Browse Articles

STF-DKANMixer: Tri-component decomposition with KAN-MLP hybrid architecture for time series forecasting

Long-term time series forecasting is critical for domains such as traffic and energy systems, yet contemporary models often fail to capture complex multiscale patterns and nonlinear dynamics, resulting in significant inaccuracies during periods of abrupt change. To overcome these limitations, we introduce STF-DKANMixer , a novel hybrid architecture combining a Multi-Layer Perceptron (MLP) with the expressive power of the Kolmogorov–Arnold Network (KAN). Our framework begins with a DFT-based decomposition strategy: long-term trends and seasonal components are extracted directly via Discrete Fourier Transform (DFT), while the remaining residual is further decomposed into high-frequency details using a Haar wavelet transform with error compensation. In the Past-Information-Mixing (PIM) stage, each component is processed by a GELU-activated KAN module for superior nonlinear feature mapping before being fused by a novel deformable feature attention (DFA) block, which adaptively learns sampling offsets and weights to capture complex dependencies. Subsequently, the Future-Information-Mixing (FIM) stage leverages an adaptive weighted ensemble of multiple lightweight predictors, enhanced by residual connections, to generate the final forecast. Extensive experiments on benchmark datasets validate the superiority of our approach. STF-DKANMixer significantly outperforms state-of-the-art models, reducing Mean Squared Error (MSE) by up to 36.1% (12.3% on average) and Mean Absolute Error (MAE) by up to 28.8% (8.8% on average). Impressively, these results are achieved while using less than half the computational resources of comparable methods. Our findings establish STF-DKANMixer as a robust, efficient, and highly accurate solution, setting a new performance standard for complex, long-horizon forecasting tasks.

Clonal cell states link gastroesophageal junction tissues with metaplasia and cancer

Abstract Barrett’s esophagus is a common type of metaplasia and a precursor of esophageal adenocarcinoma. However, the cell states and lineage connections underlying the origin, maintenance, and progression of Barrett’s esophagus have not been resolved in humans. Here, we perform single-cell lineage tracing and transcriptional profiling of patient cells isolated from metaplastic and healthy tissue. Our analysis unexpectedly reveals evidence for lineages spanning squamous esophagus, gastric cardia, and transitional basal cells at the tissue junction. We also identify lineages connecting Barrett’s esophagus to both esophageal and gastric tissues. Barrett’s esophagus biopsies consist of multiple distinct clones, with lineages that contain all progenitor and differentiated cell types. We discover Barrett’s esophagus cell types, including tuft, ciliated, and BEST4+ cells, which we validate through both lineage relationships and spatial transcriptomics. In contrast, the precancerous dysplastic lesions show expansion from a single molecularly aberrant Barrett’s esophagus clone. Together, these findings provide a single-cell view of the cell dynamics of Barrett’s esophagus, linking cell states along the disease trajectory, from its origin to cancer.

Diagnostic performance of real-time artificial intelligence using deep learning analysis of endoscopic ultrasound videos for gallbladder polypoid lesions

Building extraction from remote sensing imagery using SegFormer with post-processing optimization

Traditional methods for building extraction from remote sensing images rely on feature classification techniques, which often suffer from high usage thresholds, cumbersome data processing, slow recognition speeds, and poor adaptability. With the rapid advancement of artificial intelligence, particularly machine learning and deep learning, significant progress has been achieved in the intelligent extraction of remote sensing images. Building extraction plays a crucial role in geographic information applications, such as urban planning, resource management, and ecological protection. This study proposes an efficient and accurate building extraction method based on the SegFormer model, a state-of-the-art Transformer-based architecture for semantic segmentation. The workflow includes data preparation, model construction, model deployment, and application. The SegFormer model is selected for its hierarchical Transformer encoder and lightweight MLP decoder, which enable high-precision binary classification of buildings in remote sensing images. Additionally, post-processing techniques, such as noise filtering, boundary cleanup, and building regularization, are applied to refine the inference results, significantly improving both the visual presentation and accuracy of the extracted buildings. Experimental validation is conducted using the publicly available WHU building dataset, demonstrating the effectiveness of the proposed method in urban, rural, and mountainous areas. The results show that the SegFormer model achieves high accuracy, with the MiT-B5 backbone network reaching 94.13% Intersection over Union (IoU) after 100 training epochs. The study highlights the robustness and scalability of the method, providing a solid technical foundation for remote sensing image analysis and practical applications in geographic information systems.

