Search Results (1,298)

Valproic acid (VPA) is a human developmental toxicant that causes neural tube defects and neurobehavioral deficits. Recent work has implicated VPA-induced oxidative stress in cell models of neurodifferentiation, where oxidative post-translational modifications (PTMs) in undifferentiated cells, primarily protein sulfenylation (Pr-SOH), were unique compared to differentiated neurons, primarily protein S-glutathionylation (Pr-SSG). Many of these effects could be mitigated by pretreatments with an Nrf2 inducer. However, it is unclear how early-stage mouse embryos (gestational day 8.5) respond to VPA treatments. Using whole embryo culture, mouse embryos were treated with VPA. A time course assessment of glutathione/glutathione disulfide redox potentials was performed via HPLC throughout 24 h of culture. At 6 h of VPA treatment, embryos were collected for the assessment of protein redox states and specific protein PTMs via various blotting techniques. Also, at 6 h of treatment, the localization of specific PTMs was determined via whole mount staining. Some embryos were pretreated with an Nrf2 inducer. Our data demonstrated that VPA caused a sharp oxidation of redox potentials, which were the greatest between 2 and 6 h, but reverted to control levels by 24 h. Preemptive Nrf2 activation prevented VPA-induced oxidation. Redox blotting showed that VPA caused oxidation of the proteome but this could be reversed by D3T pretreatment. More specifically, Pr-SOH levels increased but Pr-SSG levels were unchanged. Increased Pr-SOH could also be reversed with prior Nrf2 activation. We conclude that embryos at these early stages of development are highly sensitive to VPA and respond more like undifferentiated cells, promoting a more pro-oxidizing outcome for proteins, increasing Pr-SOH formation vs. Pr-SSG. These findings may support specific windows of development where embryos are more susceptible to VPA-induced oxidative injury. Further understanding of redox control and regulation at these susceptible states may serve to develop preventative strategies to reduce poor developmental outcomes after exposures. Full article

Respiratory diseases are a major global health challenge. However, identification of respiratory diseases is often limited by subjectivity, environmental noise and inter-clinician variability. This study presents an explainable multimodal deep learning framework for recording-level multiclass classification of respiratory audio signals. The proposed system integrates two complementary representations—a spectro-temporal encoder based on a CNN–BiLSTM-attention architecture and a handcrafted acoustic-feature encoder capturing acoustic descriptors commonly used in respiratory-audio analysis, including MFCCs, zero-crossing rate, spectral centroid, spectral bandwidth, chroma, RMS energy, and spectral rolloff features. These branches are combined through late-stage fusion to leverage both data-driven representation learning and domain-informed acoustic cues. The proposed model was trained and internally evaluated on the Asthma Detection Dataset Version 2, comprising five respiratory categories: bronchial disease, asthma, COPD, healthy, and pneumonia. Mono conversion, resampling to 16 kHz, 100–2000 Hz band-pass filtering, amplitude normalisation, fixed 4 s trimming or zero-padding, training-only augmentation, handcrafted-feature extraction, mel-spectrogram generation, quality control auditing, and stratified recording-level partitioning have been applied in the pre-processing steps. Across five repeated experiments with different random seeds, the proposed hybrid model achieved a mean held-out recording-level test accuracy of $0.9099\pm 0.0163$, balanced accuracy of $0.8936\pm 0.0152$, macro F1-score of $0.8937\pm 0.0177$, macro ROC–AUC of $0.9867\pm 0.0010$, and macro PR–AUC of $0.9489\pm 0.0044$. Conventional machine learning baseline comparisons showed that the proposed model achieved stronger internal accuracy, balanced accuracy, macro recall, macro F1-score, and macro ROC–AUC than classical machine learning algorithms trained on handcrafted acoustic features, although Random Forest remained competitive in macro PR–AUC. Ablation analysis shows that the deep spectro-temporal branch was the primary contributor to predictive performance, while the handcrafted branch provided complementary interpretable acoustic information rather than consistently improving all classification metrics. Explainability was incorporated using Grad-CAM and Integrated Gradients for spectrogram-based interpretation and SHAP for handcrafted-feature attribution. Domain-shift evaluation on the ICBHI Respiratory Sound Database and a COPD-focused cohort revealed substantial dataset shift effects, including poor healthy-case recognition on ICBHI and seed-dependent COPD recognition in the COPD-focused cohort. Identifier-aware sensitivity analyses showed lower performance than the main recording-level split, suggesting that subject-like or source-level overlap may inflate internal performance estimates. The findings should be interpreted as promising internal held-out recording-level algorithmic performance with limited external transfer, rather than evidence of readiness for clinical use. Full article

