Ethical Incorporation of Artificial Intelligence into Neurosurgery: A Generative Pre-Trained Transformer Chatbot-Based, Human-Modified Approach.

INTRODUCTION: Artificial intelligence (AI) has become increasingly used in neurosurgery. Generative pre-trained transformers (GPTs) have been of particular interest. However, ethical concerns regarding the incorporation of AI into the field remain underexplored. We delineate key ethical considerations using a novel GPT-based, human-modified approach, synthesize the most common considerations, and present an ethical framework for the involvement of AI in neurosurgery. METHODS: GPT-4, ChatGPT, Bing Chat / Copilot, You, Perplexity.ai, and Google Bard were queried with the prompt "How can artificial intelligence be ethically incorporated into neurosurgery?". Then, a layered GPT-based thematic analysis was performed. The authors synthesized the results into considerations for the ethical incorporation of AI into neurosurgery. Separate Pareto analyses with 20% threshold and 10% threshold were conducted to determine salient themes. The authors refined these salient themes. RESULTS: Twelve key ethical considerations focusing on stakeholders, clinical implementation, and governance were identified. Refinement of the Pareto analysis of the top 20% most salient themes in the aggregated GPT outputs yielded ten key considerations. Additionally, from the top 10% most salient themes, five considerations were retrieved. An ethical framework for the use of AI in neurosurgery was developed. CONCLUSION: It is critical to address the ethical considerations associated with the use of AI in neurosurgery. The framework described in this manuscript may facilitate the integration of AI into neurosurgery, benefitting both patients and neurosurgeons alike. We urge neurosurgeons to use AI only for validated purposes and caution against automatic adoption of its outputs without neurosurgeon interpretation.

Influence of Preoperative Motor Score and Patient Comorbidities on Transcranial Motor-Evoked Potential Acquisition in Intracranial Surgery: A Retrospective Cohort Study.

BACKGROUND AND OBJECTIVES: Intraoperative neurophysiological monitoring plays a pivotal role in modern neurosurgery, aiding in real-time assessment of eloquent neural structures to mitigate iatrogenic neural injury. This study represents the largest retrospective series to date in monitoring corticospinal tract integrity during intracranial surgery with transcranial motor-evoked potentials (TCMEPs), focusing on the influence of demographic factors, comorbidities, and preoperative motor deficits on the reliability of intraoperative neurophysiological monitoring. While the impact of patient-specific factors affecting TCMEP monitoring in spine surgery is well-documented, similar insights for intracranial surgery are lacking. METHODS: A total of 420 craniotomy patients were retrospectively analyzed from electronic medical records from December 2017 to February 2023, excluding patients without preoperative Medical Research Council scores or medical histories. Using intrinsic hand muscles as a robust data set, 840 hand TCMEPs acquired during intracranial surgery were assessed. Demographic and clinical factors, including preoperative motor scores, were analyzed to identify associations with TCMEP acquisition and amplitude. Nonparametric statistics and multivariate regression analysis were employed. RESULTS: TCMEPs were successfully acquired in 734 (87.7%) patient hands, even in the presence of preoperative motor deficits in 13.9% of total patient hands. Preoperative motor scores did not predict the ability to acquire baseline TCMEPs (P = .6). Notably, older age (P < .001) and hypertension (P = .01) were independent predictors of lower TCMEP acquisition rates. Preoperative motor scores significantly influenced TCMEP amplitudes, with higher scores correlating with higher amplitudes (1771 [SD = 1550] eve vs 882 [SD = 856] µV, P < .0001). Older age (P < .001) and chronic kidney disease (P = .04) were also associated with reduced TCMEP amplitudes. CONCLUSION: Our investigation into TCMEPs during intracranial surgery demonstrated a notably high acquisition rate in hand muscles, irrespective of preoperative motor deficits. Preoperative motor scores reliably correlated with TCMEP amplitudes in a linear fashion while advanced age and renal disease emerged as independent predictors of lower TCMEP amplitudes.

Results from a first-in-human phase I safety trial to evaluate the use of a vascularized pericranial/temporoparietal fascial flap to line the resection cavity following resection of newly diagnosed glioblastoma.