Structure, function, and implications of fucosyltransferases in health and disease

Improving wind power prediction with advanced temporal and frequency domain processing combined with error correction

Stakeholder communication and client understanding of viral load suppression in Blantyre, Malawi—What role did U = U play before the flip the script campaign?

Background Various studies have demonstrated the effectiveness of viral load suppression (VLS) in preventing sexual transmission of the human immunodeficiency virus (HIV), leading to the slogan “U ndetectable = U ntransmittable” or “U = U”. As few studies have examined health stakeholders’ understandings of U = U, the communication of a suppressed viral load (VL) to male clients, and male clients’ understandings of the benefits of a suppressed VL in Sub-Saharan Africa, we explore these in this study. In the field of HIV, it is only in recent years that men who have sex with women (MSW) have been given some attention. The treatment cascade shows that MSW lag behind women in knowing their HIV-status, accessing treatment and suppressing their VL. Methods and findings Our findings are based on in-depth qualitative semi-structured in-person interviews with 16 local stakeholders (health facility personnel, NGO and church-based HIV programme implementers and applied researchers) and 39 men on antiretroviral treatment (ART) in Blantyre, Malawi: Twenty-three men with a detectable VL and 16 with a suppressed VL were included in the study. Thematic analysis of the audio-recorded and transcribed interviews show that stakeholders were mostly unfamiliar with the slogan U = U. Despite realizing the advantages for clients and their partners, stakeholders were cautious about how they conveyed information about an undetectable VL: they were worried about clients either misinterpreting an undetectable VL as being cured of HIV or engaging in risky sexual behaviour. This attitude prevailed irrespective of the degree of stakeholders’ endorsement of the evidence of U = U. Male clients were aware of personal health benefits of a suppressed VL, yet unaware of their non-infectiousness. In order not to infect others, some men ceased sexual relations or forwent procreation. Conclusion For U = U to become part of a client’s actionable knowledge with regard to treatment benefits, U = U needs to be communicated comprehensively. To what extent recent policy changes, the Flip the Script campaign to make U = U known and staff training programs have led to a different practice will need to be further researched, using our findings as a reference point. Integrated health services are already in place; however, VL monitoring, and testing and treatment for sexually transmitted infections (STIs) should ideally be conducted more frequently than annually, so that Malawian clients can be assured of their VL status and of not passing on the virus while also being able to get any STIs treated.

Adjuvant benmelstobart plus anlotinib in patients with high-risk recurrence after resection of hepatocellular carcinoma: a phase II study (ALTER-H006)

B lymphocytes contribute Angiotensin II induced cardiac hypertrophy

IconicITA: Iconicity ratings of the Italian affective lexicon

Iconicity, defined as the potential of linguistic signs to resemble properties or features of their referents, is increasingly recognized as a general property of language. One common approach for quantifying iconicity is to collect iconicity ratings. Although iconicity datasets have been developed for several languages, no comprehensive dataset of iconicity ratings is currently available for Italian. The current study presents IconicITA , the first dataset of Italian iconicity ratings for the 1,121 words of the Italian adaptation of Affective Norms for English Words (ANEW). Ratings were collected from both Italian native speakers (L1) and English native speakers with Italian as a second language (L2). Including L2 participants allowed us to contribute to the debate on whether iconicity ratings genuinely measure form-meaning resemblance, rather than exclusively reflecting semantic properties. We showed that L1 Italian iconicity ratings are positively associated with perceptual strength in the auditory and haptic modalities, and with specificity ratings. Conversely, we found a negative correlation between iconicity and concreteness, age of acquisition, word frequency, and letter frequency. In general, the relationship between Italian iconicity norms and various psycholinguistic variables largely replicated previous findings in the literature on iconicity. Considering L2 data, the ratings provided by L2 speakers correlated more strongly with the Italian L1 data compared to the translation-equivalent English L1 data. This finding suggests that participants’ judgments were influenced not only by the semantic information of the words but also by language-specific form-level properties. We take this result as evidence of the validity of iconicity ratings to operationalize the degree of resemblance between words’ form and meaning.

