Search Results (2,776)

Background/Objectives: Botanical nootropics are increasingly sought as natural alternatives to synthetic stimulants. Mangifera indica leaf extract, standardized to the polyphenol mangiferin, has shown promise in regulating brain activity and enhancing cognitive performance. This study aimed to replicate, in an independent cohort, the acute cognitive benefits of a soluble mango leaf extract (Zynamite^® S) following a previous proof-of-concept trial. Methods: In a double-blind, randomized, placebo-controlled crossover trial, 88 healthy young adults (aged 18–25) received either a single 100 mg dose of Zynamite^® S or a matched placebo. Cognitive performance was assessed using the Trail Making Test (TMT), Digit Symbol Substitution Test (DSST), and Stroop Color-Word Test. Emotional states were assessed via the Profile of Mood States (POMS) at baseline, and then at 30 min, 3 h, and 5 h post-ingestion. Results: Zynamite^® S supplementation resulted in the replication of previous findings, with significant improvements in mental processing speed, attention, and cognitive flexibility (p < 0.05). Notably, this trial identified a rapid onset of action, with significant cognitive and mood improvements detectable as early on as 30 min post-dose and a sustained duration of the effect with benefits observed up to 5 h post-administration. Mood assessments confirmed an overall improvement in emotional balance (p < 0.05) by reducing tension, mental fatigue, and low mood during the completion of cognitive-demanding tasks. Conclusions: These findings reproduce the acute benefits of Zynamite^® S in a healthy young adult population, including a rapid onset of effects and a sustained improvement in cognitive performance and mood over a 5 h window. Full article

Humans possess unique advantages in dealing with few-shot visual recognition scenarios, inspiring the development of meta-learning methods aimed at emulating these abilities. Current mainstream meta-learning primarily utilizes the monocular vision or the dual asymmetric complementary architectures, collectively referred to as Machine-visual Meta-Learning (MvML). Nevertheless, research has not yet developed an architecture for simulating the human binocular visual system, named Humanoid-visual Meta-Learning (HvML). This paper innovatively proposes an excellent paradigm BPIAL belonging to the HvML: Binocular Perception Instance Authentication Learning, which can alleviate the monocular shallowness and dual-branch processing instability of MvML. Structurally, our BPIAL comprises two interconnected binocular perception and information processing modules: BSEM and IAPM. The former module can simulate binocular visual field extraction, feature extraction and compression, and channel dimension reduction, while IAPM can simulate the logic, reasoning, and judgment processes of the two human visual branches and the brain. We have demonstrated the feasibility and correctness of BPIAL on five benchmarks. Sufficient and comparative experiments with the state-of-the-art methods have proved its superiority and effectiveness. Full article

Background: Danshen–Gegen is a classic herb pair in traditional Chinese medicine, which has been used to treat cardiovascular and cerebrovascular diseases. Ischemic stroke (IS) is a prevalent cerebrovascular condition; ferroptosis is one of the contributing factors driving the progression of IS. This study aims to determine the underlying mechanism and examine if Danshen–Gegen (DG) extract may prevent cerebral ischemia–reperfusion injury by preventing ferroptosis. Methods: The comprehensive compositional characterization of DG was analyzed by ultra-high-performance liquid chromatography coupled with hybrid quadrupole-orbitrap high-resolution mass spectrometry (UPLC-Q-orbitrap MS). The experiments were conducted in middle cerebral artery occlusion/reperfusion (MCAO/R) rats and oxygen-glucose deprivation/re-oxygenation (OGD/R) cells. The neuroprotective effects of DG on IS were assessed by examining rat survival rates, infarct volume, behavioral scores, and cerebral water content. Then, we tested the accumulation of Fe²⁺ and lipid peroxidation products such as reactive oxygen species (ROS), glutathione (GSH), malondialdehyde (MDA), myeloperoxidase (MPO), and 4-hydroxynonenal (4-HNE) in rats and cells. The expression of nuclear factor erythroid-derived 2-like 2 (Nrf2), Solute Carrier Family 7 Member 11 (-xCT), Glutathione peroxidase 4 (GPX4), Cyclooxygenase-2 (COX-2), Transferrin Receptor 1 (TFR1), and Long-chain-fatty-acid–CoA ligase 4 (ACSL4) was also assessed in vivo and in vitro. Results: UPLC-Q-orbitrap MS analysis was performed to characterize the chemical profile of DG, and a total of 33 chemical constituents were successfully identified. DG significantly alleviated the ischemic damage to brain tissue, reduced infarct volume, and improved neurological dysfunction. The content of Fe²⁺ and lipid peroxidation products was markedly decreased. Furthermore, DG could restore the expression of Nrf2, -xCT, and GPX4 with the inhibition of COX-2, TFR1, and ACSL4, thus achieving a suppressive effect on ferroptosis. Conclusions: The regulatory influence of DG via the Nrf2/System xc-/GPX4 axis may play a crucial role in alleviating ferroptosis and enhancing recovery from cerebral ischemia injury. Full article