Accurate prediction of effluent total nitrogen (TN) is important for early exceedance warning and operational control in wastewater treatment plants. Existing decomposition-based models may overestimate performance when full-series decomposition is performed before data splitting, causing potential temporal information leakage. To address this issue, this study compares noncausal and strictly causal Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise combined with Long Short-Term Memory (ICEEMDAN-LSTM) and Variational Mode Decomposition–Long Short-Term Memory network (VMD-LSTM) settings, and proposes a Level–Trend Persistence-Residual LSTM (LT-PR-LSTM) for univariate effluent TN prediction. The model uses Persistence as the short-term state baseline, extracts level features from historical TN, and introduces first- and second-order differences to learn residual corrections relative to the current state. Multi-model comparison, ablation experiments, stability tests, SHapley Additive exPlanations (SHAP) interpretation, supplementary dataset validation, and efficiency analysis were conducted. Results show that noncausal decomposition inflates predictive performance. LT-PR-LSTM achieves the best main-test performance, with RMSE 1.1273, MAE 0.6082, MAPE 7.5455%, and $R^2$ 0.8512, reducing RMSE, MAE, and MAPE by 6.73%, 7.64%, and 8.56% compared with Persistence. SHAP identifies $TN(t-2h)$ as the dominant predictor, and the model requires only 0.5348 ms/sample, indicating potential for online TN early warning. Full article

This study, based on phase equilibrium theory, uses reservoir crude oil systems as the research object and adopts the Perturbed Chain-Statistical Associating Fluid Theory (PC-SAFT) equation of state. By combining the Panuganti characterization method with the three-phase Rachford–Rice algorithm, an integrated RRPC-SAFT engineering workflow is established, which effectively addresses the drawbacks of traditional PC-SAFT models, including low computational efficiency and poor convergence under extreme working conditions. On this basis, systematic performance comparisons are conducted between the RRPC-SAFT workflow and classical cubic equations of state (PR and SRK). Furthermore, the asphaltene phase behavior under gas injection development conditions is simulated, and the quantitative effects of the four SARA fractions on the critical precipitation conditions and precipitation intensity of asphaltenes are determined, clarifying the evolution rules and main controlling factors of asphaltene phase instability under various development scenarios. The research results reveal that the average relative errors of bubble point pressure and asphaltene onset precipitation pressure (AOP) for the three crude oil samples are all less than or equal to 5%. Compared with the PR and SRK models, the average prediction errors are reduced by 1.27% and 2.01%, respectively. Gas injection simulation results demonstrate that nitrogen poses the highest risk of triggering asphaltene precipitation under equimolar injection, with the asphaltene onset precipitation pressure increasing up to 114.94%. Single-factor analysis of SARA fractions verifies that saturates and asphaltenes aggravate precipitation, while aromatics and resins suppress asphaltene destabilization. In terms of computational efficiency, the computational speed of the RRPC-SAFT algorithm is four times higher than that of the traditional Gibbs free energy minimization algorithm. This model can be applied to calculate the thermodynamic critical equilibrium conditions of asphaltene precipitation, providing a thermodynamic basis for early screening of asphaltene deposition risks, optimization of gas injection schemes, and design of deposition prevention and control technologies. Full article

