Spontaneous isospin ordering fundamentally reshapes quantum nonlinear transport, yet the nature of nonlinear responses in strongly correlated and symmetry-broken quantum states remain largely unexplored. In this work, we investigate a regime where symmetry breaking is not static but emerges dynamically as a tunable property of the electronic system, driven by strong correlations. Using high-quality, dual-gated Bernal bilayer graphene, we observe a giant nonlinear Hall conductivity (9.1 µm S V⁻¹) that manifests strongly in a field-driven isospin-polarized phase. In this correlated regime, the nonlinear conductivity scales exponentially with the linear conductivity, in stark contrast to the quadratic scaling expected in conventional nonlinear systems. Combined with quantum oscillation measurements and self-consistent theoretical modelling, our results indicate that this giant response is closely linked to an interaction-driven valley-polarized state. Our findings establish spontaneous isospin symmetry breaking as an effective route to giant nonlinear quantum transport, where interaction-enhanced skew scattering emerges as the dominant transport mechanism.
Although many neural network (NN) adaptive controllers have been proposed to deal with cooperation of nonlinear multiagent systems (MASs), it is still unknown how to achieve asymptotical cooperative goals over a general directed topology. A main challenge is the coupling of nonlinearities learning and cooperative control. Within this context, a novel class of adaptive controllers based on an NN-based cooperative modified state observer (CMSO) is proposed, where the CMSO can approximate unknown nonlinearities so that nonlinearities learning is decoupled into local tracking control under the proposed framework. It is proven that the controllers can achieve asymptotic consensus if the topology has a directed spanning tree. Note that both nonsmooth controllers and smooth controllers are proposed, where smooth controllers can avoid chattering, which may be induced by nonsmooth ones. Finally, a simulation over multiple-robot systems is given to validate the theoretical results.
In this paper, a novel electronic cochlear model is presented based on design philosophies of an ergodically enabled sequential logic (ESL) biomimetic circuit. It is shown that the presented cochlear model can reproduce key characteristics of typical nonlinear sound processing functions of mammalian cochleae: combination tone generation and pitch shift effect. A prototype of the presented cochlear model is implemented by a field programmable gate array (FPGA) and its functionality is validated through hardware experiments. It is then shown that the presented ESL cochlear model can be implemented by much fewer electronic circuit elements and much lower power consumption compared to a digital signal processor cochlear model. Finally, the significance of these results for the development of future hardware-efficient cochlear implants are discussed.
The aim of this study was to analyze the impact of an Anterior Cruciate Ligament (ACL) injury on movement variability (MV) and biomechanical variables in female basketball players during a drop vertical jump (DVJ) task, both with and without a ball. Female basketball players (n = 15) participated in this study. The participants performed two jump in each of six conditions (total = 12 jumps) in randomizer order: (i) Bilateral No Ball (2NB)-two-leg jump without a ball; (ii) Bilateral Ball (2B)-two-leg jump with a ball; (iii) ACL No Ball (ACLNB)-injured-leg jump without a ball; (iv) ACL Ball (ACLB)-injured-leg jump with a ball; (v) No ACL No Ball (NOACLNB) -uninjured-leg jump without a ball; (vi) No ACL Ball (NOACLB)-uninjured-leg jump with a ball. DVJ performance parameters were assessed using an accelerometer placed on the lower back and a force platform. MV was quantified using the multiscale entropy (MSE) derived from the acceleration data. The Complexity Index (CI) was also calculated. The ACLB condition exhibited the highest sample entropy (SampEn) value across all scales, followed by the 2B and NOACLB conditions, whereas ACLNB showed the lowest values. The CI also indicated significant variability across conditions. For linear biomechanical variables, significant differences were only observed in contact time (CT) between NOACLB and NOACLNB. The inclusion of a task constraint did not result in differences between ACL and NOACL groups in linear jumping performance metrics. However, nonlinear MV analyses, using MSE and CI, revealed condition differences.
