Background: Neurosurgical conditions impose a significant burden on the Canadian healthcare system. This study quantifies the economic impact and explores predictive models for postoperative length of stay. Methods: We analyzed data from the Canadian Institute for Health Information National Health Expenditure Trends database for 2015-2019, focusing on case volumes, healthcare costs, and lengths of stay (LOS) across age groups and conditions. Decision tree models were created to predict total LOS from patient age and average acute LOS. Results: There was a modest increase in case volumes from 6,220 ± 3,103 in 2015 to 6,492 ± 3,240 in 2018, with a slight decrease in 2019. The total estimated hospital costs ranged from 2.27 ± 0.38 million CAD in 2015 to 2.23 ± 0.44 million CAD in 2019. The highest costs were seen in the 18-59 age group, at 2.53 ± 0.43 million CAD. Decision tree models showed high accuracy for predicting LOS in cases like spinal injury (F1-score: 0.98) but were less accurate for interventions with trauma or complications (F1-scores from 0.66 to 0.97). Conclusions: The study delineates the financial demands of neurosurgery in Canada and suggests decision tree models as useful tools for predicting hospital stay, with variable accuracy depending on the case complexity.

Background: The formation of pial-pial collateral network aneurysms due to carotid occlusion is a rare neurological phenomenon. This case details a 69-year-old male who developed a pial-pial collateral network aneurysm secondary to left internal carotid artery occlusion, leading to intracranial hemorrhage. Methods: The patient presented with altered consciousness due to left temporal intracerebral hemorrhage, subdural hematoma, and intraventricular hemorrhage. Cerebral angiography revealed an occluded left internal carotid artery, with superficial temporal artery (STA) and superior orbital artery anastomosis, and extensive pial-pial collaterals from the posterior temporal artery. A 4 mm aneurysm arising from this collateral network was identified. Surgical intervention involved a left temporal craniectomy and excision of the aneurysm, prioritizing the preservation of the STA. N.B., Informed patient consent was obtained in this study. Results: Successful aneurysm removal and preservation of collateral pathways were confirmed by postoperative imaging. The patient exhibited rapid neurological improvement; by postoperative day (POD) one, the patient showed limited response to stimuli. He was extubated by POD4 and discharged on POD27, where he conversed well, was independently ambulatory, and needed minimal to no assistance in activities of daily living. Conclusions: This case highlights the need for careful preoperative planning and intraoperative precision, especially in preserving vital collateral vascular pathways.

Background: The integration of Artificial Intelligence (AI) in medical education is an area of growing importance. While AI models have been evaluated extensively in multiple-choice question formats, their proficiency in written exams remains to be explored. Methods: Four AI models—GPT-4 (OpenAI), Claude-2.1 (Anthropic), Gemini Pro (Google), and Perplexity 70B (Perplexity)—were tested using the Canadian Royal College Sample Neurosurgery Exam. The written exam covered diagnostic reasoning, knowledge of neurosurgical conditions, and understanding of radiographic imaging techniques. Results: GPT-4 and Perplexity 70B both achieved a score of 68.42%, followed by Claude-2.1 with 60.53%, and Gemini Pro with 57.89%. The models showed proficiency in answering questions that required factual knowledge, such as identifying pathogens in spinal epidural abscess. However, they struggled with more complex diagnostic reasoning tasks, particularly in explaining the pathophysiology behind a sudden rise in blood pressure during surgery and interpreting radiographic characteristics of intracranial abscesses on MRI. Conclusions: The findings indicate that while AI models like GPT-4 and Perplexity 70B are adept at handling factual neurosurgical questions, their performance in complex diagnostic reasoning in a written format is less consistent. This underscores the need for more advanced and specialized AI training, particularly in the nuances of medical diagnostics and decision-making.

Background: Chordomas are rare, malignant bone tumors that present significant challenges in management and treatment due to their complex anatomical locations and propensity for recurrence. Advancements in artificial intelligence (AI) and machine learning (ML) show promise in improving chordoma management. Methods: A comprehensive literature search was conducted following PRISMA guidelines across multiple databases, including MEDLINE, Cochrane, Embase, Scopus, and Web of Science. The search targeted articles related to AI and ML applications in clinical tasks associated with chordoma management. The selection process involved systematic screening, data extraction, and assessment of inter-rater variability. Results: The search yielded 1,006 records, with 18 included for analysis. Convolutional neural networks (CNNs) excelled in tumor volume estimation, with the state-of-the-art model achieving a Dice similarity score of 74.2%, sensitivity of 79.4%, and positive predictive value of 74.3%. Clustering algorithms were effective in prognostic evaluations. Bayesian models and logistic regression demonstrated robustness in diagnostics. Support vector machines (SVMs) were noted for their diagnostic precision. Conclusions: AI and ML algorithms, particularly CNNs, clustering algorithms, Bayesian models, logistic regression, and SVMs, show promise in improving chordoma management through enhanced imaging, diagnostics, and prognostics. Future research should focus on larger, externally validated datasets and explore underutilized techniques like multi-modal data integration.

