RESUMO
To examine the safety and efficacy of chemoablation using UGN-101 in patients with upper tract urothelial cancer (UTUC). We conducted a systematic search through 7 databases/registries to identify key observational and experimental studies reporting either the efficacy or safety of UGN-101 in UTUC patients regardless of the risk or grade of the disease. The outcomes included efficacy (complete/partial/no response, survival, death, recurrence, or progression) and safety endpoints. All meta-analyses were conducted through STATA. The prevalence rate and its 95% CI were pooled across studies. A subgroup meta-analysis was conducted on follow-up. The quality was assessed using the Newcastle Ottawa Scale. Twenty studies (1051 patients) were analyzed. Complete response was reported in 49% (39%-60%) of cases, and 5% (0%- 15%) had disease progression. Treatment cessation was reported in 13% (3%-27%) of patients. Four percent of cases needed radical nephroureterectomy. Recurrence and death occurred in 14% (7%-23%) and 6% (2%-10%) of patients. Complications occurred in 63% (39%-85%), the majority of which were of grades I, II, and III. Ureteral stenosis was the most common complication accounting for 35% of cases. Chemoablationrelated complications occurred more than procedure-related ones. Based on available evidence, the intracavitary instillation of UGN-101 gel provides an alternative therapeutic option for upper tract urothelial cancer. Chemoablation provides good clinical outcomes in terms of complete response, disease progression and recurrence, and the need to undergo nephroureterectomy. Complications were encountered in more than half the population; however, most of them were of low grades.
RESUMO
OBJECTIVE: Despite the proven efficacy of surgical intervention for achieving seizure freedom and improved quality of life for many epilepsy patients, this treatment remains underutilized. In this study, the authors assessed sociodemographic trends in epilepsy surgery in the National Inpatient Sample (NIS) and the Kids' Inpatient Database (KID) and sought to determine whether disparities in surgical intervention for epilepsy may be attributed to insurance and comorbidity status. METHODS: This cross-sectional study utilized data from the NIS database and KID from the Healthcare Cost and Utilization Project between the years 2012 and 2018. Outcomes of interest were rates of neurosurgical intervention, including resection, neuromodulation, or laser ablation. The authors utilized logit regression models to test the association between rates of neurosurgical intervention and the variables of interest and calculated the adjusted mean proportion of patients who received surgery using marginal effects. RESULTS: Of 336,015 admissions with intractable epilepsy in the NIS, 6.1% were patients who underwent neurosurgical treatment. Of 39,655 admissions from KID, 5.0% received surgical treatment. Private insurance was associated with a greater odds of surgical intervention compared with Medicaid (NIS: OR 1.63, KID: OR 1.62; p < 0.001). Patients assigned White race had an increased odds ratio of undergoing surgery when compared with those assigned Black race, adjusted for comorbidity burden (NIS: OR 1.59, p < 0.001; KID: OR 1.44, p = 0.027). Patients with an Elixhauser Comorbidity Index score of 0 or 1 were associated with a lower likelihood of surgery when compared to their higher scoring counterparts who had 4 or more comorbidities (NIS: OR 0.74, KID: OR 0.62; both p < 0.001). CONCLUSIONS: This study demonstrates that marginalized patients and those with Medicaid had decreased odds of neurosurgical intervention for epilepsy. Results of this research support the need for increased attention toward epilepsy patients from marginalized groups. Further investigation into the root cause of socioeconomic inequities in epilepsy surgery is necessary.
RESUMO
The responsive neurostimulator system has become increasingly popular in the surgical management of refractory epilepsy, with targeting of various thalamic nuclei showing promising results in select patients. A 42-year-old female presented for evaluation of refractory epilepsy consisting of generalized tonic-clonic and focal seizures with preserved awareness. Phase I and II monitoring suggested multifocal bilateral epilepsy with bilateral frontal onset, and the patient underwent robot-guided bilateral centromedian thalamic placement of the RNS System. In this operative video, the authors share their institutional experience and protocol utilizing the ExcelsiusGPS robot in the placement of the RNS System in the thalamus. The video can be found here: https://stream.cadmore.media/r10.3171/2024.4.FOCVID243.