Trihybrid nanofluids possess promising applications in the biomedical, electronic, cosmetic, and materials processing fields owing to their enhanced thermophysical characteristics. This study examined the flow and thermal characteristics of a tri-hybrid ([Formula: see text])/blood Casson nanofluid over a rotating disk, influenced by nonlinear thermal radiation, focusing on the Darcy-Forchheimer effects, slip phenomena, Biot number, and internal heat generation or absorption. The governing equations were simplified to a system of ordinary differential equations by introducing appropriate similarity variables and solved numerically using the BVP4C solver. The computed Nusselt number values were compared with those from earlier investigations in the literature and showed strong consistency, demonstrating the reliability of the present numerical model. Additionally, a response surface approach was utilized to create a regression model that showed excellent agreement with the numerical data, with a coefficient of determination of (99.99%), indicating high predictive accuracy for the skin friction coefficient. The analysis showed that the nanoparticle concentration had the greatest impact among the studied parameters, contributing approximately (70-75%), while the magnetic parameter accounted for about (14-16%), and the Casson parameter contributed nearly (10-13%) to the variation in skin friction.
The synthesis and investigation of 2-cyano-N'-(2-hydroxybenzylidene)-3-phenylacrylohydrazide (H2L) and its Ni2+, Cu2+, Co2+ and Zn2+ complexes are discussed in this paper. These compounds were characterized with different techniques, including CHN, IR, MS, 13C & 1HNMR, TGA, ESR, UV-Visible and the magnetic moments. DFT calculations (DFT/B3LYP) level was used for geometry optimization of the suggested structures with 6-311 + + G(d, p) basis set. An octahedral stereochemistry was suggested for Ni2+ and Co2+ complexes while; a square planar geometry was proposed for Cu2+complex. The results proposed four coordinated stereochemistry for Zn2+ complex. The IR spectra of the H2L were simulated and compared with the experimental result, achieving a correlation coefficient R2 = 0.99976. The UV-visible spectra were used in determining the optical band gaps and found to be in the range 3.26-3.28 eV. Non-linear optical (NLO) properties of the DFT optimized compounds were investigated and indicated higher values in comparison with urea. The molecular docking was utilized to study the possible interactions between the synthesized compounds and the targeted proteins of liver and colon cancer. The suggested compounds were examined towards HeP2G and HCT-116 cell lines to determine their anticancer activity that are compared to Doxorubicin and Sorafenib as standards. H2L exhibited strong cytotoxicity against HeP2G cell line; on the other hand, Zn2+ complex exhibited strong cytotoxicity against HCT-116 cell line.
Mechanistic PK/PD models represent complex biological systems in which model parameters are in nonlinear, multidimensional parameter spaces and often require robust optimization informed by experimental data, making parameter estimation a critical yet challenging aspect of pharmacometric analysis. Moreover, optimization methods for nonlinear-mixed effects PK/PD modeling are highly sensitive to initial parameter values, and poor initial estimates frequently result in convergence failure or suboptimal fits. To address the challenges, this study introduces a systematic framework that integrates population-based meta-heuristic algorithms to identify optimal or near-optimal initial estimates that facilitate subsequent nonlinear mixed-effects modeling. Nineteen distinct population-based meta-heuristic algorithms, including evolutionary, swarm-based, and bio-inspired methods, were evaluated on two mechanistic ODE-based models: (1) a two-compartment pharmacokinetic model with linear and Michaelis-Menten elimination and (2) the Friberg myelosuppression model. The temporal evolution of model goodness-of-fit over iterations was evaluated for each algorithm, and the effects of population size and iteration number on the algorithm performance were also examined. The results demonstrate population-based optimization algorithms efficiently and autonomously refine parameter estimates over iterations. Jellyfish search optimizer, symbiotic organisms search, and memetic algorithm showed strong performance compared to others with respect to optimization accuracy and computational efficiency across both models. As expected, increasing population size and iteration number enhanced the optimization performance. Overall, population-based meta-heuristic algorithms demonstrate superior capability in navigating nonlinear, high-dimensional parameter spaces, generating robust initial estimates for subsequent modeling workflows, and enhancing convergence stability, computational efficiency, and parameter identifiability.
