Machine Learning and Neurosurgical Outcome Prediction: A Systematic Review

Usefulness of Artificial Intelligence in Traumatic Brain Injury: A Bibliometric Analysis and Mini-review

BACKGROUND

Traumatic brain injury (TBI) has become a major source of disability worldwide, increasing the interest in algorithms that use artificial intelligence (AI) to optimize the interpretation of imaging studies, prognosis estimation, and critical care issues. In this study we present a bibliometric analysis and mini-review on the main uses that have been developed for TBI in AI.

METHODS

The results informing this review come from a Scopus database search as of April 15, 2023. The bibliometric analysis was carried out via the mapping bibliographic metrics method. Knowledge mapping was made in the VOSviewer software (V1.6.18), analyzing the "link strength" of networks based on co-occurrence of key words, countries co-authorship, and co-cited authors. In the mini-review section, we highlight the main findings and contributions of the studies.

RESULTS

A total of 495 scientific publications were identified from 2000 to 2023, with 9262 citations published since 2013. Among the 160 journals identified, The Journal of Neurotrauma, Frontiers in Neurology, and PLOS ONE were those with the greatest number of publications. The most frequently co-occurring key words were: "machine learning", "deep learning", "magnetic resonance imaging", and "intracranial pressure". The United States accounted for more collaborations than any other country, followed by United Kingdom and China. Four co-citation author clusters were found, and the top 20 papers were divided into reviews and original articles.

CONCLUSIONS

AI has become a relevant research field in TBI during the last 20 years, demonstrating great potential in imaging, but a more modest performance for prognostic estimation and neuromonitoring.

Developing machine learning models to predict multi-class functional outcomes and death three months after stroke in Sweden

Globally, stroke is the third-leading cause of mortality and disability combined, and one of the costliest diseases in society. More accurate predictions of stroke outcomes can guide healthcare organizations in allocating appropriate resources to improve care and reduce both the economic and social burden of the disease. We aim to develop and evaluate the performance and explainability of three supervised machine learning models and the traditional multinomial logistic regression (mLR) in predicting functional dependence and death three months after stroke, using routinely-collected data. This prognostic study included adult patients, registered in the Swedish Stroke Registry (Riksstroke) from 2015 to 2020. Riksstroke contains information on stroke care and outcomes among patients treated in hospitals in Sweden. Prognostic factors (features) included demographic characteristics, pre-stroke functional status, cardiovascular risk factors, medications, acute care, stroke type, and severity. The outcome was measured using the modified Rankin Scale at three months after stroke (a scale of 0-2 indicates independent, 3-5 dependent, and 6 dead). Outcome prediction models included support vector machines, artificial neural networks (ANN), eXtreme Gradient Boosting (XGBoost), and mLR. The models were trained and evaluated on 75% and 25% of the dataset, respectively. Model predictions were explained using SHAP values. The study included 102,135 patients (85.8% ischemic stroke, 53.3% male, mean age 75.8 years, and median NIHSS of 3). All models demonstrated similar overall accuracy (69%-70%). The ANN and XGBoost models performed significantly better than the mLR in classifying dependence with F1-scores of 0.603 (95% CI; 0.594-0.611) and 0.577 (95% CI; 0.568-0.586), versus 0.544 (95% CI; 0.545-0.563) for the mLR model. The factors that contributed most to the predictions were expectedly similar in the models, based on clinical knowledge. Our ANN and XGBoost models showed a modest improvement in prediction performance and explainability compared to mLR using routinely-collected data. Their improved ability to predict functional dependence may be of particular importance for the planning and organization of acute stroke care and rehabilitation.

Predicting Outcomes Following Lower Extremity Endovascular Revascularization Using Machine Learning

BACKGROUND

Lower extremity endovascular revascularization for peripheral artery disease carries nonnegligible perioperative risks; however, outcome prediction tools remain limited. Using machine learning, we developed automated algorithms that predict 30-day outcomes following lower extremity endovascular revascularization.

METHODS AND RESULTS

The National Surgical Quality Improvement Program targeted vascular database was used to identify patients who underwent lower extremity endovascular revascularization (angioplasty, stent, or atherectomy) for peripheral artery disease between 2011 and 2021. Input features included 38 preoperative demographic/clinical variables. The primary outcome was 30-day postprocedural major adverse limb event (composite of major reintervention, untreated loss of patency, or major amputation) or death. Data were split into training (70%) and test (30%) sets. Using 10-fold cross-validation, 6 machine learning models were trained using preoperative features. The primary model evaluation metric was area under the receiver operating characteristic curve. Overall, 21 886 patients were included, and 30-day major adverse limb event/death occurred in 1964 (9.0%) individuals. The best performing model for predicting 30-day major adverse limb event/death was extreme gradient boosting, achieving an area under the receiver operating characteristic curve of 0.93 (95% CI, 0.92-0.94). In comparison, logistic regression had an area under the receiver operating characteristic curve of 0.72 (95% CI, 0.70-0.74). The calibration plot showed good agreement between predicted and observed event probabilities with a Brier score of 0.09. The top 3 predictive features in our algorithm were (1) chronic limb-threatening ischemia, (2) tibial intervention, and (3) congestive heart failure.

CONCLUSIONS

Our machine learning models accurately predict 30-day outcomes following lower extremity endovascular revascularization using preoperative data with good discrimination and calibration. Prospective validation is warranted to assess for generalizability and external validity.

Machine learning application in otology

This review presents a comprehensive history of Artificial Intelligence (AI) in the context of the revolutionary application of machine learning (ML) to medical research and clinical utilization, particularly for the benefit of researchers interested in the application of ML in otology. To this end, we discuss the key components of ML-input, output, and algorithms. In particular, some representation algorithms commonly used in medical research are discussed. Subsequently, we review ML applications in otology research, including diagnosis, influential identification, and surgical outcome prediction. In the context of surgical outcome prediction, specific surgical treatments, including cochlear implantation, active middle ear implantation, tympanoplasty, and vestibular schwannoma resection, are considered. Finally, we highlight the obstacles and challenges that need to be overcome in future research.

Predicting the Severity and Discharge Prognosis of Traumatic Brain Injury Based on Intracranial Pressure Data Using Machine Learning Algorithms

OBJECTIVE

This study aimed to explore the potential of employing machine learning algorithms based on intracranial pressure (ICP), ICP-derived parameters, and their complexity to predict the severity and short-term prognosis of traumatic brain injury (TBI).

METHODS

A single-center prospectively collected cohort of neurosurgical intensive care unit admissions was analyzed. We extracted ICP-related data within the first 6 hours and processed them using complex algorithms. To indicate TBI severity and short-term prognosis, Glasgow Coma Scale score on the first postoperative day and Glasgow Outcome Scale-Extended score at discharge were used as binary outcome variables. A univariate logistic regression model was developed to predict TBI severity using only mean ICP values. Subsequently, 3 multivariate Random Forest (RF) models were constructed using different combinations of mean and complexity metrics of ICP-related data. To avoid overfitting, five-fold cross-validations were performed. Finally, the best-performing multivariate RF model was used to predict patients' discharge Glasgow Outcome Scale-Extended score.

RESULTS

The logistic regression model exhibited limited predictive ability with an area under the curve (AUC) of 0.558. Among multivariate models, the RF model, combining the mean and complexity metrics of ICP-related data, achieved the most robust ability with an AUC of 0.815. Finally, in terms of predicting discharge Glasgow Outcome Scale-Extended score, this model had a consistent performance with an AUC of 0.822. Cross-validation analysis confirmed the performance.

CONCLUSIONS

This study demonstrates the clinical utility of the RF model, which integrates the mean and complexity metrics of ICP data, in accurately predicting the TBI severity and short-term prognosis.

Automatic Segmentation of Vestibular Schwannomas: A Systematic Review

BACKGROUND

Vestibular schwannomas (VSs) are benign tumors often monitored over time, with measurement techniques for assessing growth rates subject to significant interobserver variability. Automatic segmentation of these tumors could provide a more reliable and efficient for tracking their progression, especially given the irregular shape and growth patterns of VS.

METHODS

Various studies and segmentation techniques employing different Convolutional Neural Network architectures and models, such as U-Net and convolutional-attention transformer segmentation, were analyzed. Models were evaluated based on their performance across diverse datasets, and challenges, including domain shift and data sharing, were scrutinized.

RESULTS

Automatic segmentation methods offer a promising alternative to conventional measurement techniques, offering potential benefits in precision and efficiency. However, these methods are not without challenges, notably the "domain shift" that occurs when models trained on specific datasets underperform when applied to different datasets. Techniques such as domain adaptation, domain generalization, and data diversity were discussed as potential solutions.

CONCLUSIONS

Accurate measurement of VS growth is a complex process, with volumetric analysis currently appearing more reliable than linear measurements. Automatic segmentation, despite its challenges, offers a promising avenue for future investigation. Robust well-generalized models could potentially improve the efficiency of tracking tumor growth, thereby augmenting clinical decision-making. Further work needs to be done to develop more robust models, address the domain shift, and enable secure data sharing for wider applicability.

Identification and Evaluation of Blood-Based Biomarkers for Abdominal Aortic Aneurysm

INTRODUCTION

Blood-based biomarkers for abdominal aortic aneurysm (AAA) have been studied individually; however, we considered a panel of proteins to investigate AAA prognosis and its potential to improve predictive accuracy.

MATERIALS AND METHODS

Using a prospectively recruited cohort of patients with/without AAA (n = 452), we conducted a prognostic study to develop a model that accurately predicts AAA outcomes using clinical features and circulating biomarker levels. Serum concentrations of 9 biomarkers were measured at baseline, and the cohort was followed for 2 years. The primary outcome was major adverse aortic event (MAAE; composite of rapid AAA expansion [>0.5 cm/6 months or >1 cm/12 months], AAA intervention, or AAA rupture). Using 10-fold cross-validation, we trained a random forest model to predict 2 year MAAE using (1) clinical characteristics, (2) biomarkers, and (3) clinical characteristics and biomarkers.