Amyloidosis encompasses a spectrum of diseases in which insoluble protein aggregates are deposited in various parts of the body, including the brain, giving rise to Alzheimer’s disease, prion disease, and Parkinson’s disease, and also being a manifestation of Type II diabetes. The soluble protein gets aggregated as insoluble plaques by an unknown phenomenon, leading to the disease. If an agent is developed that can dissociate or disintegrate these plaques, it can be proposed as a lead molecule for amyloid dissociation. In the present study, we have taken the aqueous extract of a herb, Shankhapushpi (Convolvulus pluricaulis), and synthesized zinc oxide nanoflowers (ZnO-NFs-Skp). The plant extract was characterized using phytochemical analysis, and the ZnO-NFs-Skp were characterized using various photophysical tools like dynamic light scattering, zeta potential, XRD, FTIR, and scanning electron microscopy (SEM). The in vitro cytotoxicity of the ZnO-NFs-Skp was assessed in the PC12 cell line using an MTT assay and a fluorescent dual-staining assay. The effect of ZnO-NFs-Skp on zebrafish embryos was evaluated for in vivo biocompatibility. Finally, the amyloid degradation of the ZnO-NFs, after incubation with preformed insulin amyloids, the model amyloid protein used for the amyloid study, was evaluated at different time intervals using the Thioflavin T fluorescence assay. The results indicated that the Shankhapushpi extract had alkaloids, coumarins, and glycosides. The hydrodynamic diameter of ZnO-NF-Skp was found to be 181 nm, and the zeta potential was −17.7 mV. SEM imaging showed a carnation flower-like morphology with a petal thickness of 30 ± 5 nm. The ZnO-NFs-Skp did not induce any toxicity up to a dose of 160 μg/mL, both in vitro and in vivo. The amyloid degradation study revealed 38% degradation of the IA, 24 h after incubation at 37 °C. SEM analysis also evidenced the degradation of IA. Compared to ZnO nanoparticles (18%), ZnO-NFs-Skp could degrade almost double (35%) the amount of IA after 12 h incubation, as shown by the ThT assay. Overall, the data suggested that Shankhapushpi-mediated ZnO-NFs (ZnO-NFs-Skp) are biocompatible and have a good capacity to degrade amyloids. In the future, amyloid degradation using Aβ-42 and the prion protein needs to be investigated. Full article

Background: Accurate segmentation of brain tumors from multi-modal magnetic resonance imaging (MRI) is essential for diagnosis, treatment planning, and therapy monitoring. However, this task remains challenging due to tumor heterogeneity, irregular boundaries, and the complex anatomical structure of surrounding tissues. In particular, precise delineation of tumor sub-regions—including whole tumor, tumor core, and enhancing tumor—continues to be a major limitation of existing automated methods. Methods: In this study, we propose a novel hybrid CNN–Transformer framework that integrates local feature extraction with global contextual modeling for improved brain tumor segmentation. The architecture consists of three main components: a dual-pathway encoder for capturing fine-grained and contextual features, a multi-scale feature fusion module based on spatial pyramid pooling with dense connections, and a boundary-aware attention decoder designed to enhance segmentation accuracy around tumor edges. The model utilizes four MRI modalities (T1, T1ce, T2, and FLAIR) to capture complementary tumor characteristics. In addition, a hybrid loss function combining Dice, focal Tversky, and boundary losses is employed to address class imbalance and improve boundary precision. Results: Experimental results on the BraTS 2023 dataset demonstrate superior performance, achieving Dice scores of 92.3%, 88.7%, and 84.5% for whole tumor, tumor core, and enhancing tumor, respectively, while maintaining high computational efficiency. Conclusion: The proposed framework achieves accurate and robust brain tumor segmentation by effectively integrating local and global features, demonstrating its potential for automated multi-modal MRI analysis in clinical practice. Full article