Voltage source converters (VSCs), together with their inner current and outer power/voltage control loops, are fundamental building blocks in the modern, converter-dominated power systems, particularly within high-voltage DC (HVDC) frameworks. Selecting effective control methods for VSCs is essential to ensure the stability, power quality, and dynamic performance of HVDC grids. This paper seeks to advance the current body of research by delivering an in-depth, consistent, unified framework and systematic examination of VSC control architectures within HVDC networks. It thoroughly explores various control strategies for VSCs interfacing with HVDC grids, such as grid-following and grid-forming strategies, with particular emphasis on both stationary ($\alpha \beta$) and synchronous (dq) reference frames. Moreover, the paper provides a comprehensive analysis of the theoretical underpinnings and decoupled control strategies, like the feedforward decoupling of the d- and q-axis currents in the dq frame and the inherently decoupled structure of the αβ frame. Additionally, advanced filtering techniques, including Moving Average Filter (MAF), Cascaded Delayed Signal Cancellation (DSC), and LCL filters, are analyzed. In addition, harmonic mitigation strategies, like parallel/multiple resonant (PR) terms in the αβ frame and cascaded notch filters in the dq frame, are presented. Furthermore, precise power control approaches and synchronization methods are discussed in detail. Also, this paper presents a detailed comparison of the performance characteristics of phase-locked loop (PLL) and frequency-locked loop (FLL) in response to grid frequency variations. Moreover, this paper proposes circuit representations and VSC models in both synchronous and stationary reference frames. The simulation results corroborate the theoretical insights discussed in the paper under various operational conditions, including initial responses, grid disturbances, three-phase-to-ground temporary fault scenarios, harmonic distortions, and load imbalances, in terms of overshoot, settling time, active- and reactive-power fluctuation reduction, voltage unbalance factor, total harmonic distortion, and post-fault convergence time, all evaluated in accordance with the limits defined in EN-50160. This comprehensive comparison of the presented control strategies facilitates researchers in identifying the most appropriate controller depending on their specific application requirements. Full article

Background: Cerebral aneurysms and aneurysmal subarachnoid hemorrhage (aSAH) may induce cardiac electrical instability through autonomic dysregulation and an exaggerated neurohumoral stress response. Electrocardiographic (ECG) abnormalities, including QT/QTc prolongation, QTc dispersion, and P-wave dispersion, are recognized markers of ventricular repolarization heterogeneity and atrial conduction abnormalities associated with arrhythmogenic risk. However, data regarding the reversibility of these electrophysiological alterations following definitive aneurysm treatment remain limited. Methods: This retrospective, single-center study included 39 patients with cerebral aneurysms who underwent microsurgical clipping between January 2025 and May 2026 and 35 age- and sex-matched healthy controls. Standard 12-lead ECGs were evaluated at baseline (preoperative) and one month after surgery in the aneurysm group. QT interval, corrected QT (QTc) interval, QTc dispersion, and P-wave dispersion were assessed using standardized methods. Baseline transthoracic echocardiographic parameters, including left ventricular ejection fraction and left atrial diameter, were evaluated to minimize potential confounding related to structural cardiac abnormalities. Between-group and within-group comparisons were performed using appropriate statistical analyses. Results: Baseline demographic and echocardiographic characteristics were comparable between the aneurysm and control groups. Patients with cerebral aneurysms demonstrated significantly higher baseline QT interval, QTc interval, QTc dispersion, and P-wave dispersion compared with healthy controls. Following microsurgical treatment, significant reductions in QT interval, QTc interval, QTc dispersion, and P-wave dispersion were observed at one month compared with preoperative values, whereas PR interval and QRS duration remained unchanged. These findings suggest a partial normalization of cardiac electrical heterogeneity after definitive aneurysm treatment. Conclusions: Cerebral aneurysms are associated with increased ventricular repolarization and atrial conduction heterogeneity, reflecting autonomic-mediated cardiac electrical instability. The significant reduction in QTc dispersion and P-wave dispersion following microsurgical treatment suggests that these electrophysiological abnormalities may be at least partially reversible after aneurysm repair. ECG-derived markers such as QTc dispersion and P-wave dispersion may represent practical and non-invasive tools for monitoring cardiac electrical instability and recovery in patients with cerebral aneurysms. Full article