RESULTS

Two-year MAAE occurred in 114 (25%) patients. Two proteins were significantly elevated in patients with AAA compared with those without AAA (angiopoietin-2 and aggrecan), composing the protein panel. For predicting 2 year MAAE, our random forest model achieved area under the receiver operating characteristic curve (AUROC) 0.74 using clinical features alone, and the addition of the 2-protein panel improved performance to AUROC 0.86.

CONCLUSIONS

Using a combination of clinical/biomarker data, we developed a model that accurately predicts 2 year MAAE.

Development and evaluation of a prediction model for peripheral artery disease-related major adverse limb events using novel biomarker data

OBJECTIVE

Prognostic tools for individuals with peripheral artery disease (PAD) remain limited. We developed prediction models for 3-year PAD-related major adverse limb events (MALE) using demographic, clinical, and biomarker data previously validated by our group.

METHODS

We performed a prognostic study using a prospectively recruited cohort of patients with PAD (n = 569). Demographic/clinical data were recorded including sex, age, comorbidities, previous procedures, and medications. Plasma concentrations of three biomarkers (N-terminal pro-B-type natriuretic peptide [NT-proBNP], fatty acid binding protein 3 [FABP3], and FABP4) were measured at baseline. The cohort was followed for 3 years. MALE was the primary outcome (composite of open/endovascular vascular intervention or major amputation). We trained three machine learning models with 10-fold cross-validation using demographic, clinical, and biomarker data (random forest, decision trees, and Extreme Gradient Boosting [XGBoost]) to predict 3-year MALE in patients. Area under the receiver operating characteristic curve (AUROC) was the primary model evaluation metric.

RESULTS

Three-year MALE was observed in 162 patients (29%). XGBoost was the top-performing predictive model for 3-year MALE, achieving the following performance metrics: AUROC = 0.88 (95% confidence interval [CI], 0.84-0.94); sensitivity, 88%; specificity, 84%; positive predictive value, 83%; and negative predictive value, 91% on test set data. On an independent validation cohort of patients with PAD, XGBoost attained an AUROC of 0.87 (95% CI, 0.82-0.90). The 10 most important predictors of 3-year MALE consisted of: (1) FABP3; (2) FABP4; (3) age; (4) NT-proBNP; (5) active smoking; (6) diabetes; (7) hypertension; (8) dyslipidemia; (9) coronary artery disease; and (10) sex.

CONCLUSIONS

We built robust machine learning algorithms that accurately predict 3-year MALE in patients with PAD using demographic, clinical, and novel biomarker data. Our algorithms can support risk stratification of patients with PAD for additional vascular evaluation and early aggressive medical management, thereby improving outcomes. Further validation of our models for clinical implementation is warranted.

Current status of artificial intelligence technologies in pituitary adenoma surgery: a scoping review

PURPOSE

Pituitary adenoma surgery is a complex procedure due to critical adjacent neurovascular structures, variations in size and extensions of the lesions, and potential hormonal imbalances. The integration of artificial intelligence (AI) and machine learning (ML) has demonstrated considerable potential in assisting neurosurgeons in decision-making, optimizing surgical outcomes, and providing real-time feedback. This scoping review comprehensively summarizes the current status of AI/ML technologies in pituitary adenoma surgery, highlighting their strengths and limitations.

METHODS

PubMed, Embase, Web of Science, and Scopus were searched following the PRISMA-ScR guidelines. Studies discussing the use of AI/ML in pituitary adenoma surgery were included. Eligible studies were grouped to analyze the different outcomes of interest of current AI/ML technologies.

RESULTS

Among the 2438 identified articles, 44 studies met the inclusion criteria, with a total of seventeen different algorithms utilized across all studies. Studies were divided into two groups based on their input type: clinicopathological and imaging input. The four main outcome variables evaluated in the studies included: outcome (remission, recurrence or progression, gross-total resection, vision improvement, and hormonal recovery), complications (CSF leak, readmission, hyponatremia, and hypopituitarism), cost, and adenoma-related factors (aggressiveness, consistency, and Ki-67 labeling) prediction. Three studies focusing on workflow analysis and real-time navigation were discussed separately.

CONCLUSION

AI/ML modeling holds promise for improving pituitary adenoma surgery by enhancing preoperative planning and optimizing surgical strategies. However, addressing challenges such as algorithm selection, performance evaluation, data heterogeneity, and ethics is essential to establish robust and reliable ML models that can revolutionize neurosurgical practice and benefit patients.

A machine learning algorithm for peripheral artery disease prognosis using biomarker data

Peripheral artery disease (PAD) biomarkers have been studied in isolation; however, an algorithm that considers a protein panel to inform PAD prognosis may improve predictive accuracy. Biomarker-based prediction models were developed and evaluated using a model development (n = 270) and prospective validation cohort (n = 277). Plasma concentrations of 37 proteins were measured at baseline and the patients were followed for 2 years. The primary outcome was 2-year major adverse limb event (MALE; composite of vascular intervention or major amputation). Of the 37 proteins tested, 6 were differentially expressed in patients with vs. without PAD (ADAMTS13, ICAM-1, ANGPTL3, Alpha 1-microglobulin, GDF15, and endostatin). Using 10-fold cross-validation, we developed a random forest machine learning model that accurately predicts 2-year MALE in a prospective validation cohort of PAD patients using a 6-protein panel (AUROC 0.84). This algorithm can support PAD risk stratification, informing clinical decisions on further vascular evaluation and management.

Using machine learning to predict outcomes following suprainguinal bypass

OBJECTIVE

Suprainguinal bypass for peripheral artery disease (PAD) carries important surgical risks; however, outcome prediction tools remain limited. We developed machine learning (ML) algorithms that predict outcomes following suprainguinal bypass.

METHODS

The Vascular Quality Initiative database was used to identify patients who underwent suprainguinal bypass for PAD between 2003 and 2023. We identified 100 potential predictor variables from the index hospitalization (68 preoperative [demographic/clinical], 13 intraoperative [procedural], and 19 postoperative [in-hospital course/complications]). The primary outcomes were major adverse limb events (MALE; composite of untreated loss of patency, thrombectomy/thrombolysis, surgical revision, or major amputation) or death at 1 year following suprainguinal bypass. Our data were split into training (70%) and test (30%) sets. Using 10-fold cross-validation, we trained six ML models using preoperative features (Extreme Gradient Boosting [XGBoost], random forest, Naïve Bayes classifier, support vector machine, artificial neural network, and logistic regression). The primary model evaluation metric was area under the receiver operating characteristic curve (AUROC). The best performing algorithm was further trained using intra- and postoperative data. Model robustness was evaluated using calibration plots and Brier scores. Performance was assessed on subgroups based on age, sex, race, ethnicity, rurality, median Area Deprivation Index, symptom status, procedure type, prior intervention for PAD, concurrent interventions, and urgency.

RESULTS

Overall, 16,832 patients underwent suprainguinal bypass, and 3136 (18.6%) developed 1-year MALE or death. Patients with 1-year MALE or death were older (mean age, 64.9 vs 63.5 years; P < .001) with more comorbidities, had poorer functional status (65.7% vs 80.9% independent at baseline; P < .001), and were more likely to have chronic limb-threatening ischemia (67.4% vs 47.6%; P < .001) than those without an outcome. Despite being at higher cardiovascular risk, they were less likely to receive acetylsalicylic acid or statins preoperatively and at discharge. Our best performing prediction model at the preoperative stage was XGBoost, achieving an AUROC of 0.92 (95% confidence interval [CI], 0.91-0.93). In comparison, logistic regression had an AUROC of 0.67 (95% CI, 0.65-0.69). Our XGBoost model maintained excellent performance at the intra- and postoperative stages, with AUROCs of 0.93 (95% CI, 0.92-0.94) and 0.98 (95% CI, 0.97-0.99), respectively. Calibration plots showed good agreement between predicted and observed event probabilities with Brier scores of 0.12 (preoperative), 0.11 (intraoperative), and 0.10 (postoperative). Of the top 10 predictors, nine were preoperative features including chronic limb-threatening ischemia, previous procedures, comorbidities, and functional status. Model performance remained robust on all subgroup analyses.

CONCLUSIONS

We developed ML models that accurately predict outcomes following suprainguinal bypass, performing better than logistic regression. Our algorithms have potential for important utility in guiding perioperative risk mitigation strategies to prevent adverse outcomes following suprainguinal bypass.

Predicting outcomes following lower extremity open revascularization using machine learning

Lower extremity open revascularization is a treatment option for peripheral artery disease that carries significant peri-operative risks; however, outcome prediction tools remain limited. Using machine learning (ML), we developed automated algorithms that predict 30-day outcomes following lower extremity open revascularization. The National Surgical Quality Improvement Program targeted vascular database was used to identify patients who underwent lower extremity open revascularization for chronic atherosclerotic disease between 2011 and 2021. Input features included 37 pre-operative demographic/clinical variables. The primary outcome was 30-day major adverse limb event (MALE; composite of untreated loss of patency, major reintervention, or major amputation) or death. Our data were split into training (70%) and test (30%) sets. Using tenfold cross-validation, we trained 6 ML models. Overall, 24,309 patients were included. The primary outcome of 30-day MALE or death occurred in 2349 (9.3%) patients. Our best performing prediction model was XGBoost, achieving an area under the receiver operating characteristic curve (95% CI) of 0.93 (0.92-0.94). The calibration plot showed good agreement between predicted and observed event probabilities with a Brier score of 0.08. Our ML algorithm has potential for important utility in guiding risk mitigation strategies for patients being considered for lower extremity open revascularization to improve outcomes.

Development and Internal Validation of Machine Learning Models to Predict Mortality and Disability After Mechanical Thrombectomy for Acute Anterior Circulation Large Vessel Occlusion

OBJECTIVE

Mechanical thrombectomy (MT) improves outcomes in patients with LVO but many still experience mortality or severe disability. We sought to develop machine learning (ML) models that predict 90-day outcomes after MT for LVO.