PURPOSE: The efficacy of systemic therapies for glioblastoma (GBM) remains limited due to the constraints of systemic toxicity and blood-brain barrier (BBB) permeability. Temporoparietal fascial flaps (TPFFs) and vascularized peri cranial flaps (PCF) are not restricted by the blood-brain barrier (BBB), as they derive their vascular supply from branches of the external carotid artery. Transposition of a vascularized TPFF or PCF along a GBM resection cavity may bring autologous tissue not restricted by the BBB in close vicinity to the tumor bed microenvironment, permit ingrowth of vascular channels fed by the external circulation, and offer a mechanism of bypassing the BBB. In addition, circulating immune cells in the vascularized flap may have better access to tumor-associated antigens (TAA) within the tumor microenvironment. We conducted a first-in-human Phase I trial assessing the safety of lining the resection cavity with autologous TPFF/PCF of newly diagnosed patients with GBM. METHODS: 12 patients underwent safe, maximal surgical resection of newly diagnosed GBMs, followed by lining of the resection cavity with a pedicled, autologous TPFF or PCF. Safety was assessed by monitoring adverse events. Secondary analysis of efficacy was examined as the proportion of patients experiencing progression-free disease (PFS) as indicated by response assessment in neuro-oncology (RANO) criteria and overall survival (OS). The study was powered to determine whether a Phase II study was warranted based on these early results. For this analysis, subjects who were alive and had not progressed as of the date of the last follow-up were considered censored and all living patients who were alive as of the date of last follow-up were considered censored for overall survival. For simplicity, we assumed that a 70% PFS rate at 6 months would be considered an encouraging response and would make an argument for further investigation of the procedure. RESULTS: Median age of included patients was 57 years (range 46-69 years). All patients were Isocitrate dehydrogenase (IDH) wildtype. Average tumor volume was 56.6 cm3 (range 14-145 cm3). Resection was qualified as gross total resection (GTR) of all of the enhancing diseases in all patients. Grade III or above adverse events were encountered in 3 patients. No Grade IV or V serious adverse events occurred in the immediate post-operative period including seizure, infection, stroke, or tumor growing along the flap. Disease progression at the site of the original tumor was identified in only 4 (33%) patients (median 23 months, range 8-25 months), 3 of whom underwent re-operation. Histopathological analyses of those implanted flaps and tumor bed biopsy at repeat surgery demonstrated robust immune infiltrates within the transplanted flap. Importantly, no patient demonstrated evidence of tumor infiltration into the implanted flap. At the time of this manuscript preparation, only 4/12 (33%) of patients have died. Based on the statistical considerations above and including all 12 patients 10/12 (83.3%) had 6-month PFS. The median PFS was 9.10 months, and the OS was 17.6 months. 4/12 (33%) of patients have been alive for more than two years and our longest surviving patient currently is alive at 60 months. CONCLUSIONS: This pilot study suggests that insertion of pedicled autologous TPFF/PCF along a GBM resection cavity is safe and feasible. Based on the encouraging response rate in 6-month PFS and OS, larger phase II studies are warranted to assess and reproduce safety, feasibility, and efficacy. TRIAL REGISTRATION NUMBER AND DATE OF REGISTRATION FOR PROSPECTIVELY REGISTERED TRIALS: ClinicalTrials.gov ID NCT03630289, dated: 08/02/2018.

Digital health technology utilization is associated with enhanced patient perspectives of care.