RESUMO
Neurosurgery for intractable psychiatric conditions has seen a resurgence with the increasing use of deep brain stimulation (DBS). Although DBS promises reversible neuromodulation and has become more popular than older lesioning methods, lesioning might still be preferred in specific cases. Here, we review the evidence for DBS and lesions in the treatment of intractable neuropsychiatric conditions and consider the factors that favour the continued use of lesioning procedures in appropriately selected cases. Broadly, systemic factors including comparative effectiveness, cost, and ethical arguments support an ongoing role for lesioning. Such a role is also supported by practical considerations including patient experiences of this type of therapy, the relative intensity of follow-up care, access to sparse or specialised follow-up care, and relative infection risk. Overall, we argue that neurosurgical lesion procedures remain an important alternative to DBS and their continued availability is necessary to fulfil the imperatives of mental health parity and enhance access to effective mental health treatments. Nonetheless, the efficacy of DBS and recent advances in closed-loop stimulation and remote programming might provide solutions to some of the challenges associated with wider use of electrical neuromodulation. Concerns about the scarcity of high-level evidence for the efficacy of lesioning procedures as well as the potential irreversible adverse effects of lesioning remain to be addressed.
RESUMO
Neurophysiology and neuromodulation strive to understand the neural basis of behavior through a one-to-one correspondence between a particular brain and its behavioral output. Within this framework, studies with few subjects but sufficient sample sizes can be both rigorous and impactful.
Assuntos
Encéfalo , Neurofisiologia , Humanos , Encéfalo/fisiologia , Neurofisiologia/métodos , Tamanho da Amostra , Animais , Neurotransmissores/metabolismo , Comportamento/fisiologiaRESUMO
Despite the importance of informed consent in healthcare, the readability and specificity of consent forms often impede patients' comprehension. This study investigates the use of GPT-4 to simplify surgical consent forms and introduces an AI-human expert collaborative approach to validate content appropriateness. Consent forms from multiple institutions were assessed for readability and simplified using GPT-4, with pre- and post-simplification readability metrics compared using nonparametric tests. Independent reviews by medical authors and a malpractice defense attorney were conducted. Finally, GPT-4's potential for generating de novo procedure-specific consent forms was assessed, with forms evaluated using a validated 8-item rubric and expert subspecialty surgeon review. Analysis of 15 academic medical centers' consent forms revealed significant reductions in average reading time, word rarity, and passive sentence frequency (all P < 0.05) following GPT-4-faciliated simplification. Readability improved from an average college freshman to an 8th-grade level (P = 0.004), matching the average American's reading level. Medical and legal sufficiency consistency was confirmed. GPT-4 generated procedure-specific consent forms for five varied surgical procedures at an average 6th-grade reading level. These forms received perfect scores on a standardized consent form rubric and withstood scrutiny upon expert subspeciality surgeon review. This study demonstrates the first AI-human expert collaboration to enhance surgical consent forms, significantly improving readability without sacrificing clinical detail. Our framework could be extended to other patient communication materials, emphasizing clear communication and mitigating disparities related to health literacy barriers.
