搜索结果：reliable

This study introduces a new 10-item, self-administered children's voice questionnaire (CVQ-10), designed to quantify children's voice handicap and assess its validity and reliability. Observational, prospective, cross-sectional study. The 10&#xa0;items that comprise the new CVQ-10 were extracted from the full CVQ. The selected items were chosen based on their Cohen's d scores. The new questionnaire was administered to 208 children (80 dysphonic and 128 nondysphonic), aged 6-17 years. In addition, the parents of these children completed the Pediatric Voice Handicap Index (pVHI) and a brief anamnesis questionnaire. Furthermore, after 2&#xa0;weeks, a subset of 30 randomly selected children (15 dysphonic and 15 nondysphonic) completed the CVQ-10 again, to examine test-retest reliability. At that time, another subset of 30 children was randomly selected to complete the full CVQ questionnaire to evaluate validity. The CVQ-10 demonstrated high reliability (Cronbach's &#x3b1; =&#xa0;0.94). Test-retest reliability yielded a Pearson correlation coefficient of r =&#xa0;0.91 (P&#xa0;< 0.001). A highly significant group difference was found between the dysphonic and nondysphonic groups of children (t[206] = 17.00, P&#xa0;< 0.001), with a calculated cutoff of&#xa0;nine&#xa0;points.&#xa0;A significant positive correlation was found between children's subjective self-evaluation on the CVQ-10 and their parents' responses on the pVHI questionnaire (r = 0.766, P&#xa0;< 0.001). The CVQ-10 is shown to be a valid and reliable instrument. Data confirms its single factor contract. This demonstrates that children are capable of providing a reliable and consistent description of their subjective voice handicap. The CVQ-10 differentiates dysphonic from nondysphonic children and can serve as a brief and easy to-use instrument for evaluating dysphonic children in clinical and research settings.

Engineered bacteria offer unique opportunities for biosensing because they combine genetically programmable biological activity with tunable surface interfaces. However, constructing living sensing systems that enable programmable recognition, reliable signal output, and robust performance in complex samples remains challenging. Here, we report an interfacially engineered bacterial sensing platform for dual-mode microRNA detection. Engineered Escherichia coli expressing enhanced green fluorescent protein (eGFP) were used as living signal carriers, and a polyphenol coating was deposited to create a stable, functionalizable interface. This coating preserved bacterial fluorescence while enabling immobilization of nucleic acid hairpin probes. Upon target recognition, catalytic hairpin assembly (CHA) was triggered, driving the programmable bridging and aggregation between bacteria and magnetic beads. This process converts molecular recognition into a magnetically separable assembly event, enabling target enrichment and background reduction. Meanwhile, enriched bacteria provide fluorescence output via intracellular eGFP, while Fe3+ released from the coating under acidic conditions generates a Prussian blue colorimetric signal. Together, these processes establish a fluorescence-colorimetric dual-mode sensing platform with limits of detection of 5.33 pM for the fluorescence mode and 14.1 pM for the colorimetric mode. In serum samples from glioma patients, the platform effectively distinguished patients from healthy controls, with dual-mode analysis showing improved discrimination performance compared to single-mode detection. This work demonstrates that interfacial engineering transforms engineered bacteria into multifunctional living sensing units and provides a practical strategy for developing reliable biosensing systems for liquid biopsy applications.

Cold object detection and classification allow reliable identification and categorization of low-temperature objects using non-contact thermal analysis, thereby complementing standard heat-focused techniques. It supports accurate monitoring of cooling behaviour, improves fault detection, quality control and remains powerful under poor lighting or visually challenging situations. However, cold object classification remains challenging when using conventional visible-light images due to the absence of discriminative temperature information. As of now, dedicated and systematic research on cold object detection and classification using thermal images are still very limited, highlighting a clear research gap. To address this limitation, this work proposes a thermal image-based cold object detection and classification framework using machine learning algorithms, designed to achieve effectiveness, reduced cost, and lower processing time. In addition, a dedicated dataset is developed by capturing time-dependent temperature variations of multiple cold object categories, enabling well ordered analysis and classification based on their thermal behaviour. Several machine learning models, including Decision Tree, Random Forest, XGBoost, and other classification algorithms, were developed and assessed using the proposed dataset. Among these models, the Random Forest classifier gives the highest classification accuracy of 99.35%, demonstrating its effectiveness in capturing temporal thermal variations for accurate cold object detection and identification. This work establishes a new research direction in thermal image analysis by shifting the focus from heat anomaly classification toward reliable cold object classification.