BACKGROUND: Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) scores measure patient experience and perspectives on care. Novel health information technologies facilitate communication between patients and healthcare teams. Playback Health is a health information technology that incorporates multimedia for providers to communicate health information to patients, their support network, and their healthcare teams. We hypothesized implementing Playback Health may enhance patient perspectives on care. METHODS: HCAHPS scores were obtained retrospectively from a neurosurgical practice located in a metropolitan area between 2020 and 2022 for seven providers. Of these, four providers utilized Playback Health, and three did not. Individual providers' scores were compared between domains using two tailed t-tests at a significance level of p < 0.05. RESULTS: Playback Health use was associated with higher HCAHPS scores across varying domains from 2020 through 2022 as well as overall scores. In 2020, the mean overall score of HCAHPS users was higher than non-users (89.65 vs. 87.28, p = 0.0095). In 2021, again higher mean overall scores were observed in users as compared to non-users (89.11 vs. 87.79, p 0.0266). In 2022, Playback Health users maintained higher scores across communication domains and overall scores (p < 0.00001). Comparisons within domains between Playback Health users and non-users revealed no domains in which non-users had a significantly higher score than users. CONCLUSION: The addition of Playback Health multimedia health information technology was associated with improved patient satisfaction scores. When used as an adjunct to existing patient care, multimedia health information technologies may improve patient perceived care.

Connectome-guided initiation of dopamine agonists facilitates cognitive recovery after frontal lobe resection: A case report.

Abulia is a common problem that manifests following various brain conditions, including brain surgeries. Abulia is felt to be related to dysfunction with the brain's dopamine-dependent circuitry. The role of default mode network (DMN) in its pathogenesis is crucial. In this case report, we detail the presentation of abulia in an elderly woman following surgical resection of a right frontal glioblastoma involving the DMN. Connectomic imaging was used pre-operatively and post-operatively, demonstrating disruption of regions integral to the DMN and the central executive network. We observed a significant cognitive improvement following the administration of levodopa and carbidopa. Preoperative assessment of both anatomical and functional networks can help ensure surgical safety and predict postoperative deficits. This evaluation not only enhances preparedness and facilitates early case diagnosis but also expedites the initiation of prompt and potentially targeted treatments. This case highlights the potential efficacy of levodopa and carbidopa in addressing DMN dysfunction and broadly suggests the potential for connectomics-guided post-operative therapies.

A Quantitative Assessment of ChatGPT as a Neurosurgical Triaging Tool.

BACKGROUND AND OBJECTIVES: ChatGPT is a natural language processing chatbot with increasing applicability to the medical workflow. Although ChatGPT has been shown to be capable of passing the American Board of Neurological Surgery board examination, there has never been an evaluation of the chatbot in triaging and diagnosing novel neurosurgical scenarios without defined answer choices. In this study, we assess ChatGPT's capability to determine the emergent nature of neurosurgical scenarios and make diagnoses based on information one would find in a neurosurgical consult. METHODS: Thirty clinical scenarios were given to 3 attendings, 4 residents, 2 physician assistants, and 2 subinterns. Participants were asked to determine if the scenario constituted an urgent neurosurgical consultation and what the most likely diagnosis was. Attending responses provided a consensus to use as the answer key. Generative pretraining transformer (GPT) 3.5 and GPT 4 were given the same questions, and their responses were compared with the other participants. RESULTS: GPT 4 was 100% accurate in both diagnosis and triage of the scenarios. GPT 3.5 had an accuracy of 92.59%, slightly below that of a PGY1 (96.3%), an 88.24% sensitivity, 100% specificity, 100% positive predictive value, and 83.3% negative predicative value in triaging each situation. When making a diagnosis, GPT 3.5 had an accuracy of 92.59%, which was higher than the subinterns and similar to resident responders. CONCLUSION: GPT 4 is able to diagnose and triage neurosurgical scenarios at the level of a senior neurosurgical resident. There has been a clear improvement between GPT 3.5 and 4. It is likely that the recent updates in internet access and directing the functionality of ChatGPT will further improve its utility in neurosurgical triage.

Assessing views and attitudes toward the use of extended reality and its implications in neurosurgical education: a survey of neurosurgical trainees.