RESUMO
The role of neoadjuvant chemotherapy (NAC) in upper tract urothelial cancer (UTUC) is not yet confirmed. Therefore, we conducted this review to pool the available evidence in this regard. We analyzed 14117 UTUC patients reported in 21 studies after searching 5 databases. The NAC was administered in 1983 patients and the remaining 12134 controls underwent radical nephroureterectomy (RNU) alone. Efficacy endpoints included pathological, functional, and survival outcomes. Safety was determined by overall and grade 3-4 complications. For dichotomous outcomes, the log odds ratio (logOR) was pooled, and for continuous variables, the crude mean difference was calculated along with its 95% CI. The NAC was associated with 10% complete pathological response (CPR), 42% pathological downstaging, 31% post-NAC advanced disease (pT3-4), 6% positive surgical margin, 18% lymph node metastasis (pN+), 24% lymphovascular invasion, and 29% mortality and recurrence at 5 years. Compared to controls, NAC resulted in increased risk of CPR [logOR=1.67; 95% CI, 0.11-3.23] and downstaging [logOR=1.30; 95% CI, 0.41-2.18] and reduced risk of advanced disease [logOR=-0.81; 95% CI, -1.51--0.11]. Renal function did not improve from baseline; however, it increased significantly after RNU. The NAC was associated with good survival/low mortality in the short term, with a sustained increase over time. Overall and grade 3-4 complications occurred in 25% and 7% of patients, respectively. Our findings support the potential benefits of NAC in enhancing pathological outcomes and possibly improving survival in UTUC patients undergoing RNU. The variability in response and associated complications underscore the importance of careful patient selection and tailored treatment approaches.
RESUMO
OBJECTIVE: Delirium is an acute cognitive disorder associated with multiple electroencephalographic (EEG) abnormalities in non-neurological patients, though specific EEG characteristics in patients with stroke remain unclear. We aimed to compare the prevalence of EEG abnormalities in stroke patients during delirium episodes with periods that did not correspond to delirium. METHODS: We retrospectively analyzed clinical EEG reports for stroke patients who received daily delirium assessments as part of a prospective study. We compared the prevalence of EEG features corresponding to patient-days with vs. without delirium, including focal and generalized slowing, and focal and generalized epileptiform abnormalities (EAs). RESULTS: Among 58 patients who received EEGs, there were 192 days of both EEG and delirium monitoring (88% [n = 169] corresponding to delirium). Generalized slowing was significantly more prevalent on days with vs. without delirium (96% vs. 57%, p = 0.03), as were bilateral or generalized EAs (38% vs. 13%, p = 0.03). In contrast, focal slowing (53% vs. 74%, p = 0.11) and focal EAs were less prevalent on days with delirium (38% vs. 48%, p = 0.37), though these differences were not statistically significant. CONCLUSIONS: We found a higher prevalence of generalized but not focal EEG abnormalities in stroke patients with delirium. SIGNIFICANCE: These findings may reinforce the diffuse nature of delirium-associated encephalopathy, even in patients with discrete structural lesions.
Assuntos
Delírio , Eletroencefalografia , Acidente Vascular Cerebral , Humanos , Delírio/epidemiologia , Delírio/fisiopatologia , Delírio/diagnóstico , Masculino , Eletroencefalografia/métodos , Feminino , Idoso , Acidente Vascular Cerebral/fisiopatologia , Acidente Vascular Cerebral/complicações , Acidente Vascular Cerebral/epidemiologia , Prevalência , Pessoa de Meia-Idade , Idoso de 80 Anos ou mais , Estudos Retrospectivos , Estudos ProspectivosRESUMO
Understanding the role of the pulvinar nucleus may be critical for guiding circuit-targeted neurosurgical intervention in some patients. In this report, a 33-year-old female presented with focal onset occipital epilepsy with secondary generalization and with a previously radiated arteriovenous malformation within the right primary visual cortex. Phase II monitoring demonstrated the pulvinar nucleus was not involved in subclinical seizures restricted to the primary visual cortex, but it did become involved in clinical events with more extensive seizure spread into higher visual cortical regions. She underwent responsive neurostimulation (RNS) with implantation of leads within the primary visual cortex. This case demonstrates the late propagation of epileptic activity from the visual cortex to the pulvinar nucleus and illustrates the pulvinar nucleus' connections with higher-order visual areas.