This study utilized the Surveillance, Epidemiology, and End Results (SEER) database to identify clinicopathological prognostic factors and develop a prognostic nomogram for predicting 1-, 3-, and 5-year overall survival (OS) in patients with stage II-IV epithelial ovarian cancer (EOC), aiming to enhance survival prediction accuracy and inform personalized therapeutic strategies. Clinical data of patients diagnosed between 2004 and 2020 were extracted and randomly divided into training, tuning, and internal validation cohorts at a 7:2:1 ratio. Univariate and multivariate Cox regression analyses identified age, tumor differentiation grade, AJCC stage, tumor size, number of positive lymph nodes, number of lymph nodes examined, surgery, chemotherapy, sequence of systemic therapy and surgery, and time from diagnosis to treatment as independent prognostic factors for OS (all P&#x2009;<&#x2009;0.05). A prognostic nomogram was subsequently developed and externally validated using an independent cohort of 115 EOC patients from Chengde Medical University (2014-2020). The model exhibited acceptable discriminative performance, with concordance indexes (C-index) of 0.693, 0.683, and 0.700 for the training, internal validation, and external validation cohorts, respectively. Calibration curves, receiver operating characteristic (ROC) curves, and area under the curve (AUC) values all exceeded 0.7, indicating reliable prediction of 1-, 3-, and 5-year survival rates. Kaplan-Meier curves revealed significant survival differences across risk groups, and decision curve analysis (DCA) confirmed the clinical utility of the nomogram. In conclusion, this externally validated nomogram can assist in predicting OS in patients with stage II-IV EOC, demonstrating moderate yet clinically useful performance with a C-index of 0.700. It addresses the need for personalized prognostic assessment and aids in tailoring therapeutic strategies, thereby potentially improving patient outcomes.

Affective state investigation requires precise measurement that satisfies parametric statistical assumptions for reliable and valid cross-data comparisons. The psychometric properties of the 20-item Positive and Negative Affect Schedule (PANAS) are satisfactory, but provide scores on an ordinal scale, which could be unsuitable for parametric statistics. The present study employed the Rasch model to assess the psychometric statistics of the PANAS, enhancing the scale's precision using community samples from four countries. I analysed responses from a randomly selected sample of 1000 individuals (250 from each country) out of a total sample of 1822 recruited from Germany (475), Ghana (523), India (411), and New Zealand (413). The analyses indicated that both the positive affect and negative affect subscales demonstrated satisfactory model fit after applying the testlet creation approach. Each subscale reflected a clear, single underlying construct with strong reliability and structural validity. Both scales also functioned equivalently across demographic groups, suggesting that the items measured affective states consistently, regardless of participants' sociodemographic backgrounds. The scales further exhibited strong convergent and discriminant validity. The study developed an algorithm to convert ordinal scores to interval data, enhancing precision and validity in parametric analyses.

Weakly supervised video anomaly detection (WSVAD) is fundamentally constrained by the absence of frame-level annotations, which leads to noisy instance selection in Multiple Instance Learning (MIL) and weak correspondence between temporal video segments and semantic descriptions. Vision-language models address this by enabling cross-modal alignment between visual features and textual labels, but when these representations are learned in Euclidean space, they struggle to capture subtle semantic variations and often produce ambiguous instance ranking under weak supervision. To address this limitation, we propose PoinCLIP-VAD, a vision-language framework that performs cross-modal fusion in hyperbolic space. The model embeds visual and textual features into a shared Poincar&#xe9; ball geometry, where non-linear distance scaling provides a more expressive representation of latent semantic relationships induced by cross-modal interactions, without relying on predefined hierarchical structures. This geometry-consistent formulation enables more reliable similarity estimation and better preserves distinctions between normal and anomalous patterns. The framework adopts a dual-block architecture consisting of a classification block for coarse anomaly scoring and a video-text alignment block for fine-grained correspondence using negative Poincar&#xe9; distance. Extensive experiments on benchmark datasets demonstrate that PoinCLIP-VAD achieves an AUC of 90.62% on UCF-Crime and an AP of 86.93% on XD-Violence, confirming improved anomaly discrimination and more consistent cross-modal alignment under weak supervision.