Background: Intraoperative testing for awake craniotomies requires a multidisciplinary team which may not be available in low-resource settings. We explored the creation of an AI tool for automated testing. Methods: We developed a NodeJS application, EloquentAid (https://www.eloquentaid.com/), for language testing automation. The workflow was as follows: users select an image-based naming task and verbally identify the image in English. Then, the application transcribes the response using OpenAI’s Whisper transcription service. Finally, the application evaluates response correctness. Feedback is provided through auditory and color signals. To assess its reliability, we tested the EloquentAid versus a human rater using a 57-item test based on the Boston Naming Test. Participants were neurosurgery and neurology residents from the Philippines. Qualitative surveys were obtained post-test. Results: A total of 798 observations were recorded (N=14). Human-application agreement was 60.52%. Cohen’s kappa was 0.31 (fair agreement). There were no false positive identifications by EloquentAid. Noun-type was felt to affect human error (i.e. “knocker,” “yolk,” “trellis”). Accent and pronunciation were felt to affect EloquentAid errors. Conclusions: EloquentAid is a promising tool to facilitate intraoperative testing and brain mapping using AI for speech recognition and response evaluation. Preliminary data shows fair human-app agreements. Improvements in test items and pronunciation recognition may be made.

Background: OSCE-GPT (https://learnmedicine.ca/) is an AI-based app that integrates history, physical exam, and relevant components for case guidance across medical disciplines to help trainees improve clinical skills. With global users across 60+ countries, this preliminary quality improvement study gathers user feedback on neurology and otolaryngology cases. Methods: A survey was distributed to users at the University of Ottawa and Cumming School of Medicine. Participants provided insights on the app’s use, perceived benefits, and suggested improvements. Results: Using 5-point Likert scales, 13 respondents, 9 of which evaluated an otolaryngology case, rated the overall usefulness of the learning tool 4.57± 0.51 (1=very poor, 5=very good), with a score of 4.00±0.65 relative to other teaching methods, such as didactic lectures or grand rounds (1=much worse, 5=much better). Users noted realistic interactions and self-paced learning as beneficial factors. Areas for improvement included a more fluid transition between physical exams and history, geographic variations in cases, and the addition of elements such as non-verbal patient cues or emotional. Conclusions: This study demonstrates utility of OSCE-GPT for medical trainees, particularly for otolaryngology and neurology cases. As cases continue to be added, feedback will be implemented to further improve user experience.

Time-varying flow separation on an accelerating prolate spheroid has been studied at various angles of incidence. Instantaneous pressure and scanning stereoscopic particle image velocimetry were used to shed light on the evolution of cross-flow structures for the Reynolds number ($Re$) range of $1.0\times 10^6\leq Re \leq 1.5\times 10^6$. The movement of separation lines is examined for various model accelerations to investigate on the interplay between acceleration and flow separation. The results demonstrate that for axial accelerations, the streamwise pressure distribution in the rear part of the prolate spheroid switches from an adverse to a favourable pressure gradient. At the same time, the circumferential adverse pressure gradient present during steady motion vanishes during said accelerations. In contrast, both streamwise and circumferential adverse pressure gradients strengthen when the model is axially decelerated. These dynamic pressure distributions influence the location of the separation line, which in turn moves closer to the model meridian during accelerations while moving outwards during decelerations. The streamwise vorticity distribution and the streamwise circulation both show how the separation-line position impacts the vortex formation. A high-vorticity region near the model surface is established during acceleration. In contrast, a decelerating model leads to transport of high-vorticity fluid into the outer area of the cross-flow separation. We further assess the memory effects following the near-impulsive velocity changes. The cross-flow retains the memory of moving separation lines shortly after the acceleration. However, the separation recovers quickly to a steady state.