RESUMO
BACKGROUND: Medical knowledge during the medieval ages flourished under the influence of great scholars of the Islamic Golden age such as Ibn Sina (Latinized as Avicenna), Abu Bakr al-Razi (Rhazes), and Abu al-Qasim Khalaf ibn al-Abbas al-Zahrawi, known as Albucasis. Much has been written on al-Zahrawi's innovation in various disciplines of medicine and surgery. In this article, we focus for on the contributions of al-Zahrawi toward the treatment of neurological disorders in the surgical chapters of his medical encyclopedia, Kitab al-Tasrif (The Method of Medicine). METHODS: Excerpts from a modern copy of volume 30 of al-Zahrawi's Kitab al-Tasrif were reviewed and translated by the primary author from Arabic to English, to further provide specific details regarding his neurosurgical knowledge. In addition, a literature search was performed using PubMed and Google Scholar to review prior reports on al-Zahrawi's neurosurgical instructions. RESULTS: In addition to what is described in the literature of al-Zahrawi's teachings in cranial and spine surgery, we provide insight into his diagnosis and management of cranial and spinal trauma, the devices he used, and prognostication of various traumatic injuries. CONCLUSIONS: Al-Zahrawi was a renowned physician during the Islamic Golden age who made significant contributions to the diagnosis and treatment of neurological conditions, particularly cranial and spinal cord injuries. He developed innovative surgical techniques for trephination and spinal traction, which are still used in modern neurosurgery. His insights make him worthy of recognition as an important figure in the history of neurological surgery.
Assuntos
Medicina Arábica , Doenças do Sistema Nervoso , Neurocirurgia , Traumatismos da Medula Espinal , Traumatismos da Coluna Vertebral , Humanos , Masculino , Doenças do Sistema Nervoso/cirurgia , Neurocirurgia/história , Procedimentos Neurocirúrgicos , Medicina Arábica/históriaRESUMO
Deep brain stimulation (DBS) and dopaminergic therapy (DA) are common interventions for Parkinson's disease (PD). Both treatments typically improve patient outcomes, and both can have adverse side effects on decision making (e.g., impulsivity).1,2 Nevertheless, they are thought to act via different mechanisms within basal ganglia circuits.3 Here, we developed and formally evaluated their dissociable predictions within a single cost/benefit effort-based decision-making task. In the same patients, we manipulated DA medication status and subthalamic nucleus (STN) DBS status within and across sessions. Using a series of descriptive and computational modeling analyses of participant choices and their dynamics, we confirm a double dissociation: DA medication asymmetrically altered participants' sensitivities to benefits vs. effort costs of alternative choices (boosting the sensitivity to benefits while simultaneously lowering sensitivity to costs); whereas STN DBS lowered the decision threshold of such choices. To our knowledge, this is the first study to show, using a common modeling framework, a dissociation of DA and DBS within the same participants. As such, this work offers a comprehensive account for how different mechanisms impact decision making, and how impulsive behavior (present in DA-treated patients with PD and DBS patients) may emerge from separate physiological mechanisms.
Assuntos
Estimulação Encefálica Profunda , Doença de Parkinson , Núcleo Subtalâmico , Humanos , Dopamina/uso terapêutico , Núcleo Subtalâmico/fisiologia , Testes Neuropsicológicos , Doença de Parkinson/terapia , Tomada de Decisões/fisiologiaRESUMO
Responsive neurostimulation (RNS) is a valuable tool in the diagnosis and treatment of medication refractory epilepsy (MRE) and provides clinicians with better insights into patients' seizure patterns. In this case illustration, we present a patient with bilateral hippocampal RNS for presumed bilateral mesial temporal lobe epilepsy. The patient subsequently underwent a right sided LITT amygdalohippocampotomy based upon chronic RNS data revealing predominance of seizures from that side. Analyzing electrocorticography (ECOG) from the RNS system, we identified the frequency of high amplitude discharges recorded from the left hippocampal lead pre- and post- right LITT amygdalohippocampotomy. A reduction in contralateral interictal epileptiform activity was observed through RNS recordings over a two-year period, suggesting the potential dependency of the contralateral activity on the primary epileptogenic zone. These findings suggest that early targeted surgical resection or laser ablation by leveraging RNS data can potentially impede the progression of dependent epileptiform activity and may aid in preserving neurocognitive networks. RNS recordings are essential in shaping further management decisions for our patient with a presumed bitemporal epilepsy.