ObjectiveThere is limited prospective evidence on the association of physical activity with metabolic dysfunction-associated steatotic liver disease. We aimed to use data from two major databases to systematically quantify the correlation of physical activity with the risk of metabolic dysfunction-associated steatotic liver disease.MethodsThis study used data from the National Health and Nutrition Examination Survey (2007-2020) and UK Biobank. Initially, a cross-sectional study was conducted using the data from the National Health and Nutrition Examination Survey data to assess the links between physical activity and metabolic dysfunction-associated steatotic liver disease using weighted logistic regression analysis with appropriate sample weights to ensure national representativeness. Subsequently, Cox proportional hazards regression and restricted cubic splines models were applied to infer causality and evaluate the potential dose-response relationships in the UK Biobank cohort over a median follow-up period of 14.18 years. Additionally, subgroup and sensitivity analyses were conducted to test the robustness of the results.ResultsIn the National Health and Nutrition Examination Survey, higher physical activity levels were inversely associated with metabolic dysfunction-associated steatotic liver disease risk (odds ratio: 0.78, 95% confidence interval: 0.63-0.97). This finding was supported in the UK Biobank cohort (hazard ratio: 0.57, 95% confidence interval: 0.34-0.76). A nonlinear dose-response relationship existed between physical activity and metabolic dysfunction-associated steatotic liver disease (p = 0.023 for a nonlinearity). Subgroup analyses highlighted stronger effects in overweight/obese individuals (hazard ratio: 0.62, 95% confidence interval: 0.43-0.88), suggesting that this high-risk population may derive particular benefit from increased physical activity.ConclusionsAnalyses of two large databases demonstrated a nonlinear dose-response association between higher physical activity levels and reduced metabolic dysfunction-associated steatotic liver disease risk. Moderate physical activity levels (600 to <3000 metabolic equivalent of task-min/week) were significantly protective, and high physical activity levels (≥3000 metabolic equivalent of task-min/week) conferred additional benefits.
Rapidly progressive Alzheimer disease and related dementias (rpADRD) is a clinically urgent syndrome characterized by accelerated deterioration. Because it is often confounded by acute etiologies, its definition remains poorly characterized. We aimed to evaluate rpADRD as a distinct clinical state, testing the hypothesis that it represents a clinical "phase transition" rather than a simple linear acceleration of typical neurodegenerative decline. Our objective was to define this transition threshold and quantify its impact on survival and disease trajectory. We conducted a retrospective cohort study using longitudinal data from participants recruited across US Alzheimer's Disease Research Centers. We defined incident rpADRD as dementia development (global Clinical Dementia Rating [CDR] ≥1) within 1 year of symptom onset, or progression to moderate-to-severe impairment (global CDR ≥2) within 2 years. To isolate primary neurodegeneration, we excluded prion, infectious, metabolic, toxic, and autoimmune etiologies. We used time-dependent Cox regression, multistate modeling, and restricted cubic spline analyses to model nonlinear survival dynamics. The final analysis cohort included 2,307 participants (mean baseline age 78.9 years; 57.1% female). Over a mean follow-up of 7.3 years, 270 patients (11.7%) developed incident rpADRD. Patients transitioning to rpADRD were older than non-rpADRD patients (mean 80.0 vs 78.8 years). Crucially, the transition to rpADRD was strongly associated with increased subsequent mortality (hazard ratio = 4.03; 95% CI 3.53-4.61, p < 0.001). Multistate and spline modeling confirmed that this shift represents a distinct, nonlinear clinical phase transition rather than a linear exacerbation of functional decline. Furthermore, this critical tipping point was reliably preceded by an early cascade of neuropsychiatric symptoms. These findings validate a clinical "phase transition" model for rpADRD, demonstrating that survival decline is not merely a linear function of disease severity. Rather, rpADRD constitutes a distinct state of clinical homeostatic failure fundamentally shifting subsequent mortality risk, often predicted by early neuropsychiatric symptoms. Recognizing this nonlinear shift is vital for accurate prognostication and segregation of rapidly progressing phenotypes in trials. Major limitations include reliance on observational data, potential recall bias regarding symptom onset, and the exclusion of neuroimaging and fluid biomarkers because of high missingness.