METHODS

Consecutive patients who underwent MT for LVO between 2015-2021 at a Comprehensive Stroke Center were reviewed. Outcomes included 90-day favorable functional status (mRS 0-2), severe disability (mRS 4-6), and mortality. ML models were trained for each outcome using prethrombectomy data (pre) and with thrombectomy data (post).

RESULTS

Three hundred and fifty seven patients met the inclusion criteria. After model screening and hyperparameter tuning the top performing ML model for each outcome and timepoint was random forest (RF). Using only prethrombectomy features, the AUCs for the RFpre models were 0.73 (95% CI 0.62-0.85) for favorable functional status, 0.77 (95% CI 0.65-0.86) for severe disability, and 0.78 (95% CI 0.64-0.88) for mortality. All of these were better than a standard statistical model except for favorable functional status. Each RF model outperformed Pre, SPAN-100, THRIVE, and HIAT scores (P < 0.0001 for all). The most predictive features were premorbid mRS, age, and NIHSS. Incorporating MT data, the AUCs for the RFpost models were 0.80 (95% CI 0.67-0.90) for favorable functional status, 0.82 (95% CI 0.69-0.91) for severe disability, and 0.71 (95% CI 0.55-0.84) for mortality.

CONCLUSIONS

RF models accurately predicted 90-day outcomes after MT and performed better than standard statistical and clinical prediction models.

Exploring the Landscape of Machine Learning Applications in Neurosurgery: A Bibliometric Analysis and Narrative Review of Trends and Future Directions

BACKGROUND

The field of neurosurgery has consistently represented an area of innovation and integration of technology since its inception. As such, machine learning (ML) has found its way into applications within neurosurgery relatively rapidly. Through this bibliometric review and cluster analysis, we seek to identify trends and emerging applications of ML within neurosurgery.

METHODS

A bibliometric analysis was carried out in the Web of Science database on publications from January 2000 to March 2023. The full data set of the 200 most cited publications including title, author information, journal, citation count, keywords, and abstracts for each publication was evaluated in CiteSpace. CiteSpace was used to elucidate publication characteristics, trends, and topic clusters via collaborate network analysis using the Kamada-Kawai algorithm.

RESULTS

The 25 most cited titles were included in our analysis. Harvard University and its affiliates represented the top institution, contributing nearly 25% of publications in the literature. WORLD NEUROSURGERY was the journal with the highest net citation count of 747 (29%). Collaborative network analysis generated 12 unique clusters, the largest of which was machine learning, followed by feature importance and deep brain stimulation.

CONCLUSION

This review highlights the most impactful articles pertaining to ML in the field of neurosurgery. ML has been applied into several sub-specialties within neurosurgery to optimize patient care, with special attention to outcome predictors, patient selection, and surgical decision making.

Machine Learning in Neurosurgery: Toward Complex Inputs, Actionable Predictions, and Generalizable Translations

Machine learning can predict neurosurgical diagnosis and outcomes, power imaging analysis, and perform robotic navigation and tumor labeling. State-of-the-art models can reconstruct and generate images, predict surgical events from video, and assist in intraoperative decision-making. In this review, we will detail the neurosurgical applications of machine learning, ranging from simple to advanced models, and their potential to transform patient care. As machine learning techniques, outputs, and methods become increasingly complex, their performance is often more impactful yet increasingly difficult to evaluate. We aim to introduce these advancements to the neurosurgical audience while suggesting major potential roadblocks to their safe and effective translation. Unlike the previous generation of machine learning in neurosurgery, the safe translation of recent advancements will be contingent on neurosurgeons' involvement in model development and validation.

Machine Learning Approach for Analyzing 3-Year Outcomes of Patients with Brain Arteriovenous Malformation (AVM) after Stereotactic Radiosurgery (SRS)

A cerebral arteriovenous malformation (AVM) is a tangle of abnormal blood vessels that irregularly connects arteries and veins. Stereotactic radiosurgery (SRS) has been shown to be an effective treatment for AVM patients, but the factors associated with AVM obliteration remains a matter of debate. In this study, we aimed to develop a model that can predict whether patients with AVM will be cured 36 months after intervention by means of SRS and identify the most important predictors that explain the probability of being cured. A machine learning (ML) approach was applied using decision tree (DT) and logistic regression (LR) techniques on historical data (sociodemographic, clinical, treatment, angioarchitecture, and radiosurgery procedure) of 202 patients with AVM who underwent SRS at the Instituto de Radiocirugía del Perú (IRP) between 2005 and 2018. The LR model obtained the best results for predicting AVM cure with an accuracy of 0.92, sensitivity of 0.93, specificity of 0.89, and an area under the curve (AUC) of 0.98, which shows that ML models are suitable for predicting the prognosis of medical conditions such as AVM and can be a support tool for medical decision-making. In addition, several factors were identified that could explain whether patients with AVM would be cured at 36 months with the highest likelihood: the location of the AVM, the occupation of the patient, and the presence of hemorrhage.

Machine learning based outcome prediction of microsurgically treated unruptured intracranial aneurysms

Machine learning (ML) has revolutionized data processing in recent years. This study presents the results of the first prediction models based on a long-term monocentric data registry of patients with microsurgically treated unruptured intracranial aneurysms (UIAs) using a temporal train-test split. Temporal train-test splits allow to simulate prospective validation, and therefore provide more accurate estimations of a model's predictive quality when applied to future patients. ML models for the prediction of the Glasgow outcome scale, modified Rankin Scale (mRS), and new transient or permanent neurological deficits (output variables) were created from all UIA patients that underwent microsurgery at the Kepler University Hospital Linz (Austria) between 2002 and 2020 (n = 466), based on 18 patient- and 10 aneurysm-specific preoperative parameters (input variables). Train-test splitting was performed with a temporal split for outcome prediction in microsurgical therapy of UIA. Moreover, an external validation was conducted on an independent external data set (n = 256) of the Department of Neurosurgery, University Medical Centre Hamburg-Eppendorf. In total, 722 aneurysms were included in this study. A postoperative mRS > 2 was best predicted by a quadratic discriminant analysis (QDA) estimator in the internal test set, with an area under the receiver operating characteristic curve (ROC-AUC) of 0.87 ± 0.03 and a sensitivity and specificity of 0.83 ± 0.08 and 0.71 ± 0.07, respectively. A Multilayer Perceptron predicted the post- to preoperative mRS difference > 1 with a ROC-AUC of 0.70 ± 0.02 and a sensitivity and specificity of 0.74 ± 0.07 and 0.50 ± 0.04, respectively. The QDA was the best model for predicting a permanent new neurological deficit with a ROC-AUC of 0.71 ± 0.04 and a sensitivity and specificity of 0.65 ± 0.24 and 0.60 ± 0.12, respectively. Furthermore, these models performed significantly better than the classic logistic regression models (p < 0.0001). The present results showed good performance in predicting functional and clinical outcomes after microsurgical therapy of UIAs in the internal data set, especially for the main outcome parameters, mRS and permanent neurological deficit. The external validation showed poor discrimination with ROC-AUC values of 0.61, 0.53 and 0.58 respectively for predicting a postoperative mRS > 2, a pre- and postoperative difference in mRS > 1 point and a GOS < 5. Therefore, generalizability of the models could not be demonstrated in the external validation. A SHapley Additive exPlanations (SHAP) analysis revealed that this is due to the most important features being distributed quite differently in the internal and external data sets. The implementation of newly available data and the merging of larger databases to form more broad-based predictive models is imperative in the future.

Machine learning-based monosaccharide profiling for tissue-specific classification of Wolfiporia extensa samples

Machine learning (ML) has been used for many clinical decision-making processes and diagnostic procedures in bioinformatics applications. We examined eight algorithms, including linear discriminant analysis (LDA), logistic regression (LR), k-nearest neighbor (KNN), random forest (RF), gradient boosting machine (GBM), support vector machine (SVM), Naïve Bayes classifier (NB), and artificial neural network (ANN) models, to evaluate their classification and prediction capabilities for four tissue types in Wolfiporia extensa using their monosaccharide composition profiles. All 8 ML-based models were assessed as exemplary models with AUC exceeding 0.8. Five models, namely LDA, KNN, RF, GBM, and ANN, performed excellently in the four-tissue-type classification (AUC > 0.9). Additionally, all eight models were evaluated as good predictive models with AUC value > 0.8 in the three-tissue-type classification. Notably, all 8 ML-based methods outperformed the single linear discriminant analysis (LDA) plotting method. For large sample sizes, the ML-based methods perform better than traditional regression techniques and could potentially increase the accuracy in identifying tissue samples of W. extensa.

Using machine learning to predict outcomes following open abdominal aortic aneurysm repair

OBJECTIVE

Prediction of outcomes following open abdominal aortic aneurysm (AAA) repair remains challenging with a lack of widely used tools to guide perioperative management. We developed machine learning (ML) algorithms that predict outcomes following open AAA repair.

METHODS

The Vascular Quality Initiative (VQI) database was used to identify patients who underwent elective open AAA repair between 2003 and 2023. Input features included 52 preoperative demographic/clinical variables. All available preoperative variables from VQI were used to maximize predictive performance. The primary outcome was in-hospital major adverse cardiovascular event (MACE; composite of myocardial infarction, stroke, or death). Secondary outcomes were individual components of the primary outcome, other in-hospital complications, and 1-year mortality and any reintervention. We split our data into training (70%) and test (30%) sets. Using 10-fold cross-validation, six ML models were trained using preoperative features (Extreme Gradient Boosting [XGBoost], random forest, Naïve Bayes classifier, support vector machine, artificial neural network, and logistic regression). The primary model evaluation metric was area under the receiver operating characteristic curve (AUROC). Model robustness was evaluated with calibration plot and Brier score. The top 10 predictive features in our final model were determined based on variable importance scores. Performance was assessed on subgroups based on age, sex, race, ethnicity, rurality, median area deprivation index, proximal clamp site, prior aortic surgery, and concomitant procedures.