Zinc (Zn) is an essential mineral and its deficiency manifests in non-specific clinical signs that require long time to develop. The response of swine intestine to Zn restriction was evaluated to identify early changes that can be indicative of Zn deficiency. Twenty-seven pigs (body weight = 77⋅5 ± 2⋅5 kg) were assigned to one of three diets: diet without added Zn (Zn-restricted diet, ZnR), and ZnR-supplemented with either 50 (Zn50) or 100 mg of Zn/kg of diet (Zn100) of Zn supplied by ZnCl2. After 32 d consuming the diets, serum Zn concentration in ZnR pigs was below the range of 0⋅59–1⋅37 μg/ml considered sufficient, thereby confirming subclinical Zn deficiency. Pigs showed no obvious health or growth changes. RNA-seq analysis followed by qPCR showed decreased expression of metallothionein-1 (MT1) (P < 0⋅05) and increased expression of Zn transporter ZIP4 (P < 0⋅05) in jejunum and ileum of ZnR pigs compared with Zn-supplemented pigs. Ingenuity pathway analysis revealed that Zn50 and Zn100 induced changes in genes related to nucleotide excision repair and integrin signalling pathways. The top gene network in the ZnR group compared with Zn100 was related to lipid and drug metabolism; and compared with Zn50, was related to cellular proliferation, assembly and organisation. Dietary Zn concentrations resulted in differences in genes related to immune pathways. Our analysis showed that small intestine presents changes associated with Zn deficiency after 32 d of Zn restriction, suggesting that the intestine could be a sentinel organ for Zn deficiency.

Background: A phase 3 trial, ADVANCE (NCT03777059), demonstrated that atogepant, an oral, CGRP receptor antagonist dosed once daily, results in clinically meaningful reductions in mean monthly migraine days. This open-label extension for ADVANCE trial completers evaluated long-term safety and tolerability of atogepant over 40-weeks. Methods: Participants in this trial (NCT03939312), rolled over from the ADVANCE trial, were treated with atogepant 60mg once daily for 40-weeks, with a 4-week safety follow-up. Only safety data were collected. Results: 685 participants took at least one dose of study drug, 74.6% completed the 40-week treatment period; mean age of 41.8 years, 88.2% female, 84.4% white, and mean BMI of 30.58 kg/m2. Mean (SD) treatment duration was 233.6 (89.32) days. 62.5% of participants experienced a treatment-emergent adverse event (TEAE), with 8.8% considered treatment-related by the investigator; serious adverse events (SAEs) occurred in 3.4% of participants, none were treatment-related. The most frequent AEs leading to discontinuation was nausea (0.4%, n=3); the most frequent TEAEs observed included upper respiratory tract infection (5.5%, n=38) and urinary tract infection (5.3%, n=36). No deaths or hepatic safety issues were observed. Conclusions: Safety results are consistent with known safety profile of atogepant and support long-term safety and tolerability of once daily dosing of atogepant 60mg.

In this paper, to address the cooperative localisation of a heterogeneous UAV swarm in the GNSS-denied environment, an adaptive simulated annealing-particle swarm optimisation (SA-PSO) cooperative localisation algorithm is proposed. Firstly, the forming principle of the communication and measurement framework is investigated in light of a heterogeneous UAV swarm composition. Secondly, a reasonably cooperative localisation function is established based on the proposed forming principle, which can minimise the relative localisation error with limited available information. Then, an adaptive weight principle is incorporated into the particle swarm optimisation (PSO) for better performance. Furthermore, in order to overcome the drawbacks of PSO algorithm easily falling into the local extreme point, an adaptive SA-PSO algorithm is improved to promote the convergence speed of cooperative localisation. Finally, comparative simulations are performed among the adaptive SA-PSO, adaptive PSO, and PSO algorithms to demonstrate the feasibility and superiority of the proposed adaptive SA-PSO algorithm. Simulation results show that the proposed algorithm has better performance in convergence speed, and the cooperative localisation precision can be guaranteed.