The advent of high-throughput sequencing technologies has generated increasingly large and complex genomic datasets, necessitating analytical approaches capable of capturing high-dimensional and potentially nonlinear genetic interactions. This situation has significantly impacted the entire field of Genome-Wide Association Study (GWAS), whose primary goal is the identification of genomic traits and variants that are statistically associated with the risk of a disease. However, traditional GWAS methods may show reduced performance when applied to highly polygenic and nonlinear genetic architectures. Computational strategies from Artificial Intelligence (AI) and, in particular, from machine- and deep-learning may provide a powerful tool to overcome such limitations, especially by capturing nonlinear interactions and complex hidden regularities in large-scale data, which traditional GWAS approaches might overlook. To date, only a few approaches have been introduced and systematically assessed. In this review, we describe the main characteristics and limitations of standard statistical approaches for GWAS, the main uses of AI methods in computational genomics, and recent attempts to leverage AI strategies in GWAS. Particular attention will be devoted to key issues, such as the interpretability of methods and results, and the curse of dimensionality. More specifically, the review presents 30 methods designed to leverage AI in GWAS, as well as presenting a comprehensive set of evaluation metrics for their performance, also providing references to the most frequently used databases, and biobanks. Overall, this work may serve as a starting point for both dry- and wet-lab researchers, aiming to extract deeper insights from genomic data by moving beyond traditional linear additive assumptions, and leveraging large-scale datasets through AI-driven approaches.
To address the accuracy bottleneck of the traditional Anderson blast-induced vibration prediction model, which assumes fixed sub-wave peak time and suffers from phase error accumulation during multi-hole superposition, a model correction method based on the nonlinear attenuation of peak time is proposed. Taking single-hole charge quantity, blast center distance and peak particle velocity (PPV) as core influencing factors, a peak time prediction formula is derived through dimensional analysis, revealing the nonlinear evolution law of peak time with propagation parameters. On this basis, an amplitude-time dual-parameter corrected Anderson nonlinear superposition model is constructed by simultaneously applying amplitude scaling and time-axis scaling to the reference wavelet. Systematic verifications are carried out through single-hole, double-hole and six-hole blasting tests as well as a 16-hole engineering case. The results show that compared with the traditional model, the relative error of PPV is reduced from the range of 15.2%-29.0% to 6.9%-14.0%, and the relative error of peak time is reduced from the range of 88%-115% to 1.3%-30.0%. The coincidence degree of waveform shape and spectral characteristics is significantly improved, and the dominant frequency prediction deviation of all test cases is controlled within 8 Hz. In the 16-hole large-scale open-pit bench blasting, the PPV prediction deviation is only 7.9%, and the peak time deviation is merely 2 ms. This model effectively solves the phase error accumulation problem in multi-hole blast-induced vibration superposition, and can provide reliable theoretical support for precise blast-induced vibration prediction and safety control.
Molecular junctions are promising for (low power) thermoelectric applications, providing their transmission landscape can be made sufficiently nonlinear at the Fermi level to efficiently break their electron/hole transport symmetry in response to a temperature gradient. We present a method to induce such nonlinearity leading to thermopower values of |S| > 0.5 mV/K. The method is applicable to metal-molecules-semimetal junctions in which a space charge region is formed within the semimetal with a characteristic length perpendicular to the interface that can be tuned to become comparable to the Fermi wavelength of the semimetal. Under such conditions, the interfacial density of states within this lead is quantized at energies that can be tuned to reside within ∼kBT from the Fermi level by varying the molecular length. The resulting conductance behavior within this energy range becomes sufficiently steep to substantially break electron/hole transport symmetry, with ensuing high thermopower values.
