Drew BJ Harris P Zègre-Hemsey JK Mammone T Schindler D Salas-Boni R Insights into the problem of alarm fatigue with physiologic monitor devices: a comprehensive observational study of consecutive intensive care unit patients PloS One 9 e110274 2014 25338067
26 Dundar TT Yurtsever I Pehlivanoglu MK Yildiz U Eker A Demir MA Machine learning-based surgical planning for neurosurgery: artificial intelligent approaches to the cranium Front Surg 9 863633 2022 35574559
27 Eddy DM Schlessinger L Validation of the Archimedes diabetes model Diabetes Care 26 3102 3110 2003 14578246
28 Ellethy H Chandra SS Nasrallah FA The detection of mild traumatic brain injury in paediatrics using artificial neural networks Comput Biol Med 135 104614 2021 34229143
29 Emami H Dong M Nejad-Davarani SP Glide-Hurst CK Generating synthetic CTs from magnetic resonance images using generative adversarial networks Med Phys 45 3627 3636 2018
30 Erol T, Mendi AF, Doğan D : The digital twin revolution in healthcare. 2020 4th international symposium on multidisciplinary studies and innovative technologies (ISMSIT); 2020 Oct 22-24; Istanbul, Turkey. New York : IEEE, c2020, pp1-7
31 Evensen KB Eide PK Measuring intracranial pressure by invasive, less invasive or non-invasive means: limitations and avenues for improvement Fluids Barriers CNS 17 34 2020 32375853
32 Farzaneh N Williamson CA Gryak J Najarian K A hierarchical expertguided machine learning framework for clinical decision support systems: an application to traumatic brain injury prognostication NPJ Digit Med 4 78 2021 33963275
33 Garg A Mago V Role of machine learning in medical research: a survey Compu Sci Rev 40 100370 2021
34 Ghajar J Traumatic brain injury Lancet 356 923 929 2000 11036909
35 Glaser J Vasquez M Cardarelli C Galvagno S Jr Stein D Murthi S Through the looking glass: early non-invasive imaging in TBI predicts the need for interventions Trauma Surg Acute Care Open 1 e000019 2016 29766062
36 Gong EJ Bang CS Interpretation of medical images using artificial intelligence: current status and future perspectives Korean J Gastroenterol 82 43 45 2023
37 Greenfield D : Artificial intelligence in medicine: applications, implications and limitations. Available at : https://sitn.hms.harvard.edu/flash/2019/artificial-intelligence-in-medicine-applications-implications-and-limitations/
38 Güiza F Depreitere B Piper I Van den Berghe G Meyfroidt G Novel methods to predict increased intracranial pressure during intensive care and long-term neurologic outcome after traumatic brain injury: development and validation in a multicenter dataset Crit Care Med 41 554 564 2013 23263587
39 Güler İ Gökçil Z Gülbandilar E Evaluating of traumatic brain injuries using artificial neural networks Expert Syst Appl 36 10424 10427 2009
40 Habibzadeh A Khademolhosseini S Kouhpayeh A Niakan A Asadi MA Ghasemi H Machine learning-based models to predict the need for neurosurgical intervention after moderate traumatic brain injury Health Sci Rep 6 e1666 2023 37908638
41 Hale AT Stonko DP Lim J Guillamondegui OD Shannon CN Patel MB Using an artificial neural network to predict traumatic brain injury J Neurosurg Pediatr 23 219 226 2018 30485240
42 Hanko M Grendár M Snopko P Opšenák R Šutovský J Benčo M Random forest-based prediction of outcome and mortality in patients with traumatic brain injury undergoing primary decompressive craniectomy World Neurosurg 148 e450 e458 2021 33444843
43 Haveman ME Van Putten MJAM Hom HW Eertman-Meyer CJ Beishuizen A Tjepkema-Cloostermans MC Predicting outcome in patients with moderate to severe traumatic brain injury using electroencephalography Crit Care 23 401 2019 31829226
44 Hosny A Parmar C Quackenbush J Schwartz LH Aerts HJWL Artificial intelligence in radiology Nat Rev Cancer 18 500 510 2018 29777175
