Background: Adult spinal deformities (ASD) are varied spinal abnormalities, often necessitating surgical intervention when associated with pain, worsening deformity, or worsening function. Predicting post-operative complications and revision surgery is critical for surgical planning and patient counseling. Due to the relatively small number of cases of ASD surgery, machine learning applications have been limited to traditional models (e.g., logistic regression or standard neural networks) and coarse clinical variables. We present the novel application of advanced models (CNN, LLM, GWAS) using complex data types (radiographs, clinical notes, genomics) for ASD outcome prediction. Methods: We developed a CNN trained on 209 ASD patients (1549 radiographs) from the Stanford Research Repository, a CNN pre-trained on VinDr-SpineXR (10,468 spine radiographs), and an LLM using free-text clinical notes from the same 209 patients, trained via Gatortron. Additionally, we conducted a GWAS using the UK Biobank, contrasting 540 surgical ASD patients with 7355 non-surgical ASD patients. Results: The LLM notably outperformed the CNN in predicting pulmonary complications (F1: 0.545 vs. 0.2881), neurological complications (F1: 0.250 vs. 0.224), and sepsis (F1: 0.382 vs. 0.132). The pre-trained CNN showed improved sepsis prediction (AUC: 0.638 vs. 0.534) but reduced performance for neurological complication prediction (AUC: 0.545 vs. 0.619). The LLM demonstrated high specificity (0.946) and positive predictive value (0.467) for neurological complications. The GWAS identified 21 significant (p < 10⁻⁵) SNPs associated with ASD surgery risk (OR: mean: 3.17, SD: 1.92, median: 2.78), with the highest odds ratio (8.06) for the LDB2 gene, which is implicated in ectoderm differentiation. Conclusions: This study exemplifies the innovative application of cutting-edge models to forecast outcomes in ASD, underscoring the utility of complex data in outcome prediction for neurosurgical conditions. It demonstrates the promise of genetic models when identifying surgical risks and supports the integration of complex machine learning tools for informed surgical decision-making in ASD.

Neuronal loss can considerably diminish neural circuit function, impairing normal behavior by disrupting information flow in the circuit. Here, we use genetically engineered electrical synapses to reroute the flow of information in a C. elegans damaged chemosensory circuit in order to restore organism behavior. We impaired chemotaxis by removing one pair of interneurons from the circuit then artificially coupled two other adjacent neuron pairs by ectopically expressing the gap junction protein, connexin, in them. This restored chemotaxis in the animals. We expected to observe linear and direct information flow between the connexin-coupled neurons in the recovered circuit but also revealed the formation of new potent left-right lateral electrical connections within the connexin-expressing neuron pairs. Our analysis suggests that these additional electrical synapses help restore circuit function by amplifying weakened neuronal signals in the damaged circuit in addition to emulating the wild-type circuit. A record of this paper's transparent peer review process is included in the Supplemental Information.

Central Nervous System Tuberculosis (CNS-TB) is a serious public health concern causing significant morbidity and mortality, especially in high TB burden countries. Despite the expanding research landscape of CNS-TB, there is no comprehensive map of this field. This work aims to (1) obtain a current and comprehensive overview of the CNS-TB research landscape, (2) investigate the intellectual and social structure of CNS-TB publications, and (3) detect geographical discrepancies in scientific production, highlighting regions requiring increased research focus. We conducted a bibliometric analysis on CNS-TB literature indexed in Web of Science from 2000 to 2022, evaluating 2130 articles. The dataset was analyzed in R for descriptive statistics. We used R-bibliometrix and VOSViewer for data visualization. Publication output grew annually at an average rate of 6·88%, driven primarily by India and China. International collaborations comprised 16·44% of total publications but contributed to 11 of the 15 top-cited papers. Additionally, we identified discrepancies of CNS-TB research in many low- and middleincome countries relative to their TB incidence. Our findings reveal a growing interest in CNS-TB research from China and India, countries with rapidly developing economies, high TB burdens, and a recent increase in research funding. Furthermore, we found that international collaborations are correlated with high impact and accessibility of CNS-TB research. Finally, we identified disparities in CNS-TB research in specific countries, particularly in many low- and middle-income countries, emphasizing the need for increased research focus in these regions.

Introduction: The advancement of artificial intelligence (AI) in neurosurgery is dependent on high quality, large, labeled datasets. Labeled neurosurgical datasets are rare, driven by the high expertise required for labeling neurosurgical data. A comprehensive resource overviewing available datasets for AI in neurosurgery is essential to identify areas for potential model building and areas of needed data construction. Methods: We conducted a systematic review according to PRISMA guidelines to identify publicly available neurosurgical datasets suitable for machine learning. A PubMed search on 8 February 2025, yielded 267 articles, of which 86 met inclusion criteria. Each study was reviewed to extract dataset characteristics, model development details, validation status, availability, and citation impact. Results: Among the 86 included studies, 83.7% focused on spine pathology, with tumor (3.5%), vascular (4.7%), and trauma (7.0%) comprising the remaining. The majority of datasets were image-based, particularly X-ray (37.2%), MRI (29.1%), and CT (20.9%). Label types included segmentation (36.0%), diagnosis (26.7%), and detection/localization (20.9%), with only 2.3% including outcome labels. While 97.7% of studies reported training a model, only 22.6% performed external validation, 20.2% shared code, and just 7.1% provided public applications. Accuracy was the most frequently reported performance metric, even for segmentation tasks, where only 60% of studies used the Dice score metric. Studies often lacked task-appropriate evaluation metrics. Conclusions: We conducted a systematic review to capture all publicly accessible datasets that can be applied to build AI models for neurosurgery. Current datasets are heavily skewed towards spine imaging and lack both clinical patient specific and outcomes information. Provided baseline models from these datasets are limited by poor external validation, lack of reproducibility, and reliance on suboptimal evaluation metrics. Future efforts should prioritize developing multi-institutional datasets with outcome labels, validated models, public access, and domain diversity to accelerate the safe and effective integration of AI into neurosurgical care.

Background: Intracranial hypertension is a life-threatening complication of acute brain injuries such as traumatic brain injury (TBI), subarachnoid hemorrhage (SAH), or intracerebral hemorrhage (ICH). In low-income and middle-income countries (LMICs), limited resources can delay timely neurocritical interventions. Smartphone-based quantitative pupillometry offers a scalable solution for early detection of elevated intracranial pressure (ICP). Here, we assessed its ability to (1) detect raised optic nerve sheath diameter (ONSD), a noninvasive surrogate for elevated ICP, and (2) classify severe TBI. Methods: Thirty-eight Nepali ICU patients with TBI (n=16), SAH (n=10), or ICH (n=12) underwent daily sonographic ONSD and pupillary light reflex (PLR) assessments through the PupilScreen app (Apertur Inc., Seattle, WA) over 7 days. Machine learning classifiers were trained on PLR features to detect elevated ONSD (>6.0 mm). To identify severe TBI (Glasgow Coma Scale [GCS] ≤8 on admission), classifiers were trained on PLR features, ONSD, or both. Results: For ONSD >6.0 mm, a random forest model achieved an AUC of 0.66, with a sensitivity of 0.31 and specificity of 0.80. For identifying severe TBI, the optimal classifier was a random forest model incorporating ONSD and a subset of PLR metrics, with a sensitivity of 0.93, specificity of 1.00, and AUC of 0.96. Conclusion: In this pilot study, smartphone-based pupillometry showed modest ability for detecting elevated ONSD. However, its high performance in severe TBI classification warrants further evaluation. Larger, multicenter studies evaluating triage utility in prehospital and resource-limited settings are warranted to validate and extend these findings.