OBJECTIVE: Extended reality (XR) systems, including augmented reality (AR), virtual reality (VR), and mixed reality, have rapidly emerged as new technologies capable of changing the way neurosurgeons prepare for cases. Thus, the authors sought to evaluate the perspectives of neurosurgical trainees on the integration of these technologies into neurosurgical education. METHODS: A 20-question cross-sectional survey was administered to neurosurgical residents and fellows to evaluate perceptions of the use of XR in neurosurgical training. Respondents evaluated each statement using a modified Likert scale (1-5). RESULTS: One hundred sixteen responses were recorded, with 59.5% of participants completing more than 90% of the questions. Approximately 59% of participants reported having institutional access to XR technologies. The majority of XR users (72%) believed it was effective for simulating surgical situations, compared with only 41% for those who did not have access to XR. Most respondents (61%) agreed that XR could become a standard in neurosurgical education and a cost-effective training tool (60%). Creating patient-specific anatomical XR models was considered relatively easy by 56% of respondents. Those with XR access reported finding it easier to create intraoperative models (58%) than those without access. A significant percentage (79%) agreed on the need for technical skill training outside the operating room (OR), especially among those without XR access (82%). There was general agreement (60%) regarding the specific need for XR. XR was perceived as effectively simulating stress in the OR. Regarding clinical outcomes, 61% believed XR improved efficiency and safety and 48% agreed it enhanced resection margins. Major barriers to XR integration included lack of ample training hours and/or time to use XR amid daily clinical obligations (63%). CONCLUSIONS: The data presented in this study indicate that there is broad agreement among neurosurgical trainees that XR holds potential as a training modality in neurosurgical education. Moreover, trainees who have access to XR technologies tend to hold more positive perceptions regarding the benefits of XR in their training. This finding suggests that the availability of XR resources can positively influence trainees' attitudes and beliefs regarding the utility of these technologies in their education and training.

Characterizing complications associated with skull clamps: A review of the manufacturer and user facility device experience database.

OBJECTIVE: Head immobilization with skull clamps is a prerequisite of many neurosurgical procedures. Adverse events relating to the use of skull clamps have been reported, however, given the paucity of published reports, we sought to conduct a more comprehensive analysis using the Manufacturer and User Facility Device Experience (MAUDE) database. METHODS: The MAUDE database was queried for neurosurgical skull clamp events over a 10-year period between January 2013 and December 2022. Reports were qualitatively analyzed and categorically assigned by the study authors as 'mechanical failure,' 'slippage,' 'contamination,' 'insufficient information,' and 'other.' Patient injury reports were also classified as 'abrasion,' 'laceration,' 'hematoma,' 'fracture,' 'intracranial hemorrhage (ICH),' 'other,' 'insufficient information,' and 'death.' RESULTS: Of 3672 reports retrieved, 2689 (73.2%) were device malfunctions, with mechanical failure (50.7%) and slippage (47.7%) being the most common causes. There were 983 reports (26.8%) involving patient injury which included lacerations (n = 776, 78.9%), fractures (n = 24, 2.4%), abrasions (n = 23, 2.3%), hematomas (n = 7, 0.71%), ICH (n = 3, 0.31%), and other causes (n = 6, 0.61%). Five (0.1%) deaths due to skull clamp related complications were also reported. CONCLUSIONS: This study provides a more comprehensive picture of adverse events in neurosurgical procedures relating to the use of skull clamps. Mechanical failures of device parts were the most common device-related complication, and lacerations the most common adverse patient-related event. While more severe patient-related events were reported, they are relatively rare. The MAUDE database is useful for characterizing underreported device-related and patient-related adverse events.

Whole brain radiation therapy resulting in radionecrosis: a possible link with radiosensitising chemoimmunotherapy.

Radionecrosis describes a rare but serious complication of radiation therapy. In clinical practice, stereotactic radiosurgery (SRS) is increasingly used in combination with systemic therapy, including chemotherapy, immune checkpoint inhibitor and targeted therapy, either concurrently or sequentially. There is a paucity of literature regarding radionecrosis in patients receiving whole brain radiation therapy (WBRT) alone (without additional SRS) in combination with immunotherapy or targeted therapies. It is observed that certain combinations increase the overall radiosensitivity of the tumorous lesions. We present a rare case of symptomatic radionecrosis almost 1 year after WBRT in a patient with non-squamous non-small cell lung cancer on third-line chemoimmunotherapy. We discuss available research regarding factors that may lead to radionecrosis in these patients, including molecular and genetic profiles, specific drug therapy combinations and their timing or increased overall survival.