45 Hsu YC Weng HH Kuo CY Chu TP Tsai YH Prediction of fall events during admission using eXtreme gradient boosting: a comparative validation study Sci Rep 10 16777 2020 33033326
46 Huanxia W : A method for patient gait real-time monitoring based on powered exoskeleton and digital twin. Fifth International Conference on Computer Information Science and Artificial Intelligence (CISAI 2022); 2022 Sep 16-18; Chongqing, China. Bellingham : SPIE, c2023, Vol 12566, pp734-743
47 Hunter OF Perry F Salehi M Bandurski H Hubbard A Ball CG Science fiction or clinical reality: a review of the applications of artificial intelligence along the continuum of trauma care World J Emerg Surg 18 16 2023 36879293
48 Imaduddin SM Fanelli A Vonberg FW Tasker RC Heldt T PseudoBayesian model-based noninvasive intracranial pressure estimation and tracking IEEE Trans Biomed Eng 67 1604 1615 2020 31535978
49 Jahns FP Miroz JP Messerer M Daniel RT Taccone FS Eckert P Quantitative pupillometry for the monitoring of intracranial hypertension in patients with severe traumatic brain injury Crit Care 23 155 2019 31046817
50 Jain S Vyvere TV Terzopoulos V Sima DM Roura E Maas A Automatic quantification of computed tomography features in acute traumatic brain injury J Neurotrauma 36 1794 1803 2019 30648469
51 Jaishankar R Fanelli A Filippidis A Vu T Holsapple J Heldt T A spectral approach to model-based noninvasive intracranial pressure estimation IEEE J Biomed Health Inform 24 2398 2406 2020 31880569
52 Jung MK Ahn D Park CM Ha EJ Roh TH You NK Prediction of serious intracranial hypertension from low-resolution neuromonitoring in traumatic brain injury: an explainable machine learning approach IEEE J Biomed Health Inform 27 1903 1913 2023
53 Kashif FM Verghese GC Novak V Czosnyka M Heldt T Model-based noninvasive estimation of intracranial pressure from cerebral blood flow velocity and arterial pressure Sci Transl Med 4 129ra144 2012
54 Keshavamurthy KN, Leary OP, Merck LH, Kimia B, Collins S, Wright DW, et al. : Machine learning algorithm for automatic detection of CT-identifiable hyperdense lesions associated with traumatic brain injury. Medical Imaging 2017: Computer-Aided Diagnosis; 2017 Feb 11-16; Olando, FL. Bellingham : SPIE, c2017, Vol 10134, pp630-638
55 Kim H Lee SB Son Y Czosnyka M Kim DJ Hemodynamic instability and cardiovascular events after traumatic brain injury predict outcome after artifact removal with deep belief network analysis J Neurosurg Anesthesiol 30 347 353 2018 28991060
56 Kim YJ The impact of time to surgery on outcomes in patients with traumatic brain injury: a literature review Int Emerg Nurs 22 214 219 2014 24680689
57 Kinoshita K Traumatic brain injury: pathophysiology for neurocritical care J Intensive Care 4 29 2016 27123305
58 Kovacs M Peluso L Njimi H De Witte O Gouvêa Bogossian E Quispe Cornejo A Optimal cerebral perfusion pressure guided by brain oxygen pressure measurement Front Neurol 12 732830 2021 34777201
59 Kristiansson H Nissborg E Bartek J Jr Andresen M Reinstrup P Romner B Measuring elevated intracranial pressure through noninvasive methods: a review of the literature J Neurosurg Anesthesiol 25 372 385 2013 23715045
60 Lal A Li G Cubro E Chalmers S Li H Herasevich V Development and verification of a digital twin patient model to predict specific treatment response during the first 24 hours of sepsis Crit Care Explor 2 e0249 2020 33225302
61 Lameski P Zdravevski E Koceski S Kulakov A Trajkovik V Suppression of intensive care unit false alarms based on the arterial blood pressure signal IEEE Access 5 5829 5836 2017
62 Laubenbacher R Sluka JP Glazier JA Using digital twins in viral infection Science 371 1105 1106 2021 33707255