RESULTS

Overall, 12,027 patients were included. The primary outcome of in-hospital MACE occurred in 630 patients (5.2%). Compared with patients without a primary outcome, those who developed in-hospital MACE were older with more comorbidities, demonstrated poorer functional status, had more complex aneurysms, and were more likely to require concomitant procedures. Our best performing prediction model for in-hospital MACE was XGBoost, achieving an AUROC of 0.93 (95% confidence interval, 0.92-0.94). Comparatively, logistic regression had an AUROC of 0.71 (95% confidence interval, 0.70-0.73). For secondary outcomes, XGBoost achieved AUROCs between 0.84 and 0.94. The calibration plot showed good agreement between predicted and observed event probabilities with a Brier score of 0.05. These findings highlight the excellent predictive performance of the XGBoost model. The top three predictive features in our algorithm for in-hospital MACE following open AAA repair were: (1) coronary artery disease; (2) American Society of Anesthesiologists classification; and (3) proximal clamp site. Model performance remained robust on all subgroup analyses.

CONCLUSIONS

Open AAA repair outcomes can be accurately predicted using preoperative data with our ML models, which perform better than logistic regression. Our automated algorithms can help guide risk-mitigation strategies for patients being considered for open AAA repair to improve outcomes.

A Pilot Remote Curriculum to Enhance Resident and Medical Student Understanding of Machine Learning in Healthcare

BACKGROUND

Despite the expanding role of machine learning (ML) in health care and patient expectations for clinicians to understand ML-based tools, few for-credit curricula exist specifically for neurosurgical trainees to learn basic principles and implications of ML for medical research and clinical practice. We implemented a novel, remotely delivered curriculum designed to develop literacy in ML for neurosurgical trainees.

METHODS

A 4-week pilot medical elective was designed specifically for trainees to build literacy in basic ML concepts. Qualitative feedback from interested and enrolled students was collected to assess students' and trainees' reactions, learning, and future application of course content.

RESULTS

Despite 15 interested learners, only 3 medical students and 1 neurosurgical resident completed the course. Enrollment included students and trainees from 3 different institutions. All learners who completed the course found the lectures relevant to their future practice as clinicians and researchers and reported improved confidence in applying and understanding published literature applying ML techniques in health care. Barriers to ample enrollment and retention (e.g., balancing clinical responsibilities) were identified.

CONCLUSIONS

This pilot elective demonstrated the interest, value, and feasibility of a remote elective to establish ML literacy; however, feedback to increase accessibility and flexibility of the course encouraged our team to implement changes. Future elective iterations will have a semiannual, 2-week format, splitting lectures more clearly between theory (the method and its value) and application (coding instructions) and will make lectures open-source prerequisites to allow tailoring of student learning to their planned application of these methods in their practice and research.

Predicting outcomes following open revascularization for aortoiliac occlusive disease using machine learning

OBJECTIVE

Open surgical treatment options for aortoiliac occlusive disease carry significant perioperative risks; however, outcome prediction tools remain limited. Using machine learning (ML), we developed automated algorithms that predict 30-day outcomes following open aortoiliac revascularization.

METHODS

The National Surgical Quality Improvement Program (NSQIP) targeted vascular database was used to identify patients who underwent open aortoiliac revascularization for atherosclerotic disease between 2011 and 2021. Input features included 38 preoperative demographic/clinical variables. The primary outcome was 30-day major adverse limb event (MALE; composite of untreated loss of patency, major reintervention, or major amputation) or death. The 30-day secondary outcomes were individual components of the primary outcome, major adverse cardiovascular event (MACE; composite of myocardial infarction, stroke, or death), individual components of MACE, wound complication, bleeding, other morbidity, non-home discharge, and unplanned readmission. Our data were split into training (70%) and test (30%) sets. Using 10-fold cross-validation, we trained six ML models using preoperative features. The primary model evaluation metric was area under the receiver operating characteristic curve (AUROC). Model robustness was evaluated with calibration plot and Brier score. Variable importance scores were calculated to determine the top 10 predictive features. Performance was assessed on subgroups based on age, sex, race, ethnicity, symptom status, procedure type, and urgency.

RESULTS

Overall, 9649 patients were included. The primary outcome of 30-day MALE or death occurred in 1021 patients (10.6%). Our best performing prediction model for 30-day MALE or death was XGBoost, achieving an AUROC of 0.95 (95% confidence interval [CI], 0.94-0.96). In comparison, logistic regression had an AUROC of 0.79 (95% CI, 0.77-0.81). For 30-day secondary outcomes, XGBoost achieved AUROCs between 0.87 and 0.97 (untreated loss of patency [0.95], major reintervention [0.88], major amputation [0.96], death [0.97], MACE [0.95], myocardial infarction [0.88], stroke [0.93], wound complication [0.94], bleeding [0.87], other morbidity [0.96], non-home discharge [0.90], and unplanned readmission [0.91]). The calibration plot showed good agreement between predicted and observed event probabilities with a Brier score of 0.05. The strongest predictive feature in our algorithm was chronic limb-threatening ischemia. Model performance remained robust on all subgroup analyses of specific demographic/clinical populations.

CONCLUSIONS

Our ML models accurately predict 30-day outcomes following open aortoiliac revascularization using preoperative data, performing better than logistic regression. They have potential for important utility in guiding risk-mitigation strategies for patients being considered for open aortoiliac revascularization to improve outcomes.

Machine learning to predict outcomes following endovascular abdominal aortic aneurysm repair

BACKGROUND

Endovascular aneurysm repair (EVAR) for abdominal aortic aneurysm (AAA) carries important perioperative risks; however, there are no widely used outcome prediction tools. The aim of this study was to apply machine learning (ML) to develop automated algorithms that predict 1-year mortality following EVAR.

METHODS

The Vascular Quality Initiative database was used to identify patients who underwent elective EVAR for infrarenal AAA between 2003 and 2023. Input features included 47 preoperative demographic/clinical variables. The primary outcome was 1-year all-cause mortality. Data were split into training (70 per cent) and test (30 per cent) sets. Using 10-fold cross-validation, 6 ML models were trained using preoperative features with logistic regression as the baseline comparator. The primary model evaluation metric was area under the receiver operating characteristic curve (AUROC). Model robustness was evaluated with calibration plot and Brier score.

RESULTS

Some 63 655 patients were included. One-year mortality occurred in 3122 (4.9 per cent) patients. The best performing prediction model for 1-year mortality was XGBoost, achieving an AUROC (95 per cent c.i.) of 0.96 (0.95-0.97). Comparatively, logistic regression had an AUROC (95 per cent c.i.) of 0.69 (0.68-0.71). The calibration plot showed good agreement between predicted and observed event probabilities with a Brier score of 0.04. The top 3 predictive features in the algorithm were 1) unfit for open AAA repair, 2) functional status, and 3) preoperative dialysis.

CONCLUSIONS

In this data set, machine learning was able to predict 1-year mortality following EVAR using preoperative data and outperformed standard logistic regression models.

Computer-vision based analysis of the neurosurgical scene - A systematic review

Introduction

With increasing use of robotic surgical adjuncts, artificial intelligence and augmented reality in neurosurgery, the automated analysis of digital images and videos acquired over various procedures becomes a subject of increased interest. While several computer vision (CV) methods have been developed and implemented for analyzing surgical scenes, few studies have been dedicated to neurosurgery.

Research question

In this work, we present a systematic literature review focusing on CV methodologies specifically applied to the analysis of neurosurgical procedures based on intra-operative images and videos. Additionally, we provide recommendations for the future developments of CV models in neurosurgery.

Material and methods

We conducted a systematic literature search in multiple databases until January 17, 2023, including Web of Science, PubMed, IEEE Xplore, Embase, and SpringerLink.

Results

We identified 17 studies employing CV algorithms on neurosurgical videos/images. The most common applications of CV were tool and neuroanatomical structure detection or characterization, and to a lesser extent, surgical workflow analysis. Convolutional neural networks (CNN) were the most frequently utilized architecture for CV models (65%), demonstrating superior performances in tool detection and segmentation. In particular, mask recurrent-CNN manifested most robust performance outcomes across different modalities.

Discussion and conclusion

Our systematic review demonstrates that CV models have been reported that can effectively detect and differentiate tools, surgical phases, neuroanatomical structures, as well as critical events in complex neurosurgical scenes with accuracies above 95%. Automated tool recognition contributes to objective characterization and assessment of surgical performance, with potential applications in neurosurgical training and intra-operative safety management.

A Machine Learning Approach to Predict Postoperative Pancreatic Fistula After Pancreaticoduodenectomy Using Only Preoperatively Known Data

BACKGROUND

Clinically-relevant postoperative pancreatic fistula (CR-POPF) following pancreaticoduodenectomy (PD) is a major postoperative complication and the primary determinant of surgical outcomes. However, the majority of current risk calculators utilize intraoperative and postoperative variables, limiting their utility in the preoperative setting. Therefore, we aimed to develop a user-friendly risk calculator to predict CR-POPF following PD using state-of-the-art machine learning (ML) algorithms and only preoperatively known variables.

METHODS

Adult patients undergoing elective PD for non-metastatic pancreatic cancer were identified from the ACS-NSQIP targeted pancreatectomy dataset (2014-2019). The primary endpoint was development of CR-POPF (grade B or C). Secondary endpoints included discharge to facility, 30-day mortality, and a composite of overall and significant complications. Four models (logistic regression, neural network, random forest, and XGBoost) were trained, validated and a user-friendly risk calculator was then developed.

RESULTS

Of the 8666 patients who underwent elective PD, 13% (n = 1160) developed CR-POPF. XGBoost was the best performing model (AUC = 0.72), and the top five preoperative variables associated with CR-POPF were non-adenocarcinoma histology, lack of neoadjuvant chemotherapy, pancreatic duct size less than 3 mm, higher BMI, and higher preoperative serum creatinine. Model performance for 30-day mortality, discharge to a facility, and overall and significant complications ranged from AUC 0.62-0.78.