The assessment of consciousness has been shaped largely by research on acquired disorders of consciousness after acute or chronic brain injury, but similar problems of unreliable behavioral expression increasingly arise in neurodegenerative disease. This translational overlap is especially relevant when preserved cognition, awareness, or intentionality cannot be reliably expressed because of severe motor impairment, fluctuating arousal, cognitive decline, aphasia, apraxia, or impaired cooperation. In neurodegenerative disease, degeneration of arousal systems, large-scale brain networks, cognition, and motor pathways may similarly make observable behavior an unreliable measure of awareness. The challenge is not only to determine if a patient responds, but also to ask if residual awareness, intentionality, or covert cognition can still be detected through physiological signals. This review discusses how contemporary modalities reshape this assessment. Electroencephalography has moved from a descriptive measure of background activity to a bedside tool capable of probing event-related responses, network organization, and cortical complexity. Magnetic resonance methods reveal altered connectivity within thalamocortical and default mode network systems, while functional near-infrared spectroscopy adds a portable hemodynamic approach that may be repeated at the bedside and integrated with active paradigms. Brain–computer interfaces provide a translational step by converting neural responses into evidence of command following or, in selected patients, into communication, and artificial intelligence strengthens these approaches by extracting clinically meaningful patterns from complex neural and hemodynamic data. Additionally, autonomic measures, including heart rate variability and baroreflex indices, are considered as auxiliary physiological context for arousal and engagement, and not as direct markers of awareness. Because the most mature evidence for covert awareness and cognitive-motor dissociation comes from acquired disorders of consciousness, this review treats brain injury literature as a methodological foundation instead of as directly interchangeable evidence for neurodegenerative disease. It then examines how these approaches may be adapted to neurodegenerative contexts, especially ALS, severe dementia, Lewy body disease with fluctuating cognition, and conditions in which communication or motor output becomes unreliable. Full article

Gingerol-rich extract derived from Zingiber officinale Roscoe (ZOGE) is clinically effective against motion sickness (MS), yet its mechanism remains unknown. Here, we demonstrate that ZOGE confers protection by modulating a specific gut–brain immune–neuroendocrine pathway. Using a rotation-induced rat model combined with behavioral tests, vestibular nuclei (VN) metabolomics, and molecular analyses, we found that ZOGE not only alleviated MS symptoms and normalized VN metabolic disturbances, particularly in phenylalanine, tyrosine, and tryptophan biosynthesis, but also potently suppressed the peripheral IL-33/ST2/PLC-γ1/TRPA1 signaling axis in the ileum, leading to reduced synthesis and release of 5-HT, a key MS mediator. Our study provides the evidence that ZOGE acts through coordinated central metabolic modulation and peripheral inhibition of a defined pro-emetic pathway, establishing a novel gut–brain immune–neuroendocrine mechanism for its therapeutic efficacy against MS. Full article

Pulmonaria rubra (Boraginaceae) is a widely distributed plant in Bulgaria, yet its phytochemical profile and therapeutic potential have remained unexplored. P. rubra methanol extract (PRE) was evaluated through phytochemical profiling and in vitro neuroprotective and antioxidant assays. Rat brain synaptosomes, mitochondria and microsomes were treated with PRE alone, and in combination with 6-hydroxydopamine and tert-butyl hydroperoxide as toxic agents. The extract exhibited concentration-dependent protective effects in all subcellular models. Additionally, it was tested on hMAOA/B and different isoforms of CYP450 enzymes, but it did not show any activity in the tested conditions. In the UHPLC-HRESIMS analysis, 26 secondary metabolites were identified, mainly hydroxycinnamic acids and caffeoyl oligomers, flavonoids, a lignan (globoidnan A), and the terpenoid glycoside roseoside. Seven compounds were identified via UHPLC-UV method using reference compounds: rosmarinic acid, rutin, quercetin-3-O-glucoside, astragalin, apigenin-7-O-glucoside, apigenin-7-O-glucuronide and alcesefoliside, with the latter two being reported for the first time in genus Pulmonaria. The quantity of rosmarinic acid in PRE was 4.35%, distinguishing the compound as the main bioactive molecule in the species. Characterized by its high content of rosmarinic acid, P. rubra represents a highly viable candidate for subsequent development into standardized phytopharmaceuticals targeting oxidative stress-related neurodegenerative diseases. Full article