Background: HIV/AIDS remains a major global public health challenge, with persistent regional disparities in burden and progress toward the UNAIDS 95–95–95 targets. This study assessed temporal trends in the HIV/AIDS burden in Kazakhstan, compared them with Central Asia and global patterns, and projected trends through 2030. Methods: We conducted a population-level analysis using Global Burden of Disease 2023 data, examining age-standardized rates (per 100,000) of incidence, prevalence, mortality, disability-adjusted life years (DALYs), years of life lost (YLLs), and years lived with disability (YLDs) from 2010 to 2023. Trends were quantified using percent change and average annual percentage change, with projections based on log-linear models. Results: Between 2010 and 2023, prevalence in Kazakhstan increased by 332.1% and incidence by 111.0%, contrasting with the decline in global incidence (−24.7%). Mortality decreased (−32.7%), along with DALYs (−28.8%) and YLLs (−37.1%), while YLDs increased by 135.5%, indicating a shift toward a chronic disease burden. In 2023, Kazakhstan had a lower overall burden than global estimates but showed steeper increases in incidence and prevalence. Age-specific analyses indicated the largest increases among adults aged 30–69 years. Under current trend assumptions, projections suggest continued growth in prevalence and incidence, with modest mortality declines through 2030, though these trajectories do not account for future changes in prevention coverage, treatment access, or policy. Conclusions: Kazakhstan is undergoing a transition toward a chronic HIV epidemic, underscoring the need to strengthen prevention, expand PrEP and testing coverage, and address structural barriers to achieve epidemic control. Full article

Endometriosis is a long-term gynecological condition marked by the growth of endometrial-like tissue outside the uterus, which undergoes proliferation, bleeding, and regeneration. This disease is associated with disrupted steroid hormone signaling, notably progesterone (P4) resistance and estradiol (E2) dominance. P4 resistance has been associated with impaired activation of the progesterone receptor (PR) and reduced transcription of P4 target genes, while elevated E2 levels induce estrogen receptor (ER)-mediated signaling, enhancing estrogen-dependent lesion growth. This hormonal imbalance contributes to a pro-inflammatory microenvironment, chronic pelvic pain, infertility, and enhanced neuroangiogenesis. Emerging evidence indicates that the coordinated regulation of neurotrophins and sex hormones promotes nerve fibers and blood vessel growth and invasion within endometriotic lesions. P4 and E2 have been shown to modulate the expression of key neurotrophins, including nerve growth factor (NGF) and brain-derived neurotrophic factor (BDNF). This review presents current evidence on the interplay between neurotrophins and ovarian steroids in endometriosis, with a specific focus on their contribution to neuroangiogenesis and pain pathophysiology. The review includes articles in English containing the Medical Subject Headings (MeSH) terms: “endometriosis”, “neurotrophins”, “nerve growth factor”, “brain-derived neurotrophic factor”, “neuroangiogenesis”, “progesterone”, and “estradiol”, found in the PubMed database published between 2000 and 24 May 2026. This review included a range of original research articles, systematic reviews, meta-analyses, prospective observational studies, case–control studies, and review papers, for a total of 122 articles. Full article