CONCLUSIONS

In this study, we developed and validated an ML model using only preoperatively known variables to predict CR-POPF following PD. The risk calculator can be used in the preoperative setting to inform clinical decision-making and patient counseling.

Artificial Intelligence Modeling to Predict Periprosthetic Infection and Explantation following Implant-Based Reconstruction

BACKGROUND

Despite improvements in prosthesis design and surgical techniques, periprosthetic infection and explantation rates following implant-based reconstruction (IBR) remain relatively high. Artificial intelligence is an extremely powerful predictive tool that involves machine learning (ML) algorithms. We sought to develop, validate, and evaluate the use of ML algorithms to predict complications of IBR.

METHODS

A comprehensive review of patients who underwent IBR from January of 2018 to December of 2019 was conducted. Nine supervised ML algorithms were developed to predict periprosthetic infection and explantation. Patient data were randomly divided into training (80%) and testing (20%) sets.

RESULTS

The authors identified 481 patients (694 reconstructions) with a mean ± SD age of 50.0 ± 11.5 years, mean ± SD body mass index of 26.7 ± 4.8 kg/m 2 , and median follow-up time of 16.1 months (range, 11.9 to 3.2 months). Periprosthetic infection developed in 113 of the reconstructions (16.3%), and explantation was required with 82 (11.8%) of them. ML demonstrated good discriminatory performance in predicting periprosthetic infection and explantation (area under the receiver operating characteristic curve, 0.73 and 0.78, respectively), and identified nine and 12 significant predictors of periprosthetic infection and explantation, respectively.

CONCLUSIONS

ML algorithms trained using readily available perioperative clinical data accurately predict periprosthetic infection and explantation following IBR. The authors' findings support incorporating ML models into perioperative assessment of patients undergoing IBR to provide data-driven, patient-specific risk assessment to aid individualized patient counseling, shared decision-making, and presurgical optimization.

Machine learning models to predict traumatic brain injury outcomes in Tanzania: Using delays to emergency care as predictors

Constraints to emergency department resources may prevent the timely provision of care following a patient's arrival to the hospital. In-hospital delays may adversely affect health outcomes, particularly among trauma patients who require prompt management. Prognostic models can help optimize resource allocation thereby reducing in-hospital delays and improving trauma outcomes. The objective of this study was to investigate the predictive value of delays to emergency care in machine learning based traumatic brain injury (TBI) prognostic models. Our data source was a TBI registry from Kilimanjaro Christian Medical Centre Emergency Department in Moshi, Tanzania. We created twelve unique variables representing delays to emergency care and included them in eight different machine learning based TBI prognostic models that predict in-hospital outcome. Model performance was compared using the area under the receiver operating characteristic curve (AUC). Inclusion of our twelve time to care variables improved predictability in each of our eight prognostic models. Our Bayesian generalized linear model produced the largest AUC, with a value of 89.5 (95% CI: 88.8, 90.3). Time to care variables were among the most important predictors of in-hospital outcome in our best three performing models. In low-resource settings where delays to care are highly prevalent and contribute to high mortality rates, incorporation of care delays into prediction models that support clinical decision making may benefit both emergency medicine physicians and trauma patients by improving prognostication performance.

Using machine learning to predict outcomes following carotid endarterectomy

OBJECTIVE

Prediction of outcomes following carotid endarterectomy (CEA) remains challenging, with a lack of standardized tools to guide perioperative management. We used machine learning (ML) to develop automated algorithms that predict outcomes following CEA.

METHODS

The Vascular Quality Initiative (VQI) database was used to identify patients who underwent CEA between 2003 and 2022. We identified 71 potential predictor variables (features) from the index hospitalization (43 preoperative [demographic/clinical], 21 intraoperative [procedural], and 7 postoperative [in-hospital complications]). The primary outcome was stroke or death at 1 year following CEA. Our data were split into training (70%) and test (30%) sets. Using 10-fold cross-validation, we trained six ML models using preoperative features (Extreme Gradient Boosting [XGBoost], random forest, Naïve Bayes classifier, support vector machine, artificial neural network, and logistic regression). The primary model evaluation metric was area under the receiver operating characteristic curve (AUROC). After selecting the best performing algorithm, additional models were built using intra- and postoperative data. Model robustness was evaluated using calibration plots and Brier scores. Performance was assessed on subgroups based on age, sex, race, ethnicity, insurance status, symptom status, and urgency of surgery.

RESULTS

Overall, 166,369 patients underwent CEA during the study period. In total, 7749 patients (4.7%) had the primary outcome of stroke or death at 1 year. Patients with an outcome were older with more comorbidities, had poorer functional status, and demonstrated higher risk anatomic features. They were also more likely to undergo intraoperative surgical re-exploration and have in-hospital complications. Our best performing prediction model at the preoperative stage was XGBoost, achieving an AUROC of 0.90 (95% confidence interval [CI], 0.89-0.91). In comparison, logistic regression had an AUROC of 0.65 (95% CI, 0.63-0.67), and existing tools in the literature demonstrate AUROCs ranging from 0.58 to 0.74. Our XGBoost models maintained excellent performance at the intra- and postoperative stages, with AUROCs of 0.90 (95% CI, 0.89-0.91) and 0.94 (95% CI, 0.93-0.95), respectively. Calibration plots showed good agreement between predicted and observed event probabilities with Brier scores of 0.15 (preoperative), 0.14 (intraoperative), and 0.11 (postoperative). Of the top 10 predictors, eight were preoperative features, including comorbidities, functional status, and previous procedures. Model performance remained robust on all subgroup analyses.

CONCLUSIONS

We developed ML models that accurately predict outcomes following CEA. Our algorithms perform better than logistic regression and existing tools, and therefore, have potential for important utility in guiding perioperative risk mitigation strategies to prevent adverse outcomes.

Assessing treatment switch among patients with multiple sclerosis: A machine learning approach

BACKGROUND

Patients with multiple sclerosis (MS) frequently switch their Disease-Modifying Agents (DMA) for effectiveness and safety concerns. This study aimed to develop and compare the random forest (RF) machine learning (ML) model with the logistic regression (LR) model for predicting DMA switching among MS patients.

METHODS

This retrospective longitudinal study used the TriNetX data from a federated electronic medical records (EMR) network. Between September 2010 and May 2017, adults (aged ≥18) MS patients with ≥1 DMA prescription were identified, and the earliest DMA date was assigned as the index date. Patients prescribed any DMAs different from their index DMAs were considered as treatment switch. . The RF and LR models were built with 72 baseline characteristics and trained with 70% of the randomly split data after up-sampling. Area Under the Curves (AUC), accuracy, recall, G-measure, and F-1 score were used to evaluate the model performance.

RESULTS

In this study, 7258 MS patients with ≥1 DMA were identified. Within two years, 16% of MS patients switched to a different DMA. The RF model obtained significantly better discrimination than the LR model (AUC = 0.65 vs. 0.63, p < 0.0001); however, the RF model had a similar predictive performance to the LR model with respect to F- and G-measures (RF: 72% and 73% vs. LR: 72% and 73%, respectively). The most influential features identified from the RF model were age, type of index medication, and year of index.

CONCLUSIONS

Compared to the LR model, RF performed better in predicting DMA switch in MS patients based on AUC measures; however, judged by F- and G-measures, the RF model performed similarly to LR. Further research is needed to understand the role of ML techniques in predicting treatment outcomes for the decision-making process to achieve optimal treatment goals.

Application of Machine Learning to Predict Hearing Outcomes of Tympanoplasty

OBJECTIVE

This retrospective study aimed to evaluate the performance of machine learning techniques in predicting air-bone gap after tympanoplasty compared with conventional scoring models and to identify the influential factors.

METHODS

We reviewed the charts of 105 patients (114 ears) with chronic otitis media who underwent tympanoplasty. Two numerical scoring systems (middle ear risk index [MERI] and ossiculoplasty outcome parameter staging [OOPS]) and three algorithms (random forest [RF], support vector machine [SVM], and k nearest neighbor [kNN]) were created. Experimental variables included age, preoperative air-bone gap, soft-tissue density lesion in the tympanic cavity in CT, otorrhea, surgical history, ossicular bone problems in CT, tympanic perforation location, perforation type (central or marginal), grafting material, smoking history, endoscopy use, and the operator whose experience was 20 years or longer, or shorter. Binary classification, postoperative air-bone gap ≤15 or >15 dB, and multiclass classification, classification into seven categories by 10 dB, were performed, and the percentages of correct prediction were calculated. The importance of features in the RF model was calculated to identify influential factors.

RESULTS

The percentages of correct prediction in binary classification were 62.3%, 72.8%, 81.5%, 81.5%, and 81.5% in MERI, OOPS, RF, SVM, and kNN, respectively, and those in multiclass classification were 29.8%, 21.9%, 63.1%, 44.7%, and 50% in the same order. The RF model suggested larger preoperative air-bone gap, and older age could make the postoperative air-bone gap larger.

CONCLUSION

The machine learning techniques, especially the RF model, are promising methods for precise postoperative air-bone gap prediction.

