Citations

Artificial intelligence (AI) is increasingly shaping pediatric spine care, leveraging its rapid advancements in healthcare to improve efficiency, accuracy, and disease understanding. Moreover, machine learning and deep learning excel at detecting complex patterns. This holds promise in processing spinal deformity data, with the potential to surpass traditional statistical methods in predictive accuracy. Challenges persist, however, including unclear clinical implementation guidelines, limited model transparency, and ethical concerns surrounding data privacy and bias. Small sample sizes and the need for larger, diverse datasets further complicate integration. In order to realize AI’s transformative potential in pediatric spine care, these critical obstacles must be addressed for effective and ethical clinical adoption. This review examines the role of AI through applications such as image sorting, surgical outcome prediction, forecasting of spinal curve progression, and vertebral volumetric analysis using deep reasoning. It also explores possible intraoperative contributions from AI, including robotics and optimized screw trajectory planning, and the potential of large language models in clinical practice.

Abstract Atraumatic and localized drug delivery to the vascular endothelium remains a critical unmet need in interventional medicine, with major implications for the management of arterial and venous diseases. Current approaches, such as drug-eluting stents and drug-coated balloons, combine mechanical revascularization with local drug release but frequently denude the endothelium and injure vessel walls, leading to restenosis, thrombosis, and chronic inflammatory repair cycles that severely limit long-term outcomes. Here, we present EndoBot, a soft, untethered millirobot designed for atraumatic navigation and targeted endovascular drug delivery during physiologic blood flow. EndoBot employs magnetically actuated corkscrew propulsion and mechanically adaptive surface crawling to maintain low radial pressure (<1 kPa) and preserve endothelial integrity. Its feasibility is established in arterial and venous phantom models, ex vivo human umbilical veins under normothermic perfusion, and in vivo rat inferior vena cava, all under clinical fluoroscopic guidance using a human-scale magnetic manipulation system. Despite periodic vessel motion, compression, and geometric irregularities, EndoBot maintains stable, mechanically adaptive navigation without endothelial injury. Under real-time fluoroscopic imaging, we identify key challenges and showcase clinically feasible strategies for 3D localization and tracking, overcoming intrinsic depth limitations of 2D clinical imaging. The device can be deployed and retrieved through standard vascular sheaths and remains stable under supraphysiologic arterial and venous flow conditions (flow rate >100 mL min-1; flow velocity >155 cm s-1, shear stress >10 Pa). Blood-compatibility testing demonstrates no increase in coagulation tendency and minimal hemolysis (<0.01%) during magnetic actuation. For drug delivery, EndoBot employs a novel endoluminal painting-based technique that transfers a hydrophobic and flow-resistant coating directly onto the inner vessel surface. This method ensures uniform deposition of a biodegradable drug depot without generating fragments larger than typical capillary diameters (>10 mm). By enabling local, atraumatic drug delivery without compromising vessel wall integrity, EndoBot first time demonstrates a translationally viable platform for exploring effectiveness in preventive interventions in early-stage vascular disease and for adjunctive therapies in advanced disease stages.

Purpose: The purpose of this study was to compare the height of velopharyngeal (VP) closure in relation to the palatal plane between individuals with velopharyngeal insufficiency (VPI) with or without cleft palate and individuals without cleft palate and with normal speech, considering the effects of age and sex. Method: Magnetic resonance imaging was used to analyze the height of VP closure in the midsagittal view during phonation of /i/ and /s/ sounds among 105 individuals with VPI and 45 individuals without cleft palate and normal speech. Results: Independent-samples t tests demonstrated that individuals with VPI display a significantly lower mean height of VP closure (3.7 mm) compared to individuals without cleft palate and normal speech during phonation of /i/ and /s/ sounds ( p < .001). Analysis of variance was used to evaluate the impact of age and sex on the height of VP closure. Significant differences were observed between the height of VP closure in individuals with VPI and those without cleft palate and normal speech across the age span during the production of /i/ and /s/ sounds. However, nonsignificant differences were found between the height of VP closure in males and females of the VPI group and those without cleft palate and normal speech during the production of these sounds. Conclusions: The results revealed the variations of the height of VP closure in relation to the palatal plane. These findings provide insights into the presurgical planning for individuals with VPI undergoing palatal surgeries, especially the location of pharyngeal flap insertion.

Foundation models can streamline imaging workflows, improve diagnostic accuracy, enable personalized treatment, and, via integrated generative artificial intelligence, create synthetic images to augment scarce real-world data; however, successful adoption requires explainability, trust, and workflow integration.