Typhoon-induced landslides in coastal mountainous regions are controlled by the coupled effects of rainfall, wind, topography, and storm-track geometry. However, conventional static susceptibility models have limited ability to represent event-scale forcing under extreme weather conditions. This study develops a physics-guided dynamic landslide susceptibility framework and retrospectively applies it to the 2017 Tropical Cyclone Cempaka event in Pacitan Regency, Indonesia, where 743 landslides were identified. The framework integrates static terrain factors, antecedent wetness, event-scale rainfall accumulation and intensity, maximum wind speed, and a typhoon geometric exposure index derived from IBTrACS best-track information that represents track proximity, topographic shielding, rainfall-favored quadrant effects, and storm-motion effects. Under spatial block cross-validation, model performance improved progressively from the static baseline to the full-factor model, with the receiver operating characteristic area under the curve (ROC-AUC) increasing from 0.648 to 0.751, the precision–recall area under the curve (PR-AUC) reaching 0.826, and the F1-score reaching 0.744. The full-factor model also reduced missed landslide cases from 328 to 205 and concentrated predicted high-susceptibility zones along the typhoon exposure corridor. Additional parameter-sensitivity analyses further indicate that the event-based E_geo setting produced positive performance increments under the event-consistent quadrant convention. These results indicate that physically meaningful typhoon-exposure information can improve the spatial discrimination and interpretability of event-scale landslide susceptibility assessment. Full article

Employee attrition is a significant organizational challenge associated with substantial financial costs and the erosion of institutional knowledge. This study presents an AI-based Management Information System (MIS) that integrates machine learning (ML) models to forecast employee turnover and support technical interpretability for HR decision-making. Using the IBM HR Analytics Dataset comprising 1480 employee records with 38 features, we implemented a rigorous preprocessing pipeline—including Synthetic Minority Over-sampling Technique (SMOTE) applied exclusively within training folds to prevent data leakage, one-hot encoding, Z-score normalization, and mean-value imputation. Four ML classifiers—Logistic Regression (LR), Random Forest (RF), Multi-Layer Perceptron (MLP), and XGBoost—were evaluated under a stratified 80/20 split with 5-fold cross-validation. XGBoost achieved the highest performance, attaining an accuracy of 87.83%, a ROC-AUC of 0.94, a PR-AUC of 0.96, and an F1-score of 93.04%, attributed to its sequential boosting mechanism and built-in L1/L2 regularization. Beyond predictive performance, the system incorporates SHapley Additive exPlanations (SHAP) to deliver feature-level transparency, enabling HR professionals to engage in proactive, informed retention interventions while retaining full decision-making authority. Within-dataset comparisons confirm that the proposed framework outperforms prior methods evaluated on the same benchmark; cross-study accuracy comparisons are reported as contextual reference only, given differences in datasets and experimental protocols. The system facilitates human oversight by positioning AI as a decision-support collaborator rather than an autonomous replacement in workforce management. Future work will address real-time deployment, controlled user studies with HR practitioners, and validation with actual organizational HR data. Full article

Banana Fusarium wilt, caused by Fusarium oxysporum f. sp. cubense tropical race 4 (Foc TR4), poses a severe threat to global banana production, and breeding resistant cultivars remains the most effective control strategy. Mutation breeding, including radiation mutagenesis and somaclonal variation, has become a primary approach for developing resistant germplasm in triploid Cavendish bananas. However, whether secondary bud-sport selection from resistant somaclonal lines inadvertently compromises original resistance mechanisms at the molecular level remains poorly understood. In this study, we generated 44 mutants from Baxi jiao via ⁶⁰Co γ-irradiation and selected five lines with distinct phenotypic variations. We also collected somaclonal variant lines GCTCV-218, GCTCV-119, GCTCV-105, their bud-sport derivatives (NK_No.1, NTH, RK_No.1), and the radiation-induced resistant mutant ‘Zhongre No.1’. Using whole-genome resequencing and transcriptome analysis, we systematically compared the genetic and transcriptomic outcomes of these breeding strategies. Radiation mutagenesis induced substantial genomic structural variations and generated novel expression patterns of defense-related genes. In contrast, while bud-sport derivatives of GCTCV-218 remained genetically similar to their parent, they exhibited significant downregulation or loss of key resistance gene expression, particularly PR-1 family members. Our findings reveal that phenotype-driven somaclonal selection can inadvertently erode original resistance mechanisms, and we recommend prioritizing radiation mutagenesis for developing banana cultivars with stable and durable resistance to Foc TR4. Full article