Sudden cardiac death after myocardial infarction (MI) remains a major clinical problem, partly driven by complex electromechanical feedback mechanisms that are not fully understood. In particular, the interplay between scar stiffness, border zone (BZ) remodeling, mechano-electric feedback (MEF) through stretch-activated channels (SACs), and cross-bridge formation, can strongly influence ventricular function and arrhythmogenic risk. In this study, we developed a 3D finite element model of human ventricular electromechanics coupled with a closed-loop circulation model to investigate the impact of scar size on cardiac function. Before addressing scar effects, we first analyzed the spatial resolution of the mesh to ensure reliable electromechanical predictions. Within the present electromechanical framework and mechanical discretization, spatial resolutions of approximately 1.0-1.3 mm were required to capture SAC-driven depolarizations in the BZ. Using this confirmed to be suitable resolution, we examined infarct volume fractions (IVFs) from 5-20% of the left ventricle (LV). Increasing IVF progressively impaired contractility and ejection fraction, with a nonlinear threshold emerging between 10% and 15% IVF in the primary models studied here, above which SACs elicited premature activations which disrupted sinus-driven dynamics, and produced irregular multi-beat PV-loop patterns. These findings highlight two key insights: spatial resolution is critical for capturing SAC-mediated feedback accurately, and scar extent strongly influences the severity of this feedback, with larger scars significantly altering ventricular dynamics. Together, this work provides mechanistic understanding of how scar enlargement can accelerate heart failure progression and arrhythmic risk through nonlinear MEF.

Self-paced reading (SPR) is widely used to investigate real-time sentence processing. In SPR, sentences can be presented either cumulatively, with previously presented words remaining visible, or non-cumulatively, with previous words disappearing. However, most prior research has avoided cumulative presentation, largely due to concerns that it allows readers to reveal multiple words through rapid key presses and then read them, thereby undermining the interpretability of reading times. As a result, cumulative SPR is widely assumed to be unsuitable for research on real-time sentence processing. The present study examines this assumption by comparing three cumulative SPR variants-ahead-visible cumulative SPR (AVC-SPR), non-ahead-visible cumulative SPR (NAVC-SPR), and partially cumulative SPR (PC-SPR)-with standard non-cumulative SPR (NC-SPR). In AVC-SPR, the positions of upcoming words are visually indicated; in NAVC-SPR, upcoming positions are not indicated; and in PC-SPR, upcoming positions are likewise not indicated, and accumulation is capped so that only a limited number of words remain&#xa0;visible. The four tasks were compared in terms of their sensitivity to detecting garden-path and number-mismatch effects. Clear effects were observed in all four tasks, with NAVC-SPR yielding the largest effect sizes. Power analyses further indicated that NAVC-SPR generally offers the highest prospective power to detect these effects. PC-SPR showed effect sizes similar to or larger than those in NC-SPR, and AVC-SPR was the least reliable task. Together, these findings challenge the assumption that cumulative presentation is unsuitable for studying real-time sentence processing and suggest that NAVC-SPR and PC-SPR are viable alternatives to NC-SPR. All cumulative SPR tasks, together with an R script for automated stimulus formatting, are openly available to facilitate their adoption.

The transition toward sustainable cities requires integrated energy planning frameworks that coordinate multiple technologies, policy instruments, and social considerations. This study proposes a robust optimization framework for rich-renewables eco-sustainable urban communities, where multi-energy hubs including electricity, thermal, cooling, and hydrogen systems are jointly managed under uncertainty. A scenario-independent static robust model is developed to ensure reliable operation under renewable intermittency, supported by sensitivity analyses. The framework introduces hydrogen chemistry consortium processes, integrating electrolyzers, methanation, fuel cells, and carbon capture, utilization, and storage to enhance renewable utilization and reduce emissions. Both stationary storage systems and electric public transportation fleets are incorporated to provide distributed and mobile energy flexibility. Demand-side management and policy mechanisms, including carbon taxation and cap-and-trade, are embedded to align operations with environmental targets. A digital-social welfare layer evaluates affordability and equitable access. Simulation results across multiple scenarios demonstrate that the proposed framework reduces operational costs by over 45%, improves grid independence by more than 35%, and achieves emission reductions exceeding 90%. Welfare indicators also show significant improvement, confirming the effectiveness of the integrated approach.