Modifiable influencing factors and their joint effects on early- and late-onset coronary heart disease

High genomic structuring in Euterpe precatoria in the Brazilian Amazônia and new implications for restoration and conservation efforts

Construction of Tetrasubstituted α‐Amino‐ and α‐Alkoxy Phosphine Oxides via Pd‐Catalyzed Regio‐ and Enantioselective Hydrophosphinylation of Dienes

Abstract Significant progress has been made in the asymmetric hydrofunctionalization of dienes to construct vinylic stereogenic carbon centers. However, achieving enantioselective hydrofunctionalization of substituted dienes with heteroatom nucleophiles to form vinylic tetrasubstituted carbon centers remains a formidable challenge. This difficulty arises primarily from issues of regio‐control, steric hindrance, and stereo‐discrimination. In this study, we present a palladium‐catalyzed regio‐ and enantioselective hydrophosphinylation of 2‐amido and 2‐alkoxyl dienes using phosphine oxides. This approach successfully constructs chiral allylic α‐aminophosphine oxides and α‐alkoxyphosphine oxides with tetrasubstituted carbon centers. Our method demonstrates a broad substrate scope with excellent enantioselectivity (up to > 99% ee) and high yields (up to 99%), achieving exclusive regio‐control under mild conditions. Additionally, the versatile post‐functionalization of the allyl group facilitates the synthesis of a wide variety of tetrasubstituted carbon centers featuring distinct heteroatom.

Magnitude of tuberculosis treatment outcomes and associated factors in public health institutions of Arba Minch town, Southern Ethiopia: A multi-centered retrospective cross-sectional study

Background In Ethiopia, the incidence rate of tuberculosis (TB) has been steadily increasing, from 119 per 100,000 in 2021–126 per 100,000 in 2022 and 146 per 100,000 in 2023. Moreover, TB remains the second leading cause of death after malaria, and the third leading cause of hospital admissions. So, the aim of this study is to determine the treatment outcome and associated factors in the public health institutions of Arba Minch town. Methods A multi-centered retrospective cross-sectional study was conducted involving 609 tuberculosis patients admitted from September 2021-August 2024 at public health institutions in Arba Minch town. A structured data extraction form was used by trained research assistants to collect the data from TB patient registration record books. Variables with a p-value <0.25 in binary logistic regression were further analyzed using multivariable logistic regression. Statistically significant factors were considered those with a p-value <0.05. Result The majority (53.7%) of the participants were aged between 21 and 40 years. Most participants were diagnosed with pulmonary tuberculosis (71.8%) and found to be new (92.9%) patients. According to this study, the magnitude of successful treatment outcomes was found to be 86.9%. In the multivariate logistic regression, being unmarried (p = 0.023), educational level (p = 0.028), and having extra-pulmonary TB (p = 0.017) have been found significantly associated with successful treatment outcome. Conclusion The study indicates a relatively positive rate of successful treatment outcomes for TB. Although the treatment outcome results are positive, targeted interventions are needed for individuals who are married, have a low educational status, and have been diagnosed with pulmonary tuberculosis.

Volumetric localization microscopy with deep learning

Developing Gaussian process regression, Lasso regression, and Nu-support vector regression models for predicting solubility of exemestane in supercritical CO2

Abstract Precise estimation of pharmaceutical solubility in supercritical carbon dioxide (scCO 2 ) is essential for optimizing pharmaceutical applications, including particle size reduction, the development of solid dispersions, and controlled-release formulations. In this research, we present a comparative analysis of three machine learning regression models—Lasso Regression, Gaussian Process Regression (GPR), and Nu-Support Vector Regression (Nu-SVR)—for predicting the solubility of exemestane (EXE), a poorly water-soluble anticancer drug, in scCO 2 under varying temperature and pressure conditions. The dataset used in this work consists of 45 experimental measurements encompassing temperature (T in K), pressure (P in MPa), and solubility (in g/L) of EXE. The dataset was divided into training and testing data subsets to facilitate reliable model validation. Model performance was thoroughly evaluated using metrics such as the R², RMSE, MAE, and AARD%. Additionally, decision surfaces and observed-versus-predicted plots were generated to visually assess model accuracy. Among the applied models, Gaussian Process Regression demonstrated superior predictive capability with an R² score of 0.996, Maximum error of 3.27, significantly outperforming both Lasso and Nu-SVR models. These results indicate that GPR effectively captures the nonlinear relationship between process variables and drug solubility, offering high generalization and precision. Feature importance analysis confirmed that pressure has the most significant influence on solubility behavior, while temperature also contributes positively to solubility trends. Residual analysis further validated the consistency and reliability of the GPR-based model. This work contributes to the growing application of machine learning techniques in pharmaceutical process modeling, particularly in supercritical fluid-based drug delivery systems. The proposed GPR model provides a reliable and efficient tool for predicting solubility, supporting the design and optimization of scCO 2 -assisted drug formulation methods.