Objectives Heritable connective tissue disorders (HCTDs), including hypermobile Ehlers-Danlos syndrome (hEDS) and hypermobile spectrum disorder (HSD), have complex clinical manifestations, but the prevalence of left common iliac vein (LCIV) compression in these groups is not well known. This study aimed to compare the prevalence of significant LCIV compression in patients with hEDS, HSD, and healthy controls. Methods This retrospective, IRB-approved, single-center study analyzed 109 patients (25 hEDS, 64 HSD, 20 controls) who underwent contrast-enhanced abdominal and pelvic CT imaging from 2020 to 2023. LCIV compression was measured on axial CT, with >50% luminal narrowing defined as significant. Statistical analyses included chi-square tests, odds ratios, and t -tests. Interrater reliability was evaluated using intraclass correlation coefficients. Results Significant LCIV compression occurred in 68.0% of hEDS, 65.6% of HSD, and 35.0% of controls ( p = 0.034). Both hEDS ( p = 0.027) and HSD ( p = 0.015) groups had significantly higher compression rates than controls, with no statistically significant difference between the two clinical groups ( p = 0.831). Odds of compression were nearly fourfold greater in hEDS (OR 3.95) and HSD (OR 3.54) compared to controls. Conclusion LCIV compression is significantly more prevalent in patients with hEDS and HSD than in healthy controls, warranting further investigation of its clinical significance.

Abstract Objective Dravet syndrome (DS) is a developmental and epileptic encephalopathy characterized by drug‐resistant seizures and developmental slowing. Although cognitive and executive function deficits have been described, their early trajectory is not well understood. Methods The prospective ENVISION natural history study (NCT04537832) assessed cognitive, executive, and adaptive function in children younger than 5 years of age with SCN1A + DS every 6 months for up to 2 years using Bayley Scales of Infant and Toddler Development, 3rd Edition (BSID‐III), Wechsler Preschool & Primary Scale of Intelligence, 4th Edition (WPPSI‐IV), Vineland Adaptive Behavior Scales, 3rd Edition (VABS‐3), Behavior Rating Inventory of Executive Function – Preschool Version (BRIEF‐P), and Pediatric Evaluation of Disability Inventory (PEDI). Results Fifty‐eight children were enrolled, with 47% younger than age 2 years. At least 80% of children did not achieve age‐appropriate milestones. Mean BSID‐III Cognitive raw scores increased minimally, with age‐equivalent gains of only 3 months over 1.5 years. Mean Cognitive Composite scores declined significantly by Month 12 (from 81.6 to 72.2; change: –11.0, 95% confidence interval [CI]: −15.3 to −6.8), signaling a widening gap compared with neurotypical development. Executive function worsened, with mean BRIEF‐P Global Executive Composite T‐scores increasing by 3.2 points/year. For some participants, scores were 5 standard deviations (SD) above the normative mean, reflecting abilities profoundly below age expectations (bottom .00003% of the population). Adaptive functioning worsened, with mean VABS‐3 Adaptive Behavior Composite decreasing from 78.7 to 68.1 over 1.5 years (change: –9.0, 95% CI: −11.9 to −6.1) and greater decline among children <2 years at enrollment, with scores decreasing by ~15 points (1 SD). Over half of children >3 years could not remove clothing independently; and when placed on a toilet, 48% could not use it. Significance Infants and young children with SCN1A + DS show significant and progressive developmental slowing across several domains, highlighting urgent need for therapies to mitigate the devastating impact on individuals and families.

ABSTRACT Background Adolescents with youth‐onset type 2 diabetes (YO‐T2D) have an increased risk for cardiometabolic complications. The impact of vertical sleeve gastrectomy (VSG) on fat distribution and cardiac morphology/function in YO‐T2D is unknown. Objectives To evaluate changes in body composition, abdominal and cardiac fat depots, and cardiometabolic health in adolescents with YO‐T2D undergoing VSG. Methods Anthropometrics, labs and imaging were used to assess participants pre‐surgery and 3–12 months post‐surgery. MRI quantified pericardial (PAT), epicardial (EAT), subcutaneous (SAT) and visceral adipose tissue (VAT), hepatic fat fraction (HFF) and cardiac morphology/function. Mixed‐effects models assessed longitudinal changes. Results By 12 months, weight decreased from 134.3 ± 5.1 to 103.3 ± 5.2 kg, VAT 1943 ± 148 to 1248 ± 150 cm 3 , HFF 23.2% ± 2.3% to 5.3% ± 2.3% and PAT and EAT by 27% and 33% (all p < 0.01). Homeostatic model assessment of insulin resistance (HOMA‐IR) improved (7.8 ± 1.2 to 1.7 ± 1.2 [ p < 0.01]). BMI, SAT, resting heart rate (RHR), left ventricular (LV) mass and cardiac output (CO) also decreased (all p < 0.05). VAT decreases correlated with decreases in HFF ( r = 0.70, p = 0.01), HOMA‐IR ( r = 0.60, p = 0.04) and CO ( r = 0.70, p = 0.03). HFF decreases correlated with decreases in BMI ( r = 0.70, p = 0.03), HOMA‐IR ( r = 0.90, p < 0.001), RHR ( r = 0.90, p = 0.002), CO ( r = 0.80, p = 0.007) and LV mass ( r = 0.70, p = 0.02). Conclusions VSG reduces ectopic and regional fat and improves associated insulin sensitivity and cardiac health in adolescents with YO‐T2D.