The versatility and broad applicability of IoT-based heterogeneous wireless sensor networks (HWSNs) make them key enablers of sustainability objectives in sustainable smart cities (SSCs). However, their heterogeneous architecture must be carefully managed to ensure reliable and energy-efficient operation. To prolong network lifetime and avoid premature node depletion, effective energy management mechanisms are essential. One promising approach is to exploit the concept of dominating sets (DSs), whereby sensor nodes are partitioned into disjoint DSs, and only one set is activated at a time. In this work, we propose EAAS-S4C-MAB, an enhanced framework for IoT-based HWSNs in SSCs that combines skyline-based DS formation, four-case, size- and lifetime-aware scheduling, and energy-aware data collection within the active DS. In the DS formation phase, the proposed Energy-Attentive Algorithm with Skyline (EAAS) generates multiple valid disjoint DS candidates per iteration, evaluates them jointly in terms of lifetime and size, and applies a BNL_SKYLINE procedure to retain only non-dominated candidates. In the scheduling phase, the proposed S4C-MAB algorithm classifies DSs into four cases according to size and lifetime and uses a multi-armed bandit to learn the relative priority of these cases online, replacing fixed case weights with adaptive case-level preferences. These learned priorities are combined with a dynamically recomputed adjusted lifetime, along with case-dependent activation caps and cooldown intervals, to guide DS activation. During communication, the active DS nodes form a chain, and the node nearest to the sink serves as the chain leader for final data delivery. Simulation results show that EAAS-S4C-MAB achieves more balanced DS utilization, reduces the premature depletion of fragile sets, and significantly extends network lifetime compared to other baseline methods, while preserving network coverage and connectivity.

Cholangiocarcinoma (CCA) is characterized by late or unclear diagnosis, high recurrence rates, and limited options for noninvasive disease monitoring. The widely used serum biomarker CA 19-9 has suboptimal sensitivity and specificity, underscoring the need for more accurate and reliable quantitative measures of tumor burden. Cell-free DNA (cfDNA) methylation represents a promising approach for longitudinal disease monitoring. In this single-center cohort study, we evaluated a cfDNA methylation-based Tumor Methylation Score (TMS) for detecting and monitoring CCA. Twenty-five patients with histologically confirmed CCA and twelve controls with benign hepatobiliary conditions were enrolled, yielding 100 plasma samples and 26 tissue samples. cfDNA methylation was quantified across 537 cancer-associated CpG loci by qPCR method, and TMS was calculated from the difference in methylation in paired plasma and buffy coat samples. For tissue samples, TMS was calculated by quantifying methylation in tumor and background liver separately. Response to treatment was assessed using blinded radiographic review. Tissue-based validation was performed using The Cancer Genome Atlas and 4 additional GEO cohorts. Finally, cellular pathways associated with the panel genes were explored using the Ingenuity Pathway Analysis (IPA). TMS was significantly elevated in CCA samples compared with controls in both tissue (P = 0.007) and plasma (P < 0.0001). Plasma TMS distinguished CCA from controls with a sensitivity of 74.3% and specificity of 80% (AUC = 0.823), while CA 19-9 showed no discriminatory value. TMS decreased consistently following curative resection. Longitudinal changes in TMS (&#x394;TMS) robustly distinguished radiographic progression from non-progression during systemic therapy (AUC = 0.968), outperforming &#x394;CA 19-9. Tissue-based analyses across independent cohorts confirmed cancer specificity and association with adverse outcomes. IPA identified HOXA9 as a significantly altered gene implicated in CCA carcinogenesis. Quantitative cfDNA methylation monitoring using TMS is feasible and accurately reflects dynamic tumor burden in CCA. TMS may serve as a complementary adjunct to imaging and CA 19-9 for diagnosis, prognosis, and longitudinal disease monitoring, warranting validation in larger prospective studies.

Measuring health, a state of complete physical, mental, and social well-being and not merely the absence of disease or infirmity, in Down syndrome (DS) was recently achieved through creation of a new health measure. However, additional information on health-related medical topics and how health measurement varies in demographic groups from that cohort has not been presented. We surveyed caregivers of individuals age 0-21 with Down syndrome (DS) from a national sample about the health of their son or daughter with DS as part of a broader study to create a valid, reliable health measure in DS. In this manuscript, we present secondary analysis of survey items on health-related topics including co-occurring medical conditions, recurrent infections, and healthcare maintenance. We then conducted an analysis to compare caregiver-reported demographic traits to total health scores through regression modeling to identify predictors of health status. Survey responses from 542 caregivers of individuals with DS were received; rates of co-occurring medical conditions generally aligned with past results, rates of recurrent infections showed lower rates of persistent fungal infections in our cohort (p&#x2009;<&#x2009;0.0001) and completion of healthcare guidelines did not correlate with health scores (p&#x2009;=&#x2009;0.13). In regression modeling, parent sex of respondent (male) and better parent health correlated with better total health scores of individuals with DS. Health-related topics show prevalence rates aligning with past literature, lower rates of some infections, and imperfect guideline adherence. Fathers and parents who feel that they are in better health reported better health of their sons and daughters with DS. Trial Registration: ClinicalTrials.gov, NCT04631237.