We described tumor response and survival in dogs with modified Adams stage 4 intranasal tumors treated with once-weekly hypofractionated radiation therapy (RT) combined with the radiosensitizer α-sulfoquinovosyl-acylpropanediol (SQAP), and compared linear and volumetric response assessments. Twenty dogs treated with intensity-modulated RT (8 Gy per fraction, once weekly) and concurrent SQAP were included in this retrospective case series. Tumor response was assessed using RECIST-like linear measurements and volumetric analysis on contrast-enhanced computed tomography. Overall survival (OS) was estimated using Kaplan–Meier analysis. Of the 20 dogs, 4 were classified as stage 4a and 16 as stage 4b. The best RECIST-like responses were complete response (CR) in 5 dogs, partial response (PR) in 12, and stable disease (SD) in 4. Volumetric responses were CR in 5 dogs, PR in 11, and SD in 5. No cases demonstrated progressive disease as the best response. The median OS for all dogs was 342 days (95% confidence interval [CI], 206–419 days). Censoring one non-tumor-related death yielded a median OS of 356 days (95% CI, 231–419 days). Exploratory analysis revealed median OS of 393 and 297 days for stage 4a and 4b dogs, respectively. Volumetric assessment appeared more sensitive for detecting tumor regrowth in selected cases. Dermatologic adverse events were limited to alopecia within the radiation field, and no complete vision loss was observed. Seizure activity was documented in eight dogs. In conclusion, once-weekly hypofractionated intensity-modulated RT combined with concurrent SQAP was associated with clinically meaningful survival outcomes in dogs with advanced intranasal tumors. However, because no radiotherapy-alone control group was available, the independent contribution of SQAP to these outcomes could not be determined. Full article

Prostate cancer (PrC) is one of the most common malignancies among men worldwide. However, the contribution of genetic variation in microRNA (miRNA) biogenesis pathway genes to PrC susceptibility remains poorly characterized in many ethnically diverse populations. We conducted a case–control study involving 532 PrC patients and 550 controls from the Volga-Ural region of Eurasia to evaluate the association of twenty-one single nucleotide polymorphisms (SNPs) with PrC risk using single-variant and polygenic approaches. Association analyses identified rs595055 in the AGO1 gene as significantly associated with PrC risk after correction for multiple testing. To evaluate the cumulative effect of genetic variation, weighted and unweighted polygenic risk scores (PRSs) were constructed. The weighted PRS was significantly associated with PrC risk (odds ratio per standard deviation increase = 1.63, 95% CI [1.43–1.85], P = 1.37 × 10⁻¹³), and demonstrated moderate discriminatory performance (AUC = 63.1%), outperforming the unweighted model. Individuals in the highest PRS quartile had approximately threefold higher odds of PrC than those in the lowest quartile. Combining the weighted PRS with prostate-specific antigen improved discrimination (AUC = 68.1%). These findings support the contribution of miRNA biogenesis pathway genes to PrC susceptibility and highlight the potential value of pathway-based polygenic risk stratification in understudied populations. Full article