Health-Related Quality of Life Outcomes in Meningioma Patients Based upon Tumor Location and Treatment Modality: A Systematic Review and Meta-Analysis.

Patients with meningiomas may have reduced health-related quality of life (HRQoL) due to postoperative neurological deficits, cognitive dysfunction, and psychosocial burden. Although advances in surgery and radiotherapy have improved progression-free survival rates, there is limited evidence regarding treatment outcomes on HRQoL. This review examines HRQoL outcomes based on tumor location and treatment modality. A systematic search in PubMed yielded 28 studies with 3167 patients. The mean age was 54.27 years and most patients were female (70.8%). Approximately 78% of meningiomas were located in the skull base (10.8% anterior, 23.3% middle, and 39.7% posterior fossae). Treatment modalities included craniotomy (73.6%), radiotherapy (11.4%), and endoscopic endonasal approach (EEA) (4.0%). The Karnofsky Performance Scale (KPS) was the most commonly utilized HRQoL instrument (27%). Preoperative KPS scores > 80 were associated with increased occurrence of postoperative neurological deficits. A significant difference was found between pre- and post-operative KPS scores for anterior/middle skull base meningiomas (SBMs) in comparison to posterior (SBMs) when treated with craniotomy. Post-craniotomy SF-36 scores were lower for posterior SBMs in comparison to those in the anterior and middle fossae. Risk factors for poor neurological outcomes include a high preoperative KPS score and patients with posterior SBMs may experience a greater burden in HRQoL.

A Bibliometric Analysis of the Rise of ChatGPT in Medical Research.

The rapid emergence of publicly accessible artificial intelligence platforms such as large language models (LLMs) has led to an equally rapid increase in articles exploring their potential benefits and risks. We performed a bibliometric analysis of ChatGPT literature in medicine and science to better understand publication trends and knowledge gaps. Following title, abstract, and keyword searches of PubMed, Embase, Scopus, and Web of Science databases for ChatGPT articles published in the medical field, articles were screened for inclusion and exclusion criteria. Data were extracted from included articles, with citation counts obtained from PubMed and journal metrics obtained from Clarivate Journal Citation Reports. After screening, 267 articles were included in the study, most of which were editorials or correspondence with an average of 7.5 +/- 18.4 citations per publication. Published articles on ChatGPT were authored largely in the United States, India, and China. The topics discussed included use and accuracy of ChatGPT in research, medical education, and patient counseling. Among non-surgical specialties, radiology published the most ChatGPT-related articles, while plastic surgery published the most articles among surgical specialties. The average citation number among the top 20 most-cited articles was 60.1 +/- 35.3. Among journals with the most ChatGPT-related publications, there were on average 10 +/- 3.7 publications. Our results suggest that managing the inevitable ethical and safety issues that arise with the implementation of LLMs will require further research exploring the capabilities and accuracy of ChatGPT, to generate policies guiding the adoption of artificial intelligence in medicine and science.

An algorithm for the treatment of concurrent pituitary adenoma and cavernous sinus aneurysm: A systematic review & case report.

BACKGROUND: Rarely, Pituitary adenomas (PA) can co-occur with intrasellar or intracavernous aneurysms. There is currently no clear guidance for the management of this dual pathology. We attempt to provide an algorithm to help guide clinical decision making for treatment of PAs co-occurring with adjacent cerebral aneurysms. METHODS: A comprehensive literature search was conducted following PRISMA guidelines using various databases. Search terms included "(Pituitary Adenoma OR Prolactinoma OR Macroadenoma OR Adenoma) AND (ICA OR Internal Carotid Artery OR paracliniod OR clinoid) Aneurysm AND (Intra-cavernous OR intracavernous OR intrasellar OR Cavernous)." RESULTS: A total of 24 studies with 24 patients were included. Twelve (50%) patients experienced visual symptoms. Ten patients (42%) had an aneurysm embedded within the adenoma. Fourteen patients (58%) had an aneurysm adjacent to the adenoma. Embedded aneurysms were significantly associated with rupture events. CONCLUSION: Vision loss is the most pressing determinant of treatment. In the absence of visual symptoms, the aneurysm should be treated first by coil embolization. If not amenable to coiling, place flow diverting stent followed by six months of anticoagulation and antiplatelet therapy. If visual loss is apparent, the adenoma-aneurysm spatial relationship becomes critical. In cases of an adjacent aneurysm, the adenoma should be removed transsphenoidally with extreme care and aneurysm rupture protocols in place. If the aneurysm is embedded within the adenoma, then a BTO is favored with permanent ICA occlusion followed by transsphenoidal resection if adequate collateral supply is demonstrated. If there is inadequate collateral supply, then an open-approach for amenable aneurysms with transcranial adenoma debulking should be performed.