63 Lee HJ Kim H Kim YT Won K Czosnyka M Kim DJ Prediction of lifethreatening intracranial hypertension during the acute phase of traumatic brain injury using machine learning IEEE J Biomed Health Inform 25 3967 3976 2021 34061756
64 Lee SB Kim H Kim YT Zeiler FA Smielewski P Czosnyka M Artifact removal from neurophysiological signals: impact on intracranial and arterial pressure monitoring in traumatic brain injury J Neurosurg 132 1952 1960 2019 31075774
65 Li H Ma H Yang B Xu C Cao L Dong X Automatic evaluation of mannitol dehydration treatments on controlling intracranial pressure using electrical impedance tomography IEEE Sens J 20 4832 4839 2020
66 Li Q Clifford GD Signal quality and data fusion for false alarm reduction in the intensive care unit J Electrocardiol 45 596 603 2012 22960167
67 Lin B Chen Z Li M Lin H Xu H Zhu Y Towards medical artificial general intelligence via knowledge-enhanced multimodal pretraining Available at : 10.48550/arXiv.2304.14204
68 Lin MY Li CC Lin PH Wang JL Chan MC Wu CL Explainable machine learning to predict successful weaning among patients requiring prolonged mechanical ventilation: a retrospective cohort study in Central Taiwan Front Med (Lausanne) 8 663739 2021 33968967
69 Lyashevska O Malone F MacCarthy E Fiehler J Buhk JH Morris L Class imbalance in gradient boosting classification algorithms: application to experimental stroke data Stat Methods Med Res 30 916 925 2021 33356965
70 Maas MB Naidech AM Batra A Chou SH Bleck TP Comment on “Can quantitative pupillometry be used to screen for elevated intracranial pressure? A retrospective cohort study” Neurocrit Care 37 597 598 2022 35761124
71 Magoulas GD Prentza A Machine learning in medical applications Paliouras G Karkaletsis V Spyropoulos CD Machine Learning and Its Applications Berlin Springer 2021 300 307
72 Majdan M Brazinova A Rusnak M Leitgeb J Outcome prediction after traumatic brain injury: comparison of the performance of routinely used severity scores and multivariable prognostic models J Neurosci Rural Pract 8 20 29 2017 28149077
73 Majdan M Mauritz W Brazinova A Rusnak M Leitgeb J Janciak I Severity and outcome of traumatic brain injuries (TBI) with different causes of injury Brain Inj 25 797 805 2011 21631184
74 Makarenko S Griesdale DE Gooderham P Sekhon MS Multimodal neuromonitoring for traumatic brain injury: a shift towards individualized therapy J Clin Neurosci 26 8 13 2016 26755455
75 Mariak Z Swiercz M Krejza J Lewko J Lyson T Intracranial pressure processing with artificial neural networks: classification of signal properties Acta Neurochir (Wien) 142 407 411 discussion 411-412 2000 10883337
76 Marshall GT James RF Landman MP O’Neill PJ Cotton BA Hansen EN Pentobarbital coma for refractory intra-cranial hypertension after severe traumatic brain injury: mortality predictions and one-year outcomes in 55 patients J Trauma 69 275 283 2010 20699736
77 Matsushima K Inaba K Siboni S Skiada D Strumwasser AM Magee GA Emergent operation for isolated severe traumatic brain injury: does time matter? J Trauma Acute Care Surg 79 838 842 2015 26317818
78 McIntyre LA Fergusson DA Hébert PC Moher D Hutchison JS Prolonged therapeutic hypothermia after traumatic brain injury in adults: a systematic review JAMA 289 2992 2999 2003 12799408
79 McIver KG The Application of High-Performance Computing to Create and Analyze Simulations of Human Injury West Lafayette Purdue University Graduate School 2022
80 Melinosky C Yang S Hu P Li H Miller CHT Khan I Continuous vital sign analysis to predict secondary neurological decline after traumatic brain injury Front Neurol 9 761 2018 30319521
81 Meyfroidt G Bouzat P Casaer MP Chesnut R Hamada SR Helbok R Management of moderate to severe traumatic brain injury: an update for the intensivist Intensive Care Med 48 649 666 2022 35595999