Exploring the immune responses triggered by vaccine formulations containing the recombinant Schistosoma mansoni 14kDa fatty acid-binding protein

Many different Schistosoma antigens have been evaluated as vaccine candidates, including the recombinant form of the Schistosoma mansoni 14-kDa fatty acid-binding protein (rSm14). However, recombinant proteins often lack intrinsic immunostimulatory activity, a limitation that can be addressed by using vaccine formulations that contain adjuvants. In this work, we describe the immune response triggered by rSm14, a vaccine candidate against schistosomiasis currently under clinical trial, formulated with either (i) Monophosphoryl Lipid A (MPLA), (ii) MPLA/Alum, or (iii) Freund’s adjuvant. rSm14/MPLA and rSm14/MPLA/Alum formulations induced increased frequency of effector and memory CD4 + T and central memory CD8 + T cells, respectively. Both formulations induced significant production of rSm14-specific IgG and IgG1 antibodies, which could recognize the protein’s native form. The rSm14/Freund’s formulation elicited a robust immune response characterized by increased levels of IFN-γ, TNF, IgG, IgG1, and IgG2c antibodies, and expansion of memory B cell. These soluble factors have been implicated in the efficacy of Sm14-based vaccines. Despite inducing both humoral and cellular immune responses, the different formulations did not impact worm burden and the number of eggs trapped in the liver and intestine. Altogether, our findings indicate a limitation in the use of the molecules assessed in this study, such as IFN-γ, TNF, and specific antibodies, as correlates of protection and vaccine efficacy.

Mammo-AGE: deep learning estimation of breast age from mammograms

Abstract Biological age is an important indicator of organ functions and health. Although mammograms are widely used in breast cancer screening, the potential of mammogram-based biological age predictors remains underexplored. Here, we propose a deep learning model to estimate the biological age of the breast using healthy mammograms. The model is developed on three large datasets and externally validated on two additional datasets, encompassing 95,826 mammograms from 44,497 women aged 18 to 98 years. It demonstrates accurate age estimation (mean absolute error: 4.2 − 6.1 years) with strong correlation to chronological age. Predicted breast age stratifies breast cancer risk similarly to chronological age. Occlusion analysis, employed for model interpretation, reveals the aging-related pattern of the breast. The breast age gap (the difference between system-bias-corrected breast age and chronological age) may reflect breast health status. Breast cancer patients show higher breast age gaps than the healthy population. In two longitudinal datasets, larger breast age gaps are associated with increased future breast cancer risk, with hazard ratios of 1.013 − 1.022. Furthermore, we finetune the model specifically for downstream breast cancer diagnosis and risk prediction. Our approach outperforms other comparative methods, showing its potential for supporting both early detection and personalized screening strategies.

Chymotrypsin digestion analysis of glycyl radical and B12-dependent radical enzymes indicates common substrate-induced structural shifts

Beauty isn’t everything: An agent-based model of imperfect food acceptance and market utility balance

Imperfect or suboptimal foods—cosmetically flawed yet nutritionally sound—are frequently discarded across the food supply chain, contributing significantly to global food waste. Although farmers, retailers, and consumers all influence this waste dynamic, existing research often treats their behaviors in isolation and fails to capture the evolving, systemic interplay among them. This study presents a novel agent-based model simulating interactions between farmers, retailers, and consumers in local produce markets to examine how preferences, stocking strategies, and marketing interventions shape market behavior over time. The model integrates behavioral mechanisms such as the mere-exposure effect and Prospect Theory to realistically represent consumer adaptation and decision-making. Through simulations across varying market configurations, we identify critical points in consumer behavior and market-wide utility. Our findings reveal that promoting imperfect foods can lead to substantial utility gains for both consumers and retailers—particularly when retailers stock high levels of imperfect products consistently over time. These results emerge from simulations showing that the mere-exposure effect drives consumer preference shifts even in the absence of marketing, enabling widespread acceptance under certain stocking strategies.