First in-human intrathecal delivery of bevacizumab for leptomeningeal spread from recurrent glioblastoma: rationale for a dose escalation trial.

PURPOSE: To outline the dose rationale for the first in-human intrathecal delivery of bevacizumab for LMS of GBM. METHODS: A 19-year-old female patient presented to Lenox Hill Hospital following thalamic GBM recurrence. She subsequently underwent two infusions of intra-arterial BEV (NCT01269853) and experienced a period of relative disease stability until progression in 2022. One month later, MRI disclosed diffuse enhancement representative of LMS of GBM. The patient subsequently underwent five cycles of IT BEV in mid-2022 (IND 162119). Doses of 25 mg, 37.5 mg, 50 mg, 50 mg, and 37.8 mg were delivered at two-week intervals between doses 1-4. The final 37.8 mg dose was given one day following her fourth dose, given that the patient was to be discharged, traveled several hours to our center, and was tolerating therapy well. Dosage was decreased due to the short interval between the final two treatments. Shortly after IT BEV completion, she received a third dose of IA BEV. RESULTS: Our patient did not show any signs of serious adverse effects or dose limiting toxicities following any of the treatments. It is difficult to determine PFS due to the rapid progression associated with LMS of GBM and rapid timeframe of treatment. CONCLUSION: LMS continues to be a devastating progression in many types of cancer, including GBM, and novel ways to deliver therapeutics may offer patients symptomatic and therapeutic benefits.

Hippocampal avoidance in whole brain radiotherapy and prophylactic cranial irradiation: a systematic review and meta-analysis.

PURPOSE: We systematically reviewed the current landscape of hippocampal-avoidance radiotherapy, focusing specifically on rates of hippocampal tumor recurrence and changes in neurocognitive function. METHODS: PubMed was queried for studies involving hippocampal-avoidance radiation therapy and results were screened using PRISMA guidelines. Results were analyzed for median overall survival, progression-free survival, hippocampal relapse rates, and neurocognitive function testing. RESULTS: Of 3709 search results, 19 articles were included and a total of 1611 patients analyzed. Of these studies, 7 were randomized controlled trials, 4 prospective cohort studies, and 8 retrospective cohort studies. All studies evaluated hippocampal-avoidance whole brain radiation treatment (WBRT) and/or prophylactic cranial irradiation (PCI) in patients with brain metastases. Hippocampal relapse rates were low (overall effect size = 0.04; 95% confidence interval [0.03, 0.05]) and there was no significant difference in risk of relapse between the five studies that compared HA-WBRT/HA-PCI and WBRT/PCI groups (risk difference = 0.01; 95% confidence interval [- 0.02, 0.03]; p = 0.63). 11 out of 19 studies included neurocognitive function testing. Significant differences were reported in overall cognitive function and memory and verbal learning 3-24 months post-RT. Differences in executive function were reported by one study, Brown et al., at 4 months. No studies reported differences in verbal fluency, visual learning, concentration, processing speed, and psychomotor speed at any timepoint. CONCLUSION: Current studies in HA-WBRT/HA-PCI showed low hippocampal relapse or metastasis rates. Significant differences in neurocognitive testing were most prominent in overall cognitive function, memory, and verbal learning. Studies were hampered by loss to follow-up.