Post-stroke depression (PSD) is a common complication after stroke and contributes to poor functional recovery and reduced quality of life. Dietary quality has been associated with depression in the general population, but evidence in individuals with stroke remains limited. This study examined the cross-sectional association between the Healthy Eating Index-2020 (HEI-2020) and PSD. This study included 962 adults with a history of stroke from seven cycles of the National Health and Nutrition Examination Survey from 2005 to 2018. Depressive symptoms were assessed using the Patient Health Questionnaire-9, with a score ≥ 10 indicating probable depression. Dietary quality was evaluated using HEI-2020 and categorized into tertiles. Multivariate logistic regression, restricted cubic spline analysis, and mediation analysis were used to assess associations and the statistical indirect association of the Dietary Index for Gut Microbiota (DI-GM). After multivariate adjustment, each one-point increase in HEI-2020 score was associated with 2% lower odds of probable depression (PHQ-9-defined) (OR = 0.98). Compared with the low dietary quality group, the high dietary quality group had 59% lower odds (OR = 0.41). The dose-response relationship was nonlinear, with a significant inverse association observed only when HEI-2020 scores exceeded 57. This exploratory threshold may serve as a reference for future prospective studies but requires validation before clinical application. Furthermore, a significant interaction with race or ethnicity was identified (P for interaction = 0.033). DI-GM showed a significant indirect association (P = 0.04), but the direct and indirect components were in opposite directions, indicating an inconsistent mediation pattern. Given that DI-GM is a dietary proxy, these findings reflect statistical associations with a dietary pattern, not biological mediation. Future studies with direct microbiome measurements are needed to verify gut-brain axis involvement in PSD. In this cross-sectional study, HEI-2020 was inversely associated with PSD, with nonlinear and race-specific patterns. DI-GM showed a significant but inconsistent indirect association, suggesting a suppressive rather than mediating role. Findings are associational, not causal.
Owing to highly tunable mechanics, gelatin methacryloyl (GelMA) hydrogels are widely exploited for three-dimensional (3D) cell culture, whereas limited experimental sampling restricts efficient formulation screening. In this work, we developed a BNN-based modeling pipeline to map GelMA hydrogels with various cross-linking parameters toward linear viscoelastic moduli and nonlinear critical stress, thereby categorizing all tested formulations into low/intermediate/high stable mechanical windows. Calibration on C2C12 myoblast morphologies confirmed that nonlinear critical stress complements linear rheological parameters to refine the screening priority of cell-compatible hydrogel recipes. Subsequent validation with primary cardiomyocytes demonstrated consistent morphological trends matching the predefined mechanical windows, alongside ambiguous boundary formulations. Our findings construct a bounded prioritization strategy to rapidly select GelMA compositions under sparse experimental conditions, with further prospective validations demanded before generalized predictive use for diverse tissue engineering scenarios.
This research aims to provide a precise and computationally efficient solution for two important systems in mathematical modeling and environmental science: the fractional nonlinear coupled Burgers' system and the fractional dynamics of the coupled plankton-oxygen model in (1 + 1) dimensions. Despite the increasing complexity of these systems, existing numerical methods often struggle with the high computational costs of non-local operators. To address this, we propose a robust hybrid framework integrating the Fractional Differential Quadrature Method (FDQM) with a Newton-Raphson (NR) iterative procedure.These systems are critical for understanding various physical processes, including turbulent fluid dynamics and the biological interactions of oxygen and plankton in aquatic environments. The models use fractional derivatives, which provide greater flexibility in capturing memory and hereditary features, making them more suitable for correctly simulating real-world processes than traditional integer-order models. The proposed work solves these problems using a generalized Liouville-Caputo fractional-order model mixed with versions of the differential quadrature technique (DQM), which allows effective handling of complex boundary conditions and spatial derivatives. The nonlinearity in these equations is handled using Newton-Raphson's iterative approach, which ensures the solutions' stability and convergence. The proposed system was implemented in MATLAB, and a complete parametric analysis was carried out to investigate how various parameters, including the fractional-order derivative, the oxygen generation rate, and the maximum per capita growth rate of phytoplankton, influence model outputs. This study not only proves the suggested techniques' accuracy, convergence, and efficiency but also sheds light on their sensitivity to important parameters, making them more applicable to real-world circumstances. The findings of this work are expected to contribute to better modeling methodologies for complex systems, ultimately benefiting academics in domains ranging from environmental science to fluid dynamics.