Purpose of the review With the widespread integration of spinal cord stimulation (SCS) into clinical practice, understanding its ethical, economic, and patient-centered implications has become increasingly important. This review critically examines recent evidence across these domains to illuminate challenges and opportunities for advancing transparent, ethical, patient-centered, and value-based neuromodulation practice. Recent findings Recent analyses reveal persistent challenges with bias, conflicts of interest, and selective outcome reporting in neuromodulation research. Studies demonstrate significant disparities in access to SCS across racial and socioeconomic groups and highlight new ethical considerations associated with artificial intelligence–enabled and informed treatment in neuromodulation. Contemporary randomized trials support clinically meaningful improvements in pain, functionality, psychological outcomes, and other patient-centered outcomes, although durability remains inconsistent because of the potential for therapy habituation or adverse events. Despite high initial upfront costs, evaluations of SCS cost-effectiveness across healthcare systems generally favor SCS over conventional medical management when assessed over multiyear periods, with differential economic benefits observed based on the type of waveform and type of national healthcare system. Summary While current evidence supports the clinical and long-term economic benefits of SCS in defined clinical contexts and indications, gaps in research transparency, equitable access to care, economic considerations, and durability of effectiveness persist.

Study Design Retrospective cohort study using a nationally representative inpatient database. Objective To determine the incidence, timing, and predictors of new-onset psychiatric disorders following lumbar fusion surgery, and to propose a data-driven risk stratification model for early identification and intervention. Methods We queried the Nationwide Readmissions Database (2016-2019) to identify adults undergoing elective inpatient lumbar fusion. Patients with preexisting psychiatric diagnoses were excluded. Outcomes included new-onset depression, anxiety, suicidal ideation, and homicidal ideation within 12 months postoperatively. Predictors were assessed using odds ratios. Timing of onset was evaluated with density plots. Results Among 839,576 patients, 12,011 developed depression, 11,828 anxiety, 916 suicidal ideation, and 63 homicidal ideation. Most diagnoses occurred within 90 days postoperatively, with peak incidence around 3 weeks. Opioid use during the index admission significantly increased odds of suicidal ideation (OR 2.13, P = 0.001), anxiety (OR 1.26, P = 0.002), and showed a trend toward depression (OR 1.14, P = 0.087). Medicaid insurance was the strongest predictor of suicidal (OR 3.69) and homicidal ideation (OR 6.41). Female sex and younger age were associated with anxiety and depression; older age was protective against ideation. Conclusions New-onset psychiatric disorders are a frequent and early complication following lumbar fusion. Several modifiable and demographic factors - including perioperative opioid use and socioeconomic status - significantly predict risk. These findings underscore the need for structured mental health screening within the first 3 months post-surgery. Implementing a predictive risk stratification tool could enable earlier psychiatric referral, reduce long-term morbidity, and improve both recovery trajectories and overall quality of life.

Abstract INTRODUCTION Alzheimer's disease (AD) is a progressive neurodegenerative disorder that needs better predictive tools. Using the National Alzheimer's Coordinating Center Uniform Data Set, this study developed machine learning (ML) models and a practical clinical tool for AD prediction. METHODS Data from 52,537 individuals (22,371 with AD) and more than 200 variables were processed with MissForest imputation and genetic algorithm‐based selection. Multiple ML models were trained, and interpretability was performed using SHAP and permutation importance. A LightGBM model was refined through iterative backward feature elimination (IBFE) followed by manual refinement. RESULTS LightGBM performed best (receiver operating characteristic‐area under the curve [ROC‐AUC] 0.91, accuracy 82.0%). Key predictors included arthritis, age, body mass index, and heart rate. A 19‐feature model retained accuracy (81.2%) and ROC‐AUC (0.90). DISCUSSION This lightweight tool predicts AD using mostly routine variables. Limitations include its cross‐sectional nature, and would need external validation. An interactive web app and GitHub resource are available. Highlights Developed a lightweight ML based tool using 19 routinely available features. The lightweight model achieved an ROC-AUC of 0.90 for Alzheimer's disease prediction on NACC multicenter data. Genetic algorithm, IBFE, and manual refinement enabled optimal feature selection. Tool hosted on an open‐access platform for clinical and research use. SHAP analysis provided model interpretability and feature‐level insights.