The rapid expansion of single-cell RNA sequencing (scRNA-seq) has made accurate cell type annotation a critical bottleneck for biological discovery. Existing computational methods are often limited by reference data dependency, while emerging single Large Language Model (LLM) approaches are susceptible to model-specific biases and provide insufficient uncertainty quantification. To address these limitations, we introduce mLLMCelltype, a framework that harnesses collective intelligence-the emergent problem-solving capacity arising when multiple independent agents interact through structured deliberation to produce solutions exceeding individual capabilities-of multiple LLMs through an iterative deliberation process. Across 49 diverse datasets, our framework achieves a mean accuracy of 77.2%, a 15.7-percentage-point improvement over the best-performing single-LLM baseline (61.5%). The consensus mechanism demonstrates high robustness to noisy input and generalizes to datasets released after the LLMs' training. By providing transparent reasoning chains and robust consensus-based confidence metrics, mLLMCelltype minimizes manual annotation effort and enables reliable interpretation of complex cellular landscapes. The framework is available as an open-source package and an accessible web server.

Diffusion magnetic resonance imaging (dMRI) provides powerful insights into brain microstructure, but conventional microstructural modeling methods require long acquisition times for covering sufficient diffusion directions and are computationally intensive. While deep learning has shown promise in reducing the direction requirement and accelerating the modeling, traditional architectures such as CNNs often struggle to capture the highly nonlinear relationships between multi-shell diffusion signals and microstructural properties. We present MicroKAN, a novel framework built upon Kolmogorov-Arnold Networks with adaptive spline-based activations, specifically designed to represent complex biophysical models with enhanced flexibility and efficiency. MicroKAN supports both supervised and self-supervised paradigms: the supervised variant learns mappings from data to reference metrics, while the self-supervised variant estimates model parameters directly by reconstructing signals through the forward diffusion process, eliminating the need for ground-truth labels. Evaluated on diffusion tensor imaging (DTI) and neurite orientation dispersion and density imaging (NODDI) across multiple datasets, MicroKAN substantially accelerates acquisition and improves the fidelity of microstructural parameter estimation. Beyond supervised training, its self-supervised formulation shows strong robustness to distribution shifts, enabling reliable performance even without annotations. Furthermore, transfer learning with minimal labeled data preserves high accuracy, underscoring the framework's adaptability to diverse scenarios. These advances establish MicroKAN as a versatile and efficient tool for dMRI analysis, offering new opportunities to accelerate neuroscience research and expand the clinical utility of microstructural imaging. Our source code is available at https://github.com/JustlfC03/MicroKAN.

Evaluating sustainable levels of offtake is central to regulating the commercial harvest of wildlife species, but methods developed for temperate-zone mammal and bird species may be inapplicable to tropical rainforest reptiles. For example, population viability models require information on abundance and on rates of growth, maturation, survival, and reproduction. For some intensively harvested reptiles - especially well-camouflaged ambush predators in densely vegetated habitats - such data are difficult to obtain. The feasibility of gathering this information is central to current debates about how best to evaluate the sustainability of commercial harvesting of reticulated pythons (Malayopython reticulatus) in Asia. Over a five-year period, we conducted 527 riverside surveys (with mark-recapture) for this species in a protected area in Sabah, Malaysia. Detectability of pythons was dependent on moon phase but always low (mean&#x2009;=&#x2009;20%); and the total number of pythons captured (N&#x2009;=&#x2009;159 total, N&#x2009;=&#x2009;83 during standardized surveys) and recaptured (N&#x2009;=&#x2009;25) despite that intensive long-term effort was insufficient to enable robust mathematical modelling. Our results suggest that attempts to base sustainable-harvesting targets for M. reticulatus on mathematical models are futile, and that repeated surveys of the attributes of harvested specimens provide a more reliable basis from which to evaluate sustainability.