RESUMO
Importance: The progression of artificial intelligence (AI) text-to-image generators raises concerns of perpetuating societal biases, including profession-based stereotypes. Objective: To gauge the demographic accuracy of surgeon representation by 3 prominent AI text-to-image models compared to real-world attending surgeons and trainees. Design, Setting, and Participants: The study used a cross-sectional design, assessing the latest release of 3 leading publicly available AI text-to-image generators. Seven independent reviewers categorized AI-produced images. A total of 2400 images were analyzed, generated across 8 surgical specialties within each model. An additional 1200 images were evaluated based on geographic prompts for 3 countries. The study was conducted in May 2023. The 3 AI text-to-image generators were chosen due to their popularity at the time of this study. The measure of demographic characteristics was provided by the Association of American Medical Colleges subspecialty report, which references the American Medical Association master file for physician demographic characteristics across 50 states. Given changing demographic characteristics in trainees compared to attending surgeons, the decision was made to look into both groups separately. Race (non-White, defined as any race other than non-Hispanic White, and White) and gender (female and male) were assessed to evaluate known societal biases. Exposures: Images were generated using a prompt template, "a photo of the face of a [blank]", with the blank replaced by a surgical specialty. Geographic-based prompting was evaluated by specifying the most populous countries on 3 continents (the US, Nigeria, and China). Main Outcomes and Measures: The study compared representation of female and non-White surgeons in each model with real demographic data using χ2, Fisher exact, and proportion tests. Results: There was a significantly higher mean representation of female (35.8% vs 14.7%; P < .001) and non-White (37.4% vs 22.8%; P < .001) surgeons among trainees than attending surgeons. DALL-E 2 reflected attending surgeons' true demographic data for female surgeons (15.9% vs 14.7%; P = .39) and non-White surgeons (22.6% vs 22.8%; P = .92) but underestimated trainees' representation for both female (15.9% vs 35.8%; P < .001) and non-White (22.6% vs 37.4%; P < .001) surgeons. In contrast, Midjourney and Stable Diffusion had significantly lower representation of images of female (0% and 1.8%, respectively; P < .001) and non-White (0.5% and 0.6%, respectively; P < .001) surgeons than DALL-E 2 or true demographic data. Geographic-based prompting increased non-White surgeon representation but did not alter female representation for all models in prompts specifying Nigeria and China. Conclusion and Relevance: In this study, 2 leading publicly available text-to-image generators amplified societal biases, depicting over 98% surgeons as White and male. While 1 of the models depicted comparable demographic characteristics to real attending surgeons, all 3 models underestimated trainee representation. The study suggests the need for guardrails and robust feedback systems to minimize AI text-to-image generators magnifying stereotypes in professions such as surgery.
Assuntos
Especialidades Cirúrgicas , Cirurgiões , Estados Unidos , Humanos , Masculino , Feminino , Estudos Transversais , Inteligência Artificial , DemografiaRESUMO
Introduction: For drug resistant epilepsy patients who are either not candidates for resective surgery or have already failed resective surgery, neuromodulation is a promising option. Neuromodulatory approaches include responsive neurostimulation (RNS), deep brain stimulation (DBS), and vagal nerve stimulation (VNS). Thalamocortical circuits are involved in both generalized and focal onset seizures. This paper explores the use of RNS in the centromedian nucleus of the thalamus (CMN) and in the anterior thalamic nucleus (ANT) of patients with drug resistant epilepsy. Methods: This is a retrospective multicenter study from seven different epilepsy centers in the United States. Patients that had unilateral or bilateral thalamic RNS leads implanted in the CMN or ANT for at least 6 months were included. Primary objectives were to describe the implant location and determine changes in the frequency of disabling seizures at 6 months, 1 year, 2 years, and > 2 years. Secondary objectives included documenting seizure free periods, anti-seizure medication regimen changes, stimulation side effects, and serious adverse events. In addition, the global clinical impression scale was completed. Results: Twelve patients had at least one lead placed in the CMN, and 13 had at least one lead placed in the ANT. The median baseline seizure frequency was 15 per month. Overall, the median seizure reduction was 33% at 6 months, 55% at 1 year, 65% at 2 years, and 74% at >2 years. Seizure free intervals of at least 3 months occurred in nine patients. Most patients (60%, 15/25) did not have a change in anti-seizure medications post RNS placement. Two serious adverse events were recorded, one related to RNS implantation. Lastly, overall functioning seemed to improve with 88% showing improvement on the global clinical impression scale. Discussion: Meaningful seizure reduction was observed in patients who suffer from drug resistant epilepsy with unilateral or bilateral RNS in either the ANT or CMN of the thalamus. Most patients remained on their pre-operative anti-seizure medication regimen. The device was well tolerated with few side effects. There were rare serious adverse events. Most patients showed an improvement in global clinical impression scores.