Fixed-bottom or floating offshore structures are the foundations, platforms, and associated infrastructure that allow for oil and gas production systems, offshore wind turbines, and cabling. The remote nature of these structures and the harsh environment with high variability in wind, waves, currents, and weather make construction activity very difficult and unpredictable; the cost of variation in the schedule can lead to high construction vessel and personnel costs. The adoption of artificial intelligence using trends observed in historical data can help achieve more accurate construction costs and schedule predictions, reducing the capital expenditure cost of installation. A resource-significant activity, sometimes called a resource-critical activity or high-resource-demand activity, is an activity on a construction or project schedule that consumes a disproportionately large share of one or more resources compared with others. Plant Design Modelling (PDM) is a digital process that creates and manages a detailed 3D model of a building’s physical and functional characteristics and semantic information, such as cost and schedule. PDM serves as a single source of truth for multidisciplinary activities and, therefore, serves as a rich data source for various construction applications, including project scheduling and cost estimation. Neural networks (NNs), a subset of machine learning algorithms inspired by the human brain, excel at identifying patterns in complex datasets and making predictions, such as forecasting costs based on non-linear relationships and historical trends. Data from an offshore structure modification project were extracted from Aveva’s Everything PDM, focusing on installation activities to create a dataset for machine learning model training. The structured data extracted exhibit non-linear patterns; therefore, linear, regularised linear, robust linear, and the ensemble (tree-based) models and supervised neural network models with varied architecture and hyperparameter values were evaluated and compared. The best performance was obtained using the deep-optimised ANN model. The result obtained is consistent with previous studies. The neural network models show a superior ability to predict the non-linear nature of offshore construction activities’ time. Full article

Decoding mental states from non-invasive neural recordings is central to brain-computer interface research. Multimodal acquisition that combines electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) couples the high temporal resolution of EEG with the spatial specificity of fNIRS, compensating for the individual limitations of each modality. While such hybrid systems achieve strong intra-subject performance, cross-subject generalization remains constrained by inter-individual variability in neural responses. This study introduces the Eigenvalue-Guided Positional Encoding Classification Transformer (EPECT), an architecture that integrates eigenvalue-aware multi-head self-attention with sinusoidal positional encoding to capture both the spectral structure of the learned feature representations and the temporal ordering of multimodal sequences. Stacked one-dimensional convolutions extract local patterns prior to transformer encoding, and global average pooling aggregates the final representation for classification. EPECT was evaluated on two publicly available EEG-fNIRS datasets covering motor imagery (MI), n-back, discrimination/selection response (DSR), and word generation (WG) paradigms under a cross-subject protocol. The model achieved classification accuracies of 97.3%, 96.3%, 98.1%, and 97.9% on the MI, n-back, DSR, and WG tasks, respectively. Ablation studies quantified the contribution of each architectural component, and integrated gradients analysis revealed structured modality-specific attribution patterns aligned with task-relevant cortical regions. Additional experiments with synthetic cortical perturbations demonstrate the sensitivity of EPECT to subtle activity changes, indicating potential utility for tracking neurorehabilitation outcomes in future clinical applications. Full article

Background: The endocannabinoid system (ECS) and the Kisspeptin system (KS) play crucial roles in the central and peripheral regulation of male reproduction. The KS comprises Kisspeptins, the cleavage product of the Kiss1 protein, and its receptor Kiss1R; it is a critical central regulatory factor of the Gonadotropin Releasing Hormone (GnRH), but its role in the testis in sustaining spermatogenesis is not fully understood. Similarly, in addition to the brain, the ECS is widely expressed in the testis, where it regulates spermatogenesis, steroidogenesis, and the production of high-quality gametes. Since the possible crosstalk between KS and ECS at the gonadal level is poorly understood, this study investigates the possible mutual modulation between ECS and KS in rat testis. Methods: Experiment 1: Testis pieces collected from adult rats were treated ex vivo for 1 h with the endocannabinoid anandamide (AEA, 10⁻⁸ M) ± SR141716A (10⁻⁷ M, a cannabinoid receptor (CB) 1 antagonist), or with SR141716A alone. Experiment 2: Testis pieces were treated for 4 h with decreasing doses of Kisspeptin-10 (Kp10, 10⁻⁶–10⁻⁹ M) ± Kp234 (a Kiss1R antagonist). Proteins extracted from the treated tissues were analyzed by Western blot for Kiss1R, Kiss1, CB1, CB2, AEA-hydrolyzing enzyme Fatty Acid Amide Hydrolase (FAAH), and AEA-biosynthetic enzyme N-acylphosphatidylethanolamine-specific phospholipase D (NAPE-PLD) proteins. Results: AEA treatment, via CB1, reduced Kiss1R protein in testis. Kp10 treatment increased the expression of CBs and NAPE-PLD at all doses and increased FAAH at 10⁻⁹ M dose only. Pre-incubation with Kp234 abolished Kp10 effects on CB1, NAPE-PLD, and FAAH, suggesting a direct Kp10-dependent modulation; on the other hand, pre-incubation with Kp234 did not abolish Kp10’s effects on CB2, suggesting an indirect action of Kp10 on CB2. Conclusions: Mutual modulation between ECS and KS exists in the testis: AEA, via CB1, suppresses Kisspeptin signaling, while Kisspeptin regulates the ECS through both Kiss1R-dependent and independent mechanisms. These local interactions identify new potential mechanisms in the intratesticular communications sustaining spermatogenesis via ECS and suggest that KS might be a new therapeutic target to rescue ECS impairment in male reproductive dysfunction. Full article