This study aimed to investigate the association between methylmalonic acid (MMA) and epilepsy prevalence and to explore potential inflammatory and nutritional pathways underlying this association. This study included adults aged ≥ 20 years from the National Health and Nutrition Examination Survey (NHANES) 2011-2014. Weighted multivariable logistic regression was used to assess the association between MMA and epilepsy prevalence, and restricted cubic spline (RCS) analysis was performed to explore potential nonlinearity. Subgroup and mediation analyses were conducted to evaluate heterogeneity and the mediating effects of the systemic inflammation response index (SIRI) and albumin (ALB). An independent clinical cohort was used for preliminary clinical corroboration, and serum MMA and inflammatory/neuronal markers were measured by enzyme-linked immunosorbent assay. A total of 9996 participants were included. Higher MMA was significantly associated with epilepsy prevalence (continuous MMA: odds ratio [OR] = 1.377, 95% confidence interval [CI]: 1.017-1.865; highest versus lowest tertile [T3 vs. T1]: OR = 1.593, 95% CI: 1.024-2.477). RCS analysis suggested a nonlinear association, with an apparent inflection point around ln(MMA) = 4.8. The association was more pronounced in participants aged < 60 years (p for interaction = 0.006). Exploratory mediation analysis indicated that SIRI and ALB accounted for 3.875% and 6.340% of the observed association respectively (both p < 0.001). Findings from the clinical cohort were consistent with the NHANES results. Patients with epilepsy had higher serum MMA levels, and MMA levels were significantly correlated with both inflammatory and neuronal injury markers. Elevated MMA levels were associated with epilepsy prevalence in NHANES. The preliminary findings from the independent clinical cohort showed consistent patterns. Further prospective studies are needed to clarify the temporal relationship between MMA and epilepsy.
While conventional risk factors for liver cancer are well-established, the contribution of biological aging to hepatocarcinogenesis remains poorly understood, representing an important gap in its etiological framework. We conducted a prospective cohort study using data from the UK Biobank (UKB). Biological age (BA) was assessed using the Klemera-Doubal method (KDM-BA) and PhenoAge algorithms. Multivariable Cox proportional hazards models were used to evaluate associations between BA acceleration and liver cancer incidence, with robustness assessed using restricted cubic splines, subgroup analyses, and sensitivity analyses. Mediation analyses evaluated whether alanine aminotransferase (ALT) and aspartate aminotransferase (AST) mediate the relationship between biological aging and liver cancer. The study included 274,966 participants (mean age 55.75 ± 8.10 years; 53.8% female). Over a median follow-up of 14.67 years, 325 incident liver cancer cases were documented. After multivariable adjustment, each 1-standard deviation (SD) increase in BA acceleration was associated with an elevated liver cancer risk (KDM-BA acceleration: HR = 1.28, 95%CI 1.16-1.41; PhenoAge acceleration: HR = 1.32, 95%CI 1.22-1.43). A nonlinear dose-response relationship was observed between KDM-BA acceleration and liver cancer risk (P < 0.001). Exploratory mediation analyses indicated that the effect of KDM-BA acceleration may be partially mediated by ALT and AST, whereas PhenoAge acceleration appeared to act primarily through direct mechanisms. Accelerated biological aging, as quantified by KDM-BA and PhenoAge, appears to be independently associated with incident liver cancer. The underlying pathways may differ: KDM-BA acceleration is partially mediated by liver enzymes, while PhenoAge acceleration appears to operate largely through direct mechanisms.