LEVEL OF EVIDENCE

4 Laryngoscope, 133:2371-2378, 2023.

FDA-approved machine learning algorithms in neuroradiology: A systematic review of the current evidence for approval

Over the past decade, machine learning (ML) and artificial intelligence (AI) have become increasingly prevalent in the medical field. In the United States, the Food and Drug Administration (FDA) is responsible for regulating AI algorithms as "medical devices" to ensure patient safety. However, recent work has shown that the FDA approval process may be deficient. In this study, we evaluate the evidence supporting FDA-approved neuroalgorithms, the subset of machine learning algorithms with applications in the central nervous system (CNS), through a systematic review of the primary literature. Articles covering the 53 FDA-approved algorithms with applications in the CNS published in PubMed, EMBASE, Google Scholar and Scopus between database inception and January 25, 2022 were queried. Initial searches identified 1505 studies, of which 92 articles met the criteria for extraction and inclusion. Studies were identified for 26 of the 53 neuroalgorithms, of which 10 algorithms had only a single peer-reviewed publication. Performance metrics were available for 15 algorithms, external validation studies were available for 24 algorithms, and studies exploring the use of algorithms in clinical practice were available for 7 algorithms. Papers studying the clinical utility of these algorithms focused on three domains: workflow efficiency, cost savings, and clinical outcomes. Our analysis suggests that there is a meaningful gap between the FDA approval of machine learning algorithms and their clinical utilization. There appears to be room for process improvement by implementation of the following recommendations: the provision of compelling evidence that algorithms perform as intended, mandating minimum sample sizes, reporting of a predefined set of performance metrics for all algorithms and clinical application of algorithms prior to widespread use. This work will serve as a baseline for future research into the ideal regulatory framework for AI applications worldwide.

Advances in artificial intelligence, robotics, augmented and virtual reality in neurosurgery

Neurosurgical practitioners undergo extensive and prolonged training to acquire diverse technical proficiencies, while neurosurgical procedures necessitate a substantial amount of pre-, post-, and intraoperative clinical data acquisition, making decisions, attention, and convalescence. The past decade witnessed an appreciable escalation in the significance of artificial intelligence (AI) in neurosurgery. AI holds significant potential in neurosurgery as it supplements the abilities of neurosurgeons to offer optimal interventional and non-interventional care to patients by improving prognostic and diagnostic outcomes in clinical therapy and assisting neurosurgeons in making decisions while surgical interventions to enhance patient outcomes. Other technologies including augmented reality, robotics, and virtual reality can assist and promote neurosurgical methods as well. Moreover, they play a significant role in generating, processing, as well as storing experimental and clinical data. Also, the usage of these technologies in neurosurgery is able to curtail the number of costs linked with surgical care and extend high-quality health care to a wider populace. This narrative review aims to integrate the results of articles that elucidate the role of the aforementioned technologies in neurosurgery.

Machine Learning Approach for the Prediction of In-Hospital Mortality in Traumatic Brain Injury Using Bio-Clinical Markers at Presentation to the Emergency Department

BACKGROUND

Accurate prediction of in-hospital mortality is essential for better management of patients with traumatic brain injury (TBI). Machine learning (ML) algorithms have been shown to be effective in predicting clinical outcomes. This study aimed to identify predictors of in-hospital mortality in TBI patients using ML algorithms.

MATERIALS AND METHOD

A retrospective study was performed using data from both the trauma registry and electronic medical records among TBI patients admitted to the Hamad Trauma Center in Qatar between June 2016 and May 2021. Thirteen features were selected for four ML models including a Support Vector Machine (SVM), Logistic Regression (LR), Random Forest (RF), and Extreme Gradient Boosting (XgBoost), to predict the in-hospital mortality.

RESULTS

A dataset of 922 patients was analyzed, of which 78% survived and 22% died. The AUC scores for SVM, LR, XgBoost, and RF models were 0.86, 0.84, 0.85, and 0.86, respectively. XgBoost and RF had good AUC scores but exhibited significant differences in log loss between the training and testing sets (% difference in logloss of 79.5 and 41.8, respectively), indicating overfitting compared to the other models. The feature importance trend across all models indicates that aPTT, INR, ISS, prothrombin time, and lactic acid are the most important features in prediction. Magnesium also displayed significant importance in the prediction of mortality among serum electrolytes.

CONCLUSIONS

SVM was found to be the best-performing ML model in predicting the mortality of TBI patients. It had the highest AUC score and did not show overfitting, making it a more reliable model compared to LR, XgBoost, and RF.

Demonstrating the successful application of synthetic learning in spine surgery for training multi-center models with increased patient privacy

From real-time tumor classification to operative outcome prediction, applications of machine learning to neurosurgery are powerful. However, the translation of many of these applications are restricted by the lack of "big data" in neurosurgery. Important restrictions in patient privacy and sharing of imaging data reduce the diversity of the datasets used to train resulting models and therefore limit generalizability. Synthetic learning is a recent development in machine learning that generates synthetic data from real data and uses the synthetic data to train downstream models while preserving patient privacy. Such an approach has yet to be successfully demonstrated in the spine surgery domain. Spine radiographs were collected from the VinDR-SpineXR dataset, with 1470 labeled as abnormal and 2303 labeled as normal. A conditional generative adversarial network (GAN) was trained on the radiographs to generate a spine radiograph and normal/abnormal label. A modified conditional GAN (SpineGAN) was trained on the same task. A convolutional neural network (CNN) was trained using the real data to label abnormal radiographs. A CNN was trained to label abnormal radiographs using synthetic images from the GAN and in a separate experiment from SpineGAN. Using the real radiographs, an AUC of 0.856 was achieved in abnormality classification. Training on synthetic data generated by the standard GAN (AUC of 0.814) and synthetic data generated by our SpineGAN (AUC of 0.830) resulted in similar classifier performance. SpineGAN generated images with higher FID and lower precision scores, but with higher recall and increased performance when used for synthetic learning. The successful application of synthetic learning was demonstrated in the spine surgery domain for the classification of spine radiographs as abnormal or normal. A modified domain-relevant GAN is introduced for the generation of spine images, evidencing the importance of domain-relevant generation techniques in synthetic learning. Synthetic learning can allow neurosurgery to use larger and more diverse patient imaging sets to train more generalizable algorithms with greater patient privacy.

Association of the classification of intraoperative adverse events (ClassIntra) with complications and neurological outcome after neurosurgical procedures: a prospective cohort study

PURPOSE

To analyze the reliability of the classification of intraoperative adverse events (ClassIntra) to reflect intraoperative complications of neurosurgical procedures and the potential to predict the postoperative outcome including the neurological performance. The ClassIntra classification was recently introduced and found to be reliable for assessing intraoperative adverse events and predicting postoperative complications across different surgical disciplines. Nevertheless, its potential role for neurosurgical procedures remains elusive.

METHODS

This is a prospective, monocentric cohort study assessing the ClassIntra in 422 adult patients who underwent a neurosurgical procedure and were hospitalized between July 1, 2021, to December 31, 2021. The primary outcome was the occurrence of intraoperative complications graded according to ClassIntra and the association with postoperative outcome reflected by the Clavien-Dindo classification and comprehensive complication index (CCI). The ClassIntra is defined as intraoperative adverse events as any deviation from the ideal course on a grading scale from grade 0 (no deviation) to grade V (intraoperative death) and was set at sign-out in agreement between neurosurgeon and anesthesiologist. Secondary outcomes were the neurological outcome after surgery as defined by Glasgow Coma Scale (GCS), modified Rankin scale (mRS), Neurologic Assessment in Neuro-Oncology (NANO) scale, National Institute Health of Strokes Scale (NIHSS), and Karnofsky Performance Score (KPS), and need for unscheduled brain scan.

RESULTS

Of 442 patients (mean [SD] age, 56.1 [16.2]; 235 [55.7%] women and 187 [44.3%] men) who underwent a neurosurgical procedure, 169 (40.0%) patients had an intraoperative adverse event (iAE) classified as ClassIntra I or higher. The NIHSS score at admission (OR, 1.29; 95% CI, 1.03-1.63, female gender (OR, 0.44; 95% CI, 0.23-0.84), extracranial procedures (OR, 0.17; 95% CI, 0.08-0.61), and emergency cases (OR, 2.84; 95% CI, 1.53-3.78) were independent risk factors for a more severe iAE. A ClassIntra ≥ II was associated with increased odds of postoperative complications classified as Clavien-Dindo (p < 0.01), neurological deterioration at discharge (p < 0.01), prolonged hospital (p < 0.01), and ICU stay (p < 0.01). For elective craniotomies, severity of ClassIntra was associated with the CCI (p < 0.01) and need for unscheduled CT or MRI scan (p < 0.01). The proportion of a ClassIntra ≥ II was significantly higher for emergent craniotomies (56.2%) and associated with in-hospital mortality, and an unfavorable neurological outcome (p < 0.01).

CONCLUSION

Findings of this study suggest that the ClassIntra is sensitive for assessing intraoperative adverse events and sufficient to identify patients with a higher risk for developing postoperative complications after a neurosurgical procedure.

Machine learning algorithms for predicting outcomes of traumatic brain injury: A systematic review and meta-analysis

BACKGROUND

Traumatic brain injury (TBI) is a leading cause of death and disability worldwide. The use of machine learning (ML) has emerged as a key advancement in TBI management. This study aimed to identify ML models with demonstrated effectiveness in predicting TBI outcomes.

METHODS

We conducted a systematic review in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analysis statement. In total, 15 articles were identified using the search strategy. Patient demographics, clinical status, ML outcome variables, and predictive characteristics were extracted. A small meta-analysis of mortality prediction was performed, and a meta-analysis of diagnostic accuracy was conducted for ML algorithms used across multiple studies.

RESULTS

ML algorithms including support vector machine (SVM), artificial neural networks (ANN), random forest, and Naïve Bayes were compared to logistic regression (LR). Thirteen studies found significant improvement in prognostic capability using ML versus LR. The accuracy of the above algorithms was consistently over 80% when predicting mortality and unfavorable outcome measured by Glasgow Outcome Scale. Receiver operating characteristic curves analyzing the sensitivity of ANN, SVM, decision tree, and LR demonstrated consistent findings across studies. Lower admission Glasgow Coma Scale (GCS), older age, elevated serum acid, and abnormal glucose were associated with increased adverse outcomes and had the most significant impact on ML algorithms.

CONCLUSION

ML algorithms were stronger than traditional regression models in predicting adverse outcomes. Admission GCS, age, and serum metabolites all have strong predictive power when used with ML and should be considered important components of TBI risk stratification.