Background/Objectives: Major Depressive Disorder is a serious mental disorder that negatively affects an individual’s health and quality of life. The diagnosis of this disease is based on clinical interviews, questionnaires, and the patient’s self-reports. The objective of this study is to develop a biological diagnostic system based on the analysis of EEG signals and brain regions, rather than relying on self-reports. Methods: In this study, the EEG signals in the Multimodal Open Mental Disorder Analysis (MODMA) dataset were divided into six anatomical regions: prefrontal, frontal, central, parietal, temporal, and occipital. Empirical Wavelet Transform and Empirical Mode Decomposition methods were applied separately to the channels in each region, resulting in three IMF components. A total of 23 features, including statistical, nonlinear, spectral, and model-based (AR) features, were extracted from each IMF component. In addition to these features, asymmetry features between the left and right hemispheres were also included. Feature dimensions ranging from 10 to 40 were selected via the mRMR method, and the extracted feature sets were classified using SVM, k-NN, RUSBoost, Random Forest, and Meta-Ensemble machine learning models with Leave-One-Subject-Out (LOSO) validation. Results: According to the analysis results, the highest accuracy rate in Major Depressive Disorder (MDD) diagnosis was achieved by classifying features extracted from the frontal and prefrontal regions. The EMD signal processing method demonstrated superior performance compared to the EWT method. An accuracy rate of 98.11% was achieved using Random Forest and Meta-Ensemble models. Conclusions: In the proposed method, Explainable Artificial Intelligence (XAI) based SHAP analysis was applied to provide reliable and interpretable features for MDD diagnosis based on brain regional analysis. Full article

Understanding the neural dynamics underlying expressive musical performance remains a major challenge at the intersection of neuroscience, music cognition, and computational modeling. While Electroencephalogram (EEG) studies of emotion have largely focused on passive exposure to affective stimuli, comparatively little research has examined oscillatory brain activity during active musical expression. The present single-subject study investigated whether band-limited EEG activity recorded during expressive piano performance by a professional concert pianist contains sufficient discriminative structure to support supervised multi-class classification of musically defined emotional categories. EEG was recorded from 128 scalp sites while a professional concert pianist performed emotionally characterized excerpts from Bach, Beethoven, and Chopin in a continuous naturalistic session. Musical excerpts had been previously categorized and perceptually validated according to emotional valence, tempo, energy/arousal, and tonal structure. From the continuous EEG recording, 180 non-overlapping 2 s artifact-free segments were extracted, yielding 30 segments for each emotional category. Mean spectral power was computed within theta (3.5–7.5 Hz), alpha (7.5–12.5 Hz), and high-beta (24–30 Hz) frequency bands across selected centro-parietal and posterior electrodes, resulting in 24 EEG-derived features per segment. Linear Support Vector Machine, Random Forest, and Gradient Boosting classifiers were evaluated using an 80/20 train-test split combined with five-fold cross-validation. EEG-only classification achieved above-chance performance across models, with Random Forest yielding the highest accuracy (0.42), macro F1-score (0.414), and Cohen’s κ (0.30), exceeding the theoretical chance level of 0.167. Feature importance analysis revealed distributed contributions across theta, alpha, and high-beta oscillatory activity, particularly over parietal and occipital regions, without evidence for a single dominant neural marker. Inclusion of an additional binary arousal-related feature substantially improved Random Forest performance (accuracy = 0.58; macro F1 = 0.579; κ = 0.50), indicating that arousal organization contributed strongly to category separability within the classification framework. These findings suggest that oscillatory EEG activity accompanying expressive musical action contains measurable statistical structure associated with emotionally differentiated performance states. Rather than identifying discrete neural correlates of emotion, the present results provide a computational characterization of distributed oscillatory dynamics emerging during expressive motor-acoustic interaction, extending affective EEG research beyond passive perception paradigms toward ecologically grounded musical performance contexts. Full article