The development of rapid and reproducible image analysis approaches that support genus-level pre-classification of macrofungi is important for taxonomic pre-evaluation and controlled microscopic data analysis. In this study, an advanced deep learning-based approach, namely the Attention-Enhanced Hybrid CNN–ViT Framework, was rigorously evaluated for genus-level classification, using basidiospore micrographs of five carefully selected macrofungal genera. The proposed approach integrates the ability of convolutional neural networks to identify local texture and contour patterns with the global context-modelling capability of Vision Transformer structures. The objective is to enhance the extraction of distinctive representations from microscopic spore images through feature fusion and attention mechanisms. A series of experiments was conducted on a curated dataset consisting of light microscopy images of the genera Agaricus, Hebeloma, Inocybe, Amanita, and Russula. The models were compared using a range of evaluation metrics, including accuracy, F1-score, MCC, ROC-AUC, and PR-AUC. The results showed that the InceptionV3 + ViT-B16 + Fusion configuration was the most successful hybrid model, achieving an accuracy of 0.9213 ± 0.0182, an F1-score of 0.9212 ± 0.0179, a Matthews correlation coefficient (MCC) of 0.9040 ± 0.0222, a receiver operating characteristic (ROC)-area under the curve (AUC) of 0.9896 ± 0.0069, and a precision-recall (PR)-AUC of 0.9684 ± 0.0192, respectively. The present findings demonstrate that basidiospore images can carry distinctive visual information for genus-level automated classification under controlled conditions. However, it is important to note that these results should not be interpreted as claims of species-level identification or field generalisability. This is due to the use of a single microscope-camera system, a single preparation protocol, and the absence of an independent external test set. The present study demonstrates that deep learning-based microscopic image analysis can be evaluated as a preliminary classification tool in macrofungal taxonomy. It also shows that such tools can provide a foundation for future work supported by specimen-level validation, external test sets, and different imaging protocols. Full article

Background: Chatbots are becoming increasingly valuable in clinical settings, offering rapid access to medical information, aiding documentation, and improving perioperative patient education. Their adaptability makes them promising tools for personalized perioperative risk stratification (PRS) and anesthesia planning, but their definitive role remains uncertain. We aimed to evaluate chatbot performance in PRS compared to standard clinical judgment and to assess the certainty of the evidence supporting their use. Methods: This systematic review (PROSPERO ID: CRD42025642357) followed PRISMA extended and PRISMA-S guidelines. The population was defined according to the PICO framework: we included adult surgical patients undergoing anesthesia assessment (P), evaluated with LLM-based chatbots for perioperative risk stratification and anesthesia planning (I), compared with traditional clinician assessment (C), and extracted performance metrics (O). Comprehensive searches of PubMed, MEDLINE, Scopus, Embase, Google Scholar, Open Gray, ClinicalTrials.gov, WHO ICTRP, and Cochrane Library Central were conducted through January 2026. Risk of bias and study quality were assessed using PROBAST-AI, RoB-2, and ROBINS-I. Certainty of the evidence was assessed using GRADE system. A random-effects meta-analysis of pooled chatbot accuracy was performed, with subgroup analyses by ASA status and perioperative risk stratification. A sensitivity analysis was performed with a leave-one-out exclusion test. Results: Eleven studies published between 2023 and January 2026 were included (N = 227,059 patients). Five prospective cohorts, two large retrospective cohorts, one randomized non-inferiority trial, and three non-clinical or mixed-methods studies were found. Meta-analysis showed that the pooled accuracy of LLM-based chatbots for AI–clinician concordance in perioperative risk stratification and ASA classification was 0.90 [95% CI: 0.42–0.99; 95% prediction interval 0.03–1.00]. Subgroup analyses indicated that the ASA status prediction subgroup reached a pooled accuracy of 0.91 (95% CI: 0.46 to 0.99), whereas the exploratory perioperative risk stratification subgroup showed an accuracy of 0.73 (95% CI: 0.10 to 0.98). Performance decreased with increasing patient complexity. Evidence is limited by small sample sizes, extreme sample size skew toward a single center, geographic bias, inconsistent outcome definitions and performance metrics, and incomplete reporting of adverse events. Most studies lacked prospective trial registration or robust control for confounding, and publication bias cannot be excluded. Conclusions: LLM-based chatbots show promising performance in routine perioperative risk stratification but remain unreliable in complex cases, with potential safety concerns. Given the overall very low GRADE certainty of evidence, these tools should be used as clinician-supervised decision support aids for routine ASA assessment, and should not be relied upon for autonomous use in complex cases or for general perioperative risk stratification. Other: This research received no external funding. PROSPERO ID: CRD42025642357. Full article