This study aimed to identify gender-specific co-developmental trajectories of internalizing and externalizing problems from middle childhood to early adolescence, along with key environmental and individual predictors among Chinese youth. A total of 1653 Chinese elementary school students (Mage = 9.40; SD = 0.51; 54.57% boys) participated in assessments at six time points, using 6-month assessment intervals. Parallel process latent class growth modeling identified four trajectories for boys: Congruent-low (65.74%), moderate-decreasing internalizing and moderate-stable externalizing problems (18.40%), high increasing-internalizing and low-stable externalizing problems (8.20%), and high decreasing-internalizing and low-stable externalizing problems (7.65%). Three trajectories were identified for girls: Congruent-low (81.09%), moderate co-occurring (7.19%), and high increasing-internalizing and low-stable externalizing problems (11.72%). Multivariate logistic regression analyses revealed that peer victimization served as an environmental risk predictor for the adverse co-developmental trajectories of internalizing and externalizing problems for boys and girls. High sensation-seeking and low self-control served as individual risk variables predicting the trajectories of high increasing-internalizing and low-stable externalizing problems, and low self-control also predicted the trajectories of high decreasing-internalizing and low-stable externalizing problems for boys. The findings highlight the importance of gender differences in understanding the progression of internalizing and externalizing problems and inform effective strategies for prevention and intervention.

Drug-induced liver injury (DILI) is a common adverse drug reaction leading to the interruption of tuberculosis (TB) therapy. We aimed to identify whether the hepatitis B virus (HBV) infection would increase the risk of DILI during first-line TB treatment. A meta-analysis of cohort studies searched in PubMed, Web of Science and China National Knowledge Infrastructure was conducted. Effect sizes were reported as risk ratios (RRs) and 95% confidence intervals (CIs) and calculated by R software. Sixteen studies with 3960 TB patients were eligible for analysis. The risk of DILI appeared to be higher in TB patients co-infected with HBV (RR 2.66; 95% CI 2.13–3.32) than those without HBV infection. Moreover, patients with positive hepatitis B e antigen (HBeAg) were more likely to develop DILI (RR 3.42; 95% CI 1.95–5.98) compared to those with negative HBeAg (RR 2.30; 95% CI 1.66–3.18). Co-infection with HBV was not associated with a higher rate of anti-TB DILI in latent TB patients (RR 4.48; 95% CI 0.80–24.99). The effect of HBV infection on aggravating anti-TB DILI was independent of study participants, whether they were newly diagnosed with TB or not. Besides, TB and HBV co-infection patients had a longer duration of recovery from DILI compared to non-co-infected patients (SMD 2.26; 95% CI 1.87–2.66). To conclude, the results demonstrate that HBV infection would increase the risk of DILI during TB therapy, especially in patients with positive HBeAg, and close liver function monitoring is needed for TB and HBV co-infection patients.

Introduction: There is ongoing concern about the burden placed on healthcare systems by lab tests. Although these concerns are widespread, it is difficult to quantify the extent of the problem. One approach involves use of a metric known as the Mean Abnormal Response Rate (MARR), which is the proportion of tests ordered that return an abnormal result; a higher MARR value indicates higher yield. The primary objective of this study was to calculate MARRs for tests ordered between April 2014 and March 2019 at the four adult emergency departments (EDs) covering a metropolitan population of 1.3 million. Secondary objectives included identifying tests with highest and lowest MARRs; comparison of MARRs for nurse- and physician-initiated orders; correlation of the number of tests per order requisition to MARR; and correlation of physician experience to MARR. Methods: In total, 40 laboratory tests met inclusion criteria for this study. Administrative data on these tests as ordered at the four EDs were obtained and analyzed. Multi-component test results, such as from CBC, were consolidated such that an abnormal result for any component was coded as an abnormal result for the entire test. Repeat tests ordered within a single patient visit were excluded. Physician experience was quantified for 209 ED physicians as number of years since licensure. Analyses were descriptive where appropriate for whole-population data. Risk of bias was attenuated by the focus on administrative data. Results: The population dataset comprised 33,757,004 test results on 415,665 unique patients. Of these results, 30.3% were the outcomes of nurse-initiated orders. The 5-year MARRs for the four hospitals were 38.3%, 40.0%, 40.7% and 40.9%. The highest per-test MARRs were for BNP (80.5%) and CBC (62.6%), while the lowest were for glucose (7.9%) and sodium (11.6%). MARRs were higher for nurse-initiated orders than for physician-initiated orders (44.7% vs. 38.1%), likely due to the greater order frequency of high-yield CBC in nurse-initiated orders (38.6% vs. 18.1%). The number of tests per order requisition was inversely associated with MARR (r = -0.90, p < 0.001). Finally, the number of years since licensure was modestly but significantly associated with MARR (r = 0.28, p < 0.001). Conclusion: This is the first and largest study to apply the MARR in an ED setting. As a metric, MARR effectively identifies differences in test ordering practices on per-test and per-hospital bases, which could be useful for data-informed practice optimization.