RESUMO
BACKGROUND AND OBJECTIVES: Interest surrounding generative large language models (LLMs) has rapidly grown. Although ChatGPT (GPT-3.5), a general LLM, has shown near-passing performance on medical student board examinations, the performance of ChatGPT or its successor GPT-4 on specialized examinations and the factors affecting accuracy remain unclear. This study aims to assess the performance of ChatGPT and GPT-4 on a 500-question mock neurosurgical written board examination. METHODS: The Self-Assessment Neurosurgery Examinations (SANS) American Board of Neurological Surgery Self-Assessment Examination 1 was used to evaluate ChatGPT and GPT-4. Questions were in single best answer, multiple-choice format. χ 2 , Fisher exact, and univariable logistic regression tests were used to assess performance differences in relation to question characteristics. RESULTS: ChatGPT (GPT-3.5) and GPT-4 achieved scores of 73.4% (95% CI: 69.3%-77.2%) and 83.4% (95% CI: 79.8%-86.5%), respectively, relative to the user average of 72.8% (95% CI: 68.6%-76.6%). Both LLMs exceeded last year's passing threshold of 69%. Although scores between ChatGPT and question bank users were equivalent ( P = .963), GPT-4 outperformed both (both P < .001). GPT-4 answered every question answered correctly by ChatGPT and 37.6% (50/133) of remaining incorrect questions correctly. Among 12 question categories, GPT-4 significantly outperformed users in each but performed comparably with ChatGPT in 3 (functional, other general, and spine) and outperformed both users and ChatGPT for tumor questions. Increased word count (odds ratio = 0.89 of answering a question correctly per +10 words) and higher-order problem-solving (odds ratio = 0.40, P = .009) were associated with lower accuracy for ChatGPT, but not for GPT-4 (both P > .005). Multimodal input was not available at the time of this study; hence, on questions with image content, ChatGPT and GPT-4 answered 49.5% and 56.8% of questions correctly based on contextual context clues alone. CONCLUSION: LLMs achieved passing scores on a mock 500-question neurosurgical written board examination, with GPT-4 significantly outperforming ChatGPT.