This studyaimed to examine the association between Life's Crucial 9 (LC9) score and the prevalence of coronary heart disease (CHD) among U.S. adults, as well as explore the potential statistical contribution of Systemic Inflammatory Response Index (SIRI) in this association. Data were obtained from 23,508 participants in the National Health and Nutrition Examination Survey (NHANES) from 2005 to 2018. LC9 score quartiles were used to summarize continuous and categorical variables, which were then reported as weighted means ± standard deviations or weighted frequencies and proportions, respectively. Differences among groups were examined using weighted t-tests and weighted chi-square tests. The relationship between LC9 scores, Ln-SIRI values, and the prevalence of CHD was investigated using weighted logistic regression models following the natural logarithmic transformation of skewed SIRI results. Furthermore, restricted cubic spline regression was utilized to investigate plausible nonlinear connections, subgroup and likelihood ratio tests investigated interaction effects, and weighted quantile sum (WQS) regression was incorporated to determine the relative contributions of LC9 components to CHD. An exploratory non-causal mediation analysis using 1,000 bootstrap samples was conducted to assess SIRI as a potential statistical explanatory variable in the association between LC9 and the prevalence of CHD. In Model 3, a 10-point increase in the LC9 score correlated with a 24% decrease in the odds of CHD prevalence; participants in the Q4 group had 59% lower odds of CHD compared with those in the Q1 group (P < 0.001 trend). Each Ln-SIRI unit was associated with 56% higher odds of CHD, with Q4 showing 2.02-fold higher odds than Q1 (P < 0.001 trend). Restricted cubic spline regression revealed linear relationships between LC9, Ln-SIRI, and the prevalence of CHD. In particular, LC9 was negatively associated with CHD, whereas Ln-SIRI was positively associated with CHD. Age significantly moderated the association between LC9 and CHD (P for interaction < 0.05), with stronger inverse associations in younger adults. WQS regression identified glucose (weight = 0.34) and tobacco (weight = 0.33) as key CHD-associated factors. Exploratory non-causal mediation analysis revealed that about 6.5% of the LC9-CHD relationship was statistically accounted for by SIRI (P < 0.001), with the majority of factors associated with the link remaining unexplained. This indicates that other factors may be the primary contributors to the LC9-CHD relationship. This study showed that higher LC9 scores were significantly associated with lower odds of prevalent self-reported CHD, while higher SIRI was associated with higher odds.
A tipping point marks the threshold or critical state where a biological system shifts from one stable state to another. Deciphering critical transitions and their associated signaling molecules is essential for elucidating complex biological processes and for enabling timely interventions to avert or postpone catastrophic deteriorations. However, existing critical-state detection methods rely mainly on Euclidean-space statistics, which may overlook nonlinear dynamical behavior among molecules and distribution-based molecular patterns, leading to limited robustness and performance in high-dimensional, sparse, and noisy single-cell data. In this study, we introduce single-cell Tipping-point Identification via Distributional Embedding (scTIDE), a framework that integrates manifold-based graph representations with optimal-transport conditional flow matching (OT-CFM) to capture intrinsic topological structure and identify critical transitions at the individual-cell level. Specifically, for a given cell, scTIDE quantifies distributional differences between a distribution derived from the reference manifold graph and a perturbed distribution inferred from the cell-perturbed manifold graph using OT-CFM, thereby identifying critical stages and key signaling molecules. The reliability and effectiveness of our model are demonstrated through synthetic models and eight distinct single-cell datasets, where it outperforms existing methods. Moreover, scTIDE reveals possible critical transitions for unseen cells and visualizes the intricate biological progression.
This study developed a full-scale transfer-learning workflow integrated with an anaerobic-digestion-specific feature engineering pipeline for methane prediction in newly commissioned anaerobic digesters with limited monitoring data. The performance of anaerobic digestion is significantly influenced by heterogeneous substrates, fluctuating operating conditions, and time-dependent microbial dynamics, resulting in nonlinear and site-specific methane production behaviors. To address this, we developed a comprehensive biogas machine learning (ML) pipeline using long-term monitoring data from a data-rich mesophilic digester, employing an ensemble of ML models, including Random Forest, XGBoost, and LightGBM. The optimized model was then applied to a newly commissioned digester with limited and incomplete datasets, without additional hyperparameter optimization. Remarkably, Hyperparameter-transfer-based transfer learning achieved prediction accuracy comparable to that of a fully re-optimized target-domain model, with an R2 of 0.88, while reducing model development time from 59.1 s to approximately 5 s. The transferred ensemble model achieved an RMSE of 171.70 m3/day and an MAE of 138.58 m3/day, compared with 151.48 m3/day and 121.43 m3/day, respectively, for the fully tuned target-domain model. Overall, this framework offers a practical and computationally efficient method for early-stage methane prediction, facilitating the rapid deployment of reliable forecasting tools in newly commissioned anaerobic digesters.