82 Mikola A, Rätsep I, Särkelä M, Lipping T : Prediction of outcome in traumatic brain injury patients using long-term qEEG features. 2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC); 2015 Aug 25-29; Milan, Italy. New York : IEEE, c2015, pp1532-1535
83 Miyagawa T Sasaki M Yamaura A Intracranial pressure based decision making: prediction of suspected increased intracranial pressure with machine learning PLoS One 15 e0240845 2020 33085690
84 Moor M Banerjee O Abad ZSH Krumholz HM Leskovec J Topol EJ Foundation models for generalist medical artificial intelligence Nature 616 259 265 2023 37045921
85 Moyer JD Lee P Bernard C Henry L Lang E Cook F Machine learning-based prediction of emergency neurosurgery within 24 h after moderate to severe traumatic brain injury World J Emerg Surg 17 42 2022 35922831
86 Naharro-Abellán A Lobo-Val-buena B Gordo F Clinical decision support systems: future or present in ICU ICU Manag Pract 19 202 205 2019
87 Noh SH Cho PG Kim KN Kim SH Shin DA Artificial intelligence for neurosurgery : current state and future directions J Korean Neurosurg Soc 66 113 120 2023 36124365
88 Noor NSEM Ibrahim H Lah MHC Abdullah JM Improving outcome prediction for traumatic brain injury from imbalanced datasets using RUSBoosted trees on electroencephalography spectral power IEEE Access 9 121608 121631 2021
89 Noraky J Verghese GC Searls DE Lioutas VA Sonni S Thomas A Noninvasive intracranial pressure determination in patients with subarachnoid hemorrhage Acta Neurochir Suppl 122 65 68 2016 27165879
90 Osheroff JA Teich J Levick D Saldana L Velasco F Sittig D Improving outcomes with clinical decision support: an implementer’s guide Chicago Himss Publishing 2012
91 Pansell J Hack R Rudberg P Bell M Cooray C Can quantitative pupillometry be used to screen for elevated intracranial pressure? A retrospective cohort study Neurocrit Care 37 531 537 2022 35606562
92 Pesapane F Codari M Sardanelli F Artificial intelligence in medical imaging: threat or opportunity? Radiologists again at the forefront of innovation in medicine Eur Radiol Exp 2 35 2018 30353365
93 Pimentel MA Brennan T Lehman LW King NK Ang BT Feng M Outcome prediction for patients with traumatic brain injury with dynamic features from intracranial pressure and arterial blood pressure signals: a Gaussian process approach Acta Neurochir Suppl 122 85 91 2016 27165883
94 Popovic D Khoo M Lee S Noninvasive monitoring of intracranial pressure Recent Pat Biomed Eng 2 165 179 2009
95 Powers WJ Intracerebral hemorrhage and head trauma: common effects and common mechanisms of injury Stroke 41 10 Suppl S107 S110 2010 20876480
96 Raj R Luostarinen T Pursiainen E Posti JP Takala RSK Bendel S Machine learning-based dynamic mortality prediction after traumatic brain injury Sci Rep 9 17672 2019 31776366
97 Rajaei F Cheng S Williamson CA Wittrup E Najarian K AI-based decision support system for traumatic brain injury: a survey Diagnostics (Basel) 13 1640 2023 37175031
98 Rajpurkar P Lungren MP The current and future state of AI interpretation of medical images N Engl J Med 388 1981 1990 2023 37224199
99 Robba C Asgari S Gupta A Badenes R Sekhon M Bequiri E Lung injury is a predictor of cerebral hypoxia and mortality in traumatic brain injury Front Neurol 11 771 2020 32849225
100 Robba C Bacigaluppi S Cardim D Donnelly J Bertuccio A Czosnyka M Non-invasive assessment of intracranial pressure Acta Neurol Scand 134 4 21 2016
101 Robba C Pozzebon S Moro B Vincent JL Creteur J Taccone FS Multimodal non-invasive assessment of intracranial hypertension: an observational study Crit Care 24 379 2020 32591024
102 Rohaut B Eliseyev A Claassen J Uncovering consciousness in unresponsive ICU patients: technical, medical and ethical considerations Crit Care 23 78 2019 30850022