Artificial Intelligence in Neurosurgery: A State-of-the-Art Review from Past to Future

In recent years, there has been a significant surge in discussions surrounding artificial intelligence (AI), along with a corresponding increase in its practical applications in various facets of everyday life, including the medical industry. Notably, even in the highly specialized realm of neurosurgery, AI has been utilized for differential diagnosis, pre-operative evaluation, and improving surgical precision. Many of these applications have begun to mitigate risks of intraoperative and postoperative complications and post-operative care. This article aims to present an overview of the principal published papers on the significant themes of tumor, spine, epilepsy, and vascular issues, wherein AI has been applied to assess its potential applications within neurosurgery. The method involved identifying high-cited seminal papers using PubMed and Google Scholar, conducting a comprehensive review of various study types, and summarizing machine learning applications to enhance understanding among clinicians for future utilization. Recent studies demonstrate that machine learning (ML) holds significant potential in neuro-oncological care, spine surgery, epilepsy management, and other neurosurgical applications. ML techniques have proven effective in tumor identification, surgical outcomes prediction, seizure outcome prediction, aneurysm prediction, and more, highlighting its broad impact and potential in improving patient management and outcomes in neurosurgery. This review will encompass the current state of research, as well as predictions for the future of AI within neurosurgery.

The Application of Deep Learning to Electroencephalograms, Magnetic Resonance Imaging, and Implants for the Detection of Epileptic Seizures: A Narrative Review

Epilepsy is a neurological disorder characterized by recurrent seizures affecting millions worldwide. Medically intractable seizures in epilepsy patients are not only detrimental to the quality of life but also pose a significant threat to their safety. Outcomes of epilepsy therapy can be improved by early detection and intervention during the interictal window period. Electroencephalography is the primary diagnostic tool for epilepsy, but accurate interpretation of seizure activity is challenging and highly time-consuming. Machine learning (ML) and deep learning (DL) algorithms enable us to analyze complex EEG data, which can not only help us diagnose but also locate epileptogenic zones and predict medical and surgical treatment outcomes. DL models such as convolutional neural networks (CNNs), inspired by visual processing, can be used to classify EEG activity. By applying preprocessing techniques, signal quality can be enhanced by denoising and artifact removal. DL can also be incorporated into the analysis of magnetic resonance imaging (MRI) data, which can help in the localization of epileptogenic zones in the brain. Proper detection of these zones can help in good neurosurgical outcomes. Recent advancements in DL have facilitated the implementation of these systems in neural implants and wearable devices, allowing for real-time seizure detection. This has the potential to transform the management of drug-refractory epilepsy. This review explores the application of ML and DL techniques to Electroencephalograms (EEGs), MRI, and wearable devices for epileptic seizure detection. This review briefly explains the fundamentals of both artificial intelligence (AI) and DL, highlighting these systems' potential advantages and undeniable limitations.

Development and Assessment of Machine Learning Models for Individualized Risk Assessment of Mastectomy Skin Flap Necrosis

OBJECTIVE

To develop, validate, and evaluate ML algorithms for predicting MSFN.

BACKGROUND

MSFN is a devastating complication that causes significant distress to patients and physicians by prolonging recovery time, compromising surgical outcomes, and delaying adjuvant therapy.

METHODS

We conducted comprehensive review of all consecutive patients who underwent mastectomy and immediate implant-based reconstruction from January 2018 to December 2019. Nine supervised ML algorithms were developed to predict MSFN. Patient data were partitioned into training (80%) and testing (20%) sets.

RESULTS

We identified 694 mastectomies with immediate implant-based reconstruction in 481 patients. The patients had a mean age of 50 ± 11.5 years, years, a mean body mass index of 26.7 ± 4.8 kg/m 2 , and a median follow-up time of 16.1 (range, 11.9-23.2) months. MSFN developed in 6% (n = 40) of patients. The random forest model demonstrated the best discriminatory performance (area under curve, 0.70), achieved a mean accuracy of 89% (95% confidence interval, 83-94), and identified 10 predictors of MSFN. Decision curve analysis demonstrated that ML models have a superior net benefit regardless of the probability threshold. Higher body mass index, older age, hypertension, subpectoral device placement, nipple-sparing mastectomy, axillary nodal dissection, and no acellular dermal matrix use were all independently associated with a higher risk of MSFN.

CONCLUSIONS

ML algorithms trained on readily available perioperative clinical data can accurately predict the occurrence of MSFN and aid in individualized patient counseling, preoperative optimization, and surgical planning to reduce the risk of this devastating complication.

Developing Predictive Models to Anticipate Shunt Complications in 33,248 Pediatric Patients with Shunted Hydrocephalus Utilizing Machine Learning

INTRODUCTION

Hydrocephalus is a common pediatric neurosurgical pathology, typically treated with a ventricular shunt, yet approximately 30% of patients experience shunt failure within the first year after surgery. As a result, the objective of the present study was to validate a predictive model of pediatric shunt complications with data retrieved from the Healthcare Cost and Utilization Project (HCUP) National Readmissions Database (NRD).

METHODS

The HCUP NRD was queried from 2016 to 2017 for pediatric patients undergoing shunt placement using ICD-10 codes. Comorbidities present upon initial admission resulting in shunt placement, Johns Hopkins Adjusted Clinical Groups (JHACG) frailty-defining criteria, and Major Diagnostic Category (MDC) at admission classifications were obtained. The database was divided into training (n = 19,948), validation (n = 6,650), and testing (n = 6,650) datasets. Multivariable analysis was performed to identify significant predictors of shunt complications which were used to develop logistic regression models. Post hoc receiver operating characteristic (ROC) curves were created.

RESULTS

A total of 33,248 pediatric patients aged 6.9 ± 5.7 years were included. Number of diagnoses during primary admission (OR: 1.05, 95% CI: 1.04-1.07) and initial neurological admission diagnoses (OR: 3.83, 95% CI: 3.33-4.42) positively correlated with shunt complications. Female sex (OR: 0.87, 95% CI: 0.76-0.99) and elective admissions (OR: 0.62, 95% CI: 0.53-0.72) negatively correlated with shunt complications. ROC curve for the regression model utilizing all significant predictors of readmission demonstrated area under the curve of 0.733, suggesting these factors are possible predictors of shunt complications in pediatric hydrocephalus.

CONCLUSION

Efficacious and safe treatment of pediatric hydrocephalus is of paramount importance. Our machine learning algorithm delineated possible variables predictive of shunt complications with good predictive value.

Development of gene model combined with machine learning technology to predict for advanced atherosclerotic plaques

BACKGROUND

Atherosclerosis, as a major cause of stroke, is responsible for a quarter of deaths worldwide. In particular, rupture of late-stage plaques in large vessels such as the carotid artery can lead to serious cardiovascular disease. The aim of our study was to establish a genetic model combined with machining leaning techniques to screen out gene signatures and predict for advanced atherosclerosis plaques.

METHODS

The microarray dataset GSE28829 and GSE43292 which were publicly obtained from the Gene Expression Omnibus database were utilized to screen for potential predictive genes. Differentially expressed genes (DEGs) were identified by using the "limma" R package. Gene Ontology (GO) and Kyoto Encyclopedia of Genes Genomes (KEGG) analyses of these DEGs were performed by Metascape. Later, Random Forest (RF) algorithm was applied to further screen out top-30 genes which contribute the most. The expression data of top 30-DEGs were converted into a "Gene Score". Finally, we developed a model based on artificial neural network (ANN) to predict advanced atherosclerotic plaques. The model later was validated in an independent test dataset GSE104140.

RESULTS

A total of 176 DEGs were identified in the training datasets. GO and KEGG enrichment analysis revealed that these genes were enriched in leukocyte-mediated immune response, cytokine- cytokine interactions, and immunoinflammatory signaling. Further, top-30 genes (including 25 upregulated and 5 downregulated DEGs) were screened as predictors by RF algorithm. The predictive model was developed with a significantly predictive value (AUC = 0.913) in the training datasets, and was validated with an independent dataset GSE104140 (AUC = 0.827).

CONCLUSION

In present study, our prediction model was established and showed satisfactory predictive power in both training and test datasets. In addition, this is the first study adopted bioinformatics methods combined with machine learning techniques (RF and ANN) to explore and predict for the advanced atherosclerotic plaques. However, further investigations were needed to verify the screened DEGs and predictive effectiveness of this model.

Estimating postoperative mortality in colorectal surgery- a systematic review of risk prediction models

PURPOSE

Risk prediction models are frequently used to support decision-making in colorectal surgery but can be inaccurate. Machine learning (ML) is becoming increasingly popular, and its application may increase predictive accuracy. We compared conventional risk prediction models for postoperative mortality (based on regression analysis) with ML models to determine the benefit of the latter approach.

METHODS

The study was registered in PROSPERO(CRD42022364753). Following the PRISMA guidelines, a systematic search of three databases (MEDLINE, EMBASE, WoS) was conducted (from 1/1/2000 to 29/09/2022). Studies were included if they reported the development of a risk model to estimate short-term postoperative mortality for patients undergoing colorectal surgery. Discrimination and calibration performance metrics were compared. Studies were evaluated against CHARMS and TRIPOD criteria.

RESULTS

3,052 articles were screened, and 45 studies were included. The total sample size was 1,356,058 patients. Six studies used ML techniques for model development. Most studies (n = 42) reported the area under the receiver operating characteristic curve (AUROC) as a measure of discrimination. There was no significant difference in the mean AUROC values between regression models (0.833 s.d. ± 0.52) and ML (0.846 s.d. ± 0.55), p = 0.539. Calibration statistics, which measure the agreement between predicted estimates and observed outcomes, were less consistent. Risk of bias assessment found most concerns in the data handling and analysis domains of eligible studies.

CONCLUSIONS

Our study showed comparable predictive performance between regression and ML methods in colorectal surgery. Integration of ML in colorectal risk prediction is promising but further refinement of the models is required to support routine clinical adoption.

Machine Learning and Artificial Intelligence in Surgical Research

Machine learning, a subtype of artificial intelligence, is an emerging field of surgical research dedicated to predictive modeling. From its inception, machine learning has been of interest in medical and surgical research. Built on traditional research metrics for optimal success, avenues of research include diagnostics, prognosis, operative timing, and surgical education, in a variety of surgical subspecialties. Machine learning represents an exciting and developing future in the world of surgical research that will not only allow for more personalized and comprehensive medical care.