The diagnosis and treatment planning for brain tumors remain a complex task in medical imaging, largely due to the intricate structure of such abnormalities. This study introduces an interpretable stacked deep learning framework consisting of three sequential stages: (i) tumor segmentation, (ii) feature extraction, and (iii) tumor classification. The segmentation stage introduces a three-parameter lambda distribution (TPLD), a symmetric special case of generalized lambda distribution (GLD), used as a statistical intensity prior that is fused into the gating signal of an Attention U-Net for enhancing boundary delineation. The segmented outputs are processed using InceptionV3 for deep feature extraction and followed by a convolutional neural network (CNN) classifier. We evaluated the proposed model on the BRISC 2025 dataset, consisting of T1 weighted brain MRI images with pixel wise segmentation masks, which is validated by medical experts. The dataset consists of 3933 training images and 860 test images with ground truth masks, containing the four classes: meningioma, glioma, no tumor, and pituitary tumor. We utilized a region-of-interest–based training strategy to reduce the computational complexity and minimize overfitting. The data split followed the official image-level partition distributed with BRISC 2025; because patient identifiers are not released with the dataset, patient-level separation could not be independently verified, and this is acknowledged as a limitation. To ensure methodological transparency and clinical robustness, we systematically report the learning dynamics across 20, 60, and 100 training epochs at multiple decision thresholds (0.50, 0.60, 0.70), providing evidence of stable model convergence without overfitting. We also introduce a composite loss function by integrating cross-entropy, focal losses, and Dice to further boost performance. Experimental results demonstrate 97.8% classification accuracy on the test set, 92.4% Dice coefficient, and 85.9% IoU at the optimal threshold of 0.60. An ablation study further confirms the contribution of each loss component, supporting reproducibility and transparency in model evaluation. These findings confirm the practical utility and reliability of the proposed framework in the context of brain tumor segmentation and clinical diagnosis. Full article

(1) Background: Clinical trial data extraction from registries such as ClinicalTrials.gov remains labor-intensive and error-prone, often missing critical details hidden in unstructured protocol descriptions. Large Language Models (LLMs) offer potential to automate this process, yet systematic multi-model comparisons on real clinical trial data remain scarce. (2) Methods: Four LLMs (OpenAI o4-mini-high, Anthropic Claude-Sonnet-4, Google Gemini 2.5-Pro, and Meta Llama-4-Maverick) extracted brain stimulation parameters from 67 transcranial direct current stimulation (tDCS) trials in Parkinson’s disease via a structured JSON schema. Pairwise inter-model agreement was quantified with Cohen’s Kappa and percentage agreement across binary, categorical, and multi-component task tiers. (3) Results: Under exact-string matching, agreement was near-perfect for binary classifications (non-invasive classification: 100%; brain stimulation presence: 99.3%, $\kappa$ = 0.50) and substantial for categorical extractions (primary stimulation type: 96.4%, $\kappa$ = 0.70), but fell to 48.6% ($\kappa$ = 0.43) for complex anatomical targets. Numeric parameters revealed model-specific strengths: o4-mini-high and Claude-Sonnet-4 achieved perfect duration agreement (r = 1.000, n = 19) while Llama-4-Maverick diverged substantially (r < 0.12). Validation against an expert gold standard (100% inter-annotator agreement on a 20-trial overlap) confirmed high extraction accuracy across all features (mean 93.7–98.9%). Crucially, the low agreement on anatomical targets proved to be an artifact of exact-string scoring: under the same semantic matching used to measure accuracy, inter-model agreement rose to 97.0%, coinciding with the 95.5% expert accuracy. Inter-model agreement therefore tracks accuracy once both are measured on a common basis. (4) Conclusions: Exact-string inter-model agreement decreases with task complexity, but this decline largely reflects interchangeable free-text wording rather than reduced accuracy. Evaluated semantically, agreement and expert accuracy are both high and closely aligned. A residual risk is not low accuracy but the rare error shared across all models, which agreement cannot detect, and which overall accuracy can itself mask when one class dominates. These findings inform hybrid human–AI systematic review pipelines in which targeted expert oversight focuses on shared-error and minority-class detection. Full article