103 Rosenberg JB Shiloh AL Savel RH Eisen LA Non-invasive methods of estimating intracranial pressure Neurocrit Care 15 599 608 2011 21519957
104 Ryu JA Jung W Jung YJ Kwon DY Kang K Choi H Early prediction of neurological outcome after barbiturate coma therapy in patients undergoing brain tumor surgery PLoS One 14 e0215280 2019 30995269
105 Sadrawi M Lin YT Lin CH Mathunjwa B Hsin HT Fan SZ Non-invasive hemodynamics monitoring system based on electrocardiography via deep convolutional autoencoder Sensors (Basel) 21 6264 2021 34577471
106 Sainbhi AS Gomez A Froese L Slack T Batson C Stein KY Noninvasive and minimally-invasive cerebral autoregulation assessment: a narrative review of techniques and implications for clinical research Front Neurol 13 872731 2022 35557627
107 Scalzo F Hamilton R Asgari S Kim S Hu X Intracranial hypertension prediction using extremely randomized decision trees Med Eng Phys 34 1058 1065 2012 22401795
108 Schweingruber N Mader MMD Wiehe A Röder F Göttsche J Kluge S A recurrent machine learning model predicts intracranial hypertension in neurointensive care patients Brain 145 2910 2919 2022 35139181
109 Seelig JM Becker DP Miller JD Greenberg RP Ward JD Choi SC Traumatic acute subdural hematoma: major mortality reduction in comatose patients treated within four hours N Engl J Med 304 1511 1518 1981 7231489
110 Shah RV Grennan G Zafar-Khan M Alim F Dey S Ramanathan D Personalized machine learning of depressed mood using wearables Transl Psychiatry 11 338 2021 34103481
111 Sidey-Gibbons JAM Sidey-Gibbons CJ Machine learning in medicine: a practical introduction BMC Med Res Methodol 19 64 2019 30890124
112 Smith M Multimodality neuromonitoring in adult traumatic brain injury: a narrative review Anesthesiology 128 401 415 2018 28938277
113 Son Y Lee SB Kim H Song ES Huh H Czosnyka M Automated artifact elimination of physiological signals using a deep belief network: an application for continuously measured arterial blood pressure waveforms Inf Sci 456 145 158 2018
114 Staartjes VE Stumpo V Kernbach JM Klukowska AM Gadjradj PS Schröder ML Machine learning in neurosurgery: a global survey Acta Neurochir (Wien) 162 3081 3091 2020 32812067
115 Stangler LA Kouzani A Bennet KE Dumee L Berk M Worrell GA Microdialysis and microperfusion electrodes in neurologic disease monitoring Fluids Barriers CNS 18 52 2021 34852829
116 Stein SC Georgoff P Meghan S Mirza KL El Falaky OM Relationship of aggressive monitoring and treatment to improved outcomes in severe traumatic brain injury J Neurosurg 112 1105 1112 2010 19747054
117 Stevens AR Su Z Toman E Belli A Davies D Optical pupillometry in traumatic brain injury: neurological pupil index and its relationship with intracranial pressure through significant event analysis Brain Inj 33 1032 1038 2019 31021683
118 Steyerberg EW Mushkudiani N Perel P Butcher I Lu J McHugh GS Predicting outcome after traumatic brain injury: development and international validation of prognostic scores based on admission characteristics PLoS Med 5 e165 discussion e165 2008 18684008
119 Stocker RA Intensive care in traumatic brain injury including multimodal monitoring and neuroprotection Med Sci (Basel) 7 37 2019 30813644
120 Surendrakumar S Rabelo TK Campos ACP Mollica A Abrahao A Lipsman N Neuromodulation therapies in pre-clinical models of traumatic brain injury: systematic review and translational applications J Neurotrauma 40 435 448 2023 35983592
121 Svedung Wettervik TM Lewén A Enblad P Fine tuning of traumatic brain injury management in neurointensive care-indicative observations and future perspectives Front Neurol 12 638132 2021 33716941
122 Tao F Qi Q Make more digital twins Nature 573 490 491 2019 31554984
123 Thabtah F Abdelhamid N Peebles D A machine learning autism classification based on logistic regression analysis Health Inf Sci Syst 7 12 2019 31168365