The Use of Machine Learning for Predicting Complications of Free-Flap Head and Neck Reconstruction

BACKGROUND

Machine learning has been increasingly used for surgical outcome prediction, yet applications in head and neck reconstruction are not well-described. In this study, we developed and evaluated the performance of ML algorithms in predicting postoperative complications in head and neck free-flap reconstruction.

METHODS

We conducted a comprehensive review of patients who underwent microvascular head and neck reconstruction between January 2005 and December 2018. Data were used to develop and evaluate nine supervised ML algorithms in predicting overall complications, major recipient-site complication, and total flap loss.

RESULTS

We identified 4000 patients who met inclusion criteria. Overall, 33.7% of patients experienced a complication, 26.5% experienced a major recipient-site complication, and 1.7% suffered total flap loss. The k-nearest neighbors algorithm demonstrated the best overall performance for predicting any complication (AUROC = 0.61, sensitivity = 0.60). Regularized regression had the best performance for predicting major recipient-site complications (AUROC = 0.68, sensitivity = 0.66), and decision trees were the best predictors of total flap loss (AUROC = 0.66, sensitivity = 0.50).

CONCLUSIONS

ML accurately identified patients at risk of experiencing postsurgical complications, including total flap loss. Predictions from ML models may provide insight in the perioperative setting and facilitate shared decision making.

New insights for predicting surgery outcome in patients with temporal lobe epilepsy. A systematic review

Resective surgery is the treatment of choice for one-third of adult patients with focal, drug-resistant epilepsy. This procedure is associated with substantial clinical and cognitive risks. In clinical practice, there is no validated model for epilepsy surgery outcome prediction (ESOP). Meta-analyses on ESOP studies assessing prognostic factors report discrepancies in terms of study design. Our review aims to systematically investigate methodological and analytical aspects of studies predicting clinical and cognitive outcomes after temporal lobe epilepsy surgery. A systematic review of ESOP studies published between 2000 and 2022 from three databases (MEDLINE, Web of Science, and PsycINFO) was completed by following the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines. It yielded 4867 articles. Among them, 21 corresponded to our inclusion criteria and were therefore retained in the final review. The risk of bias was assessed using A Tool to Assess Risk of Bias and Applicability of Prediction Model Studies (PROBAST). Data extracted from the 21 studies were analyzed using narrative synthesis and descriptive statistics. Our findings show an increase in the use of multimodal datasets and machine learning analyses in recent ESOP studies, although regression remained the most frequently used approach. We also identified a more frequent use of network notions in recent ESOP studies. Nevertheless, several methodological issues were noted, such as small sample sizes, lack of information on the follow-up period, variability in seizure outcome, and the definition of neuropsychological postoperative change. Of 21 studies, only one provided a clinical tool to anticipate the cognitive outcome after epilepsy surgery. We conclude that methodological issues should be overcome before we move towards more complete models to better predict clinical and cognitive outcomes after epilepsy surgery. Recommendations for future studies to harness the possibilities of multimodal datasets and data fusion, are provided. A stronger bridge between fundamental and clinical research may result in developing accessible clinical tools.

Prediction of Mortality among Patients with Isolated Traumatic Brain Injury Using Machine Learning Models in Asian Countries: An International Multi-Center Cohort Study

Traumatic brain injury (TBI) is a significant healthcare concern in several countries, accounting for a major burden of morbidity, mortality, disability, and socioeconomic losses. Although conventional prognostic models for patients with TBI have been validated, their performance has been limited. Therefore, we aimed to construct machine learning (ML) models to predict the clinical outcomes in adult patients with isolated TBI in Asian countries. The Pan-Asian Trauma Outcome Study registry was used in this study, and the data were prospectively collected from January 1, 2015, to December 31, 2020. Among a total of 6540 patients (≥ 15 years) with isolated moderate and severe TBI, 3276 (50.1%) patients were randomly included with stratification by outcomes and subgrouping variables for model evaluation, and 3264 (49.9%) patients were included for model training and validation. Logistic regression was considered as a baseline, and ML models were constructed and evaluated using the area under the precision-recall curve (AUPRC) as the primary outcome metric, area under the receiver operating characteristic curve (AUROC), and precision at fixed levels of recall. The contribution of the variables to the model prediction was measured using the SHapley Additive exPlanations (SHAP) method. The ML models outperformed logistic regression in predicting the in-hospital mortality. Among the tested models, the gradient-boosted decision tree showed the best performance (AUPRC, 0.746 [0.700-0.789]; AUROC, 0.940 [0.929-0.952]). The most powerful contributors to model prediction were the Glasgow Coma Scale, O₂ saturation, transfusion, systolic and diastolic blood pressure, body temperature, and age. Our study suggests that ML techniques might perform better than conventional multi-variate models in predicting the outcomes among adult patients with isolated moderate and severe TBI.

Machine learning prediction of malignant middle cerebral artery infarction after mechanical thrombectomy for anterior circulation large vessel occlusion

OBJECTIVE

Prediction of malignant middle cerebral artery infarction (MMI) could identify patients for early intervention. We trained and internally validated a ML model that predicts MMI following mechanical thrombectomy (MT) for ACLVO.

METHODS

All patients who underwent MT for ACLVO between 2015 - 2021 at a single institution were reviewed. Data was divided into 80% training and 20% test sets. 10 models were evaluated on the training set. The top 3 models underwent hyperparameter tuning using grid search with nested 5-fold CV to optimize the area under the receiver operating curve (AUROC). Tuned models were evaluated on the test set and compared to logistic regression.

RESULTS

A total of 381 patients met the inclusion criteria. There were 50 (13.1%) patients who developed MMI. Out of the 10 ML models screened on the training set, the top 3 performing were neural network (median AUROC 0.78, IQR 0.72 - 0.83), support vector machine ([SVM] median AUROC 0.77, IQR 0.72 - 0.83), and random forest (median AUROC 0.75, IQR 0.68 - 0.81). On the test set, random forest (median AUROC 0.78, IQR 0.73 - 0.83) and neural network (median AUROC 0.78, IQR 0.73 - 0.83) were the top performing models, followed by SVM (median AUROC 0.77, IQR 0.70 - 0.83). These scores were significantly better than those for logistic regression (AUROC 0.72, IQR 0.66 - 0.78), individual risk factors, and the Malignant Brain Edema score (p < 0.001 for all).

CONCLUSION

ML models predicted MMI with good discriminative ability. They outperformed standard statistical techniques and individual risk factors.

Machine Learning Models Using Routinely Collected Clinical Data Offer Robust and Interpretable Predictions of 90-Day Unplanned Acute Care Use for Cancer Immunotherapy Patients

PURPOSE

Clinical management of patients receiving immune checkpoint inhibitors (ICIs) could be informed using accurate predictive tools to identify patients at risk of short-term acute care utilization (ACU). We used routinely collected data to develop and assess machine learning (ML) algorithms to predict unplanned ACU within 90 days of ICI treatment initiation.

METHODS

We used aggregated electronic health record data from 7,960 patients receiving ICI treatments to train and assess eight ML algorithms. We developed the models using pre-SARS-COV-19 COVID-19 data generated between January 2016 and February 2020. We validated our algorithms using data collected between March 2020 and June 2022 (peri-COVID-19 sample). We assessed performance using area under the receiver operating characteristic curves (AUROC), sensitivity, specificity, and calibration plots. We derived intuitive explanations of predictions using variable importance and Shapley additive explanation analyses. We assessed the marginal performance of ML models compared with that of univariate and multivariate logistic regression (LR) models.

RESULTS

Most algorithms significantly outperformed the univariate and multivariate LR models. The extreme gradient boosting trees (XGBT) algorithm demonstrated the best overall performance (AUROC, 0.70; sensitivity, 0.53; specificity, 0.74) on the peri-COVID-19 sample. The algorithm performance was stable across both pre- and peri-COVID-19 samples, as well as ICI regimen and cancer groups. Type of ICI agents, oxygen saturation, diastolic blood pressure, albumin level, platelet count, immature granulocytes, absolute monocyte, chloride level, red cell distribution width, and alcohol intake were the top 10 key predictors used by the XGBT algorithm.

CONCLUSION

Machine learning algorithms trained using routinely collected data outperformed traditional statistical models when predicting 90-day ACU. The XGBT algorithm has the potential to identify high-ACU risk patients and enable preventive interventions to avoid ACU.

Artificial Intelligence in Neurosurgery: A Bibliometric Analysis

BACKGROUND

Artificial intelligence (AI) has the potential to augment clinicians' diagnostic and decision-making capabilities. It is well suited to identify patterns and correlations within data sets and may be applied to identify elements of importance in complex and data-laden areas such as patient selection, diagnostics, treatment, and outcome prediction. The development of modern neurosurgery has been dependent on major technological advances. In line with this, a growing interest is seen in the use of AI to assist in neurosurgical research and enhance neurosurgical practices.

METHODS

A bibliometric analysis of the 50 most-cited articles alluding to the use of AI in neurosurgery, from inception until July of 2022, was undertaken using the Web of Science database. Statistical analyses were performed on R.

RESULTS

The citation count ranged from 29 to 159 (mean: 51.9, standard deviation: 24.8), and the top-cited article was a 2018 systematic review published in World Neurosurgery. Most articles were published after 2015 (85%). The United States was the largest contributing country on the list with 22 articles. Four first and last authors, each, had 2 or more publications. Female first and last authorship was attributed to 18% and 0% of the articles, respectively.

CONCLUSIONS

This review highlights the most-impactful articles pertaining to AI in the field of neurosurgery. Although female authors were significantly underrepresented on the list, their work was at least as impactful as their male peers. Finally, the striking dominance of articles originating from the developed world raises concerns as to the future of AI in attending to the global health crisis.