Assuntos
Neurocirurgia , Humanos , Procedimentos Neurocirúrgicos , Razão de Chances , Autoavaliação (Psicologia) , Coluna VertebralRESUMO
OBJECTIVE: In patients with treatment-refractory temporal lobe epilepsy (TLE), a single stereotactic laser interstitial thermotherapy (LITT) procedure is sometimes insufficient to ablate epileptogenic tissue, particularly the medial structures often implicated in TLE. In patients with seizure recurrence after initial ablation, the extent to which a second ablation may achieve improved seizure outcomes is uncertain. The objective of this study was to investigate the feasibility and potential efficacy of repeat LITT amygdalohippocampotomy as a worthwhile strategy for intractable temporal lobe epilepsy by quantifying changes to targeted mesial temporal lobe structures and seizure outcomes. METHODS: Patients who underwent two LITT procedures for drug-resistant mesial TLE at our institution were included in the study. Lesion volumes for both procedures were calculated by comparing post-ablation intraoperative sequences to preoperative anatomy. Clinical outcomes after the initial procedure and repeat procedure were classified according to Engel scores. RESULTS: Five consecutive patients were included in this retrospective case series: 3 with right- and 2 with left-sided TLE. The median interval between LITT procedures was 294 days (range: 227-1918). After the first LITT, 3 patients experienced class III outcomes, 1 experienced a class IV, and 1 experienced a class IB outcome. All patients achieved increased seizure freedom after a second procedure, with class I outcomes (3 IA, 2 IB). CONCLUSIONS: Repeat LITT may be sufficient to achieve satisfactory seizure outcomes in some individuals who might otherwise be considered for more aggressive resection or palliative neuromodulation. A larger study to establish the potential value of repeat LITT amygdalohippocampotomy vs. other re-operation strategies for persistent, intractable temporal lobe epilepsy is worth pursuing.
Assuntos
Epilepsia Resistente a Medicamentos , Epilepsia do Lobo Temporal , Terapia a Laser , Humanos , Epilepsia do Lobo Temporal/cirurgia , Epilepsia do Lobo Temporal/patologia , Estudos Retrospectivos , Resultado do Tratamento , Terapia a Laser/métodos , Convulsões/cirurgia , Epilepsia Resistente a Medicamentos/cirurgia , Lasers , Imageamento por Ressonância MagnéticaRESUMO
BACKGROUND AND OBJECTIVES: General large language models (LLMs), such as ChatGPT (GPT-3.5), have demonstrated the capability to pass multiple-choice medical board examinations. However, comparative accuracy of different LLMs and LLM performance on assessments of predominantly higher-order management questions is poorly understood. We aimed to assess the performance of 3 LLMs (GPT-3.5, GPT-4, and Google Bard) on a question bank designed specifically for neurosurgery oral boards examination preparation. METHODS: The 149-question Self-Assessment Neurosurgery Examination Indications Examination was used to query LLM accuracy. Questions were inputted in a single best answer, multiple-choice format. χ 2 , Fisher exact, and univariable logistic regression tests assessed differences in performance by question characteristics. RESULTS: On a question bank with predominantly higher-order questions (85.2%), ChatGPT (GPT-3.5) and GPT-4 answered 62.4% (95% CI: 54.1%-70.1%) and 82.6% (95% CI: 75.2%-88.1%) of questions correctly, respectively. By contrast, Bard scored 44.2% (66/149, 95% CI: 36.2%-52.6%). GPT-3.5 and GPT-4 demonstrated significantly higher scores than Bard (both P < .01), and GPT-4 outperformed GPT-3.5 ( P = .023). Among 6 subspecialties, GPT-4 had significantly higher accuracy in the Spine category relative to GPT-3.5 and in 4 categories relative to Bard (all P < .01). Incorporation of higher-order problem solving was associated with lower question accuracy for GPT-3.5 (odds ratio [OR] = 0.80, P = .042) and Bard (OR = 0.76, P = .014), but not GPT-4 (OR = 0.86, P = .085). GPT-4's performance on imaging-related questions surpassed GPT-3.5's (68.6% vs 47.1%, P = .044) and was comparable with Bard's (68.6% vs 66.7%, P = 1.000). However, GPT-4 demonstrated significantly lower rates of "hallucination" on imaging-related questions than both GPT-3.5 (2.3% vs 57.1%, P < .001) and Bard (2.3% vs 27.3%, P = .002). Lack of question text description for questions predicted significantly higher odds of hallucination for GPT-3.5 (OR = 1.45, P = .012) and Bard (OR = 2.09, P < .001). CONCLUSION: On a question bank of predominantly higher-order management case scenarios for neurosurgery oral boards preparation, GPT-4 achieved a score of 82.6%, outperforming ChatGPT and Google Bard.