124 Tierney KJ Nayak NV Prestigiacomo CJ Sifri ZC Neurosurgical intervention in patients with mild traumatic brain injury and its effect on neurological outcomes J Neurosurg 124 538 545 2016 26406795
125 Tisdall MM Smith M Multimodal monitoring in traumatic brain injury: current status and future directions Br J Anaesth 99 61 67 2007 17548431
126 Tsien CL Fackler JC Poor prognosis for existing monitors in the intensive care unit Crit Care Med 25 614 619 1997 9142025
127 Tsien CL Kohane IS McIntosh N Multiple signal integration by decision tree induction to detect artifacts in the neonatal intensive care unit Artif Intell Med 19 189 202 2000 10906612
128 Tu KC Eric Nyam TT Wang CC Chen NC Chen KT Chen CJ A computer-assisted system for early mortality risk prediction in patients with traumatic brain injury using artificial intelligence algorithms in emergency room triage Brain Sci 12 612 2022 35624999
129 Tunthanathip T Oearsakul T Application of machine learning to predict the outcome of pediatric traumatic brain injury Chin J Traumatol 24 350 355 2021 34284922
130 van de Sande D van Genderen ME Huiskens J Gommers D van Bommel J Moving from bytes to bedside: a systematic review on the use of artificial intelligence in the intensive care unit Intensive Care Med 47 750 760 2021 34089064
131 Vilela GH Cabella B Mascarenhas S Czosnyka M Smielewski P Dias C Validation of a new minimally invasive intracranial pressure monitoring method by direct comparison with an invasive technique Acta Neurochir Suppl 122 97 100 2016 27165885
132 Wang X, Gao Y, Lin J, Rangwala H, Mittu R : A machine learning approach to false alarm detection for critical arrhythmia alarms. 2015 IEEE 14th international conference on machine learning and applications (ICMLA); 2015 Dec 9-11; Miami, FL. New York : IEEE, 2015, pp202-207
133 Wang Y Huang C Tian R Yang X Target temperature management and therapeutic hypothermia in sever neuroprotection for traumatic brain injury: clinic value and effect on oxidative stress Medicine (Baltimore) 102 e32921 2023 36897685
134 Wang Z Wang H Becker R Rufo J Yang S Mace BE Acoustofluidic separation enables early diagnosis of traumatic brain injury based on circulating exosomes Microsyst Nanoeng 7 20 2021 34567734
135 Whalen S Schreiber J Noble WS Pollard KS Navigating the pitfalls of applying machine learning in genomics Nat Rev Genet 23 169 181 2022 34837041
136 Ye G Balasubramanian V Li JK Kaya M Machine learning-based continuous intracranial pressure prediction for traumatic injury patients IEEE J Transl Eng Health Med 10 4901008 2022 35795876
137 Yokobori S Hosein K Burks S Sharma I Gajavelli S Bullock R Biomarkers for the clinical differential diagnosis in traumatic brain injury--a systematic review CNS Neurosci Ther 19 556 565 2013 23710877
138 Young AMH Guilfoyle MR Donnelly J Smielewski P Agarwal S Czosnyka M Multimodality neuromonitoring in severe pediatric traumatic brain injury Pediatr Res 83 41 49 2018 29084196
139 Yu R Wang S Xu J Wang Q He X Li J Machine learning approaches-driven for mortality prediction for patients undergoing craniotomy in ICU Brain Inj 35 1658 1664 2021 35080996
140 Zeiler FA Iturria-Medina Y Thelin EP Gomez A Shankar JJ Ko JH Integrative neuroinformatics for precision prognostication and personalized therapeutics in moderate and severe traumatic brain injury Front Neurol 12 729184 2021 34557154
141 Zhang X Medow JE Iskandar BJ Wang F Shokoueinejad M Koueik J Invasive and noninvasive means of measuring intracranial pressure: a review Physiol Meas 38 R143 R182 2017 28489610
142 Zhang X Yan C Gao C Malin BA Chen Y Predicting missing values in medical data via XGBoost regression J Healthc Inform Res 4 383 394 2020 33283143