Malignant melanoma is responsible for most skin cancer-related deaths due to its unpredictable behavior. Serum biomarkers have been widely investigated to improve prognostic assessment, yet their clinical utility remains inconclusive. This systematic review and meta-analysis evaluated the prognostic significance of serum biomarkers in melanoma. Following PRISMA guidelines and a registered protocol (PROSPERO: CRD42023486532), PubMed, EMBASE, and CENTRAL were searched for studies assessing biomarkers and survival outcomes. Univariate analyses revealed that elevated levels of LDH (HR 2.29, 95%-CI:1.97-2.67), S100B (HR 2.52, 95%-CI:1.59-3.99), circulating tumor DNA (ctDNA) (HR 2.61, 95%-CI:1.90-3.58), neutrophil-to-lymphocyte ratio (NLR) (HR 2.34, 95%-CI:1.86-2.93), and interleukin-6 (HR 3.11, 95%-CI:2.44-3.96) were significantly associated with reduced overall survival. Similarly, higher levels of LDH (HR 2.04, 95%-CI:1.68-2.47), S100B (HR 1.94, 95%-CI:1.39-2.70), ctDNA (HR 2.57, 95%-CI:1.95-3.39), and NLR (HR 2.38, 95%-CI:1.52-3.73) predicted shorter progression-free survival. These associations persisted in multivariable-adjusted analyses for LDH, ctDNA, NLR, IL-6, S100B, and CRP supporting their predictive relevance. LDH remains a reliable and cost-effective biomarker, while NLR may provide complementary prognostic information in patients receiving immune checkpoint inhibitors. Emerging biomarkers such as ctDNA demonstrate promising prognostic potential, but further evaluation is required before routine clinical implementation.

The detection of cancer cells not only facilitates the early diagnosis of cancers but also provides valuable support for formulating personalized treatment plans for patients. In this work, we develop a new aptamer-based electrochemical biosensing method for cancer cell detection, employing an electrochemically activated DNA circuit as a signal generation module. Specifically, target cells are captured onto the indium tin oxide electrodes through aptamer-mediated recognition of surface proteins. Then, a DNA circuit begins with the potential-driven selective bioconjugation of inherent tyrosine residues on the cell surface. This is subsequently followed by the incorporation of a facile click chemistry reaction and a netlike cascade assembly involving multiple hairpin probes. As a result, a large number of electroactive methylene blue molecules are recruited to the cell surface, thereby ensuring highly efficient electrochemical signal generation. Taking breast cancer cells MDA-MB-231 as a model, this method enables the detection of target cells in the linear range from 5 to 1&#x202f;&#xd7;&#x202f;104&#x202f;cells with a limit of detection down to 4.2&#x202f;cells, and also demonstrates high specificity and applicability to complex samples with good recoveries from 95.0% to 106.2%. Furthermore, the DNA circuit-based signal generation module exhibits excellent universality and signal stability, allowing the method to be adapted for detecting other cancer cells, such as HER-2-positive BT-474 and MDA-MB-453&#x202f;cells through simple aptamer replacement. Therefore, this work presents a reliable new tool for cancer cell detection, which holds great promise for playing an important role in future cancer diagnostic practices.

It is urgent to find novel non-invasive biomarkers that can accurately diagnose alcohol-associated liver disease (ALD). The objective of this study was to explore dysregulated metabolites in the serum of ALD patients by metabolomics, and establish a reliable diagnostic model by machine learning algorithms. A total of 1800 participants, including ALD, metabolic dysfunction-associated steatotic liver disease (MASLD), chronic hepatitis B (CHB), alcohol use disorder (AUD) and normal control (NC) individuals were recruited from four medical centers. Steroid hormone and bile acid metabolism was identified to be dysregulated in ALD in the discovery cohort by untargeted metabolomic analysis, and further confirmed in the training cohort by absolute quantitative metabolomic analysis. A machine learning model named "Bashald" was built based on the training cohort, and further validated in three independent validation cohorts. Our Bashald model exhibited great diagnostic performance with an AUC of 0.942 (95% CI, 0.880-1.000) in an internal test subset. In the validation cohorts, Bashald maintained good predictive performances, with the AUCs of &#x2265;0.821 for diagnosing ALD. In addition, Bashald demonstrated superior performance for the detection of early-stage ALD patients, with the AUCs of 0.870 and 0.792 for the training cohort and validation cohort 2, respectively, which had greatly surpassed traditional clinical indicators. Our research uncovered the specific metabolic profile of ALD and identified a distinct set of biomarkers that facilitate early detection, thereby promoting the application of precision diagnosis for ALD.