Browse Articles

Understanding machine learning weather prediction by designing a cost-efficient model with knowledge-oriented modules

Abstract Deep learning-based models are gaining prevalence in global weather forecasting, surpassing the performance of existing numerical models. However, training these models with high-resolution global weather data requires massive computational resources, making it difficult to conduct extensive experiments to understand the model processes. In addition, the reason for region- or variable-dependent accuracy in the machine learning models, along with the extra predictability provided by each component, remains unknown. Therefore, we propose a novel data-driven model named KARINA, which combines Geocyclic Padding and SENet modules with the ConvNeXt backbone to enhance weather forecasting while minimizing training resources. Despite its much lower training cost, KARINA achieved competitive performance compared to the recently developed data-driven models such as Pangu-Weather and GraphCast, while surpassing the numerical weather prediction of ECMWF IFS at a lead time of up to 10 days. The efficient training process and KARINA’s modular structure allow us to demonstrate the effectiveness of Geocyclic Padding and SENet through comprehensive trials. Geocyclic Padding significantly improves the modeling of horizontal advection, while SENet particularly captures the dynamics of atmospheric convection. These findings suggest that incorporating knowledge-oriented techniques can lead to reliable performance. This paper presents a framework for gaining a deeper understanding of the model mechanism and proposes ways to improve machine learning weather prediction models.

Structural basis of dimerization and cascade formation by Cas5

Prediction of remaining useful life for electronic equipment based on online PINN

Low bias negative differential resistance in WSe2/MoS2 Planar Superlattice Diodes

Design and optimization of a polarization-insensitive Ti/TiO2 metamaterial absorber using particle swarm optimization for broadband solar–thermal applications

Abstract In this paper, a polarization-insensitive and ultra-broadband metamaterial absorber composed of titanium (Ti) and titanium dioxide (TiO 2 ) resonators is proposed. The multilayer architecture integrates three square resonators and one disk resonator, which are optimized to maximize absorption across the 0.25–4 μm spectral range. Particle swarm optimization (PSO) yields an average absorption of 98.99% and a bandwidth of 3533 nm (467–4000 nm). The structure maintains absorption above 80% for transverse magnetic (TM) polarization and above 90% for transverse electric (TE) polarization for wavelengths above 600 nm, even at incidence angles up to 60°. Under the Air Mass 1.5 Global spectrum (AM 1.5G) solar spectrum, the device achieves a solar absorption efficiency of 98.17% and a thermal emission efficiency of 99.26% at 1800 K. This performance is a result of the combined effects of surface plasmon polaritons, magnetic plasmons, and localized surface plasmon resonances. The absorber further demonstrates robustness against structural variations, maintaining over 90% absorption across the wavelength range of 250–4919 nm (BW = 4669 nm). The designed absorber exhibits a compact electrical size, with a total thickness of approximately 0.2λ and periodicity of 0.1λ, calculated at the longest operating wavelength (4 μm), confirming its subwavelength characteristics and suitability for planar integration. In comparison to existing Ti-based absorbers, this design provides both the broadest operational bandwidth and the highest efficiency, indicating strong potential for solar energy harvesting applications.

Next-generation Candida albicans vaccine VXV-01 containing recombinant Als3p and Hyr1p antigens for invasive Candida infections

Study on the anti-tumor effect of Wall-broken ganoderma lucidum spore powder on papillary thyroid cancer

Generalized Galton’s boards explain social phenomena via statistical physics

222Rn, 226Ra, and 238U in soil samples at a depth of 50 cm in KUFA City of AL-NAJAF Governorate, IRAQ

Statins attenuate PD-L1 sorting to small extracellular vesicles dependent on ubiquitin-like 3 modification

Abstract Small extracellular vesicles (sEVs) mediate cell-to-cell communication by carrying RNAs and proteins. Ubiquitin-like 3 (UBL3) functions as a posttranslational modification factor, regulating protein sorting to sEVs. Programmed cell death ligand 1 (PD-L1) binds to programmed cell death 1 (PD-1) on immune cells, suppressing their function. Although immune checkpoint inhibitors, anti-PD-L1 and anti-PD-1 antibodies, have improved cancer treatment, efficacy remains limited (~ 25%). Per recent studies, PD-L1-containing sEVs are elevated in cancer patients, contributing to impaired immunotherapy responses. Herein, we discovered that PD-L1 is modified by UBL3 and that its sorting to sEVs is regulated by UBL3. Furthermore, we found that statins, commonly prescribed for hypercholesterolemia, inhibit UBL3 modification, thereby reducing PD-L1 sorting to sEVs. Among patients with a high tumor proportion score, serum levels of PD-L1-containing sEVs were significantly lower in those using statins. Consistently, bioinformatic analysis revealed that UBL3 and PD-L1 expression levels affect lung cancer survival. Integrating statins into existing combination therapies may therefore offer a promising strategy to enhance immunotherapy efficacy.

Engineering dielectric properties and charge transport in PANI/CuO nanocomposites via microstructural control

Abstract This study systematically investigates the structure-property relationships in polyaniline/copper oxide (PANI/CuO) nanocomposites, with a specific focus on how controlled CuO incorporation (0.5 to 1.25 mol%) tunes their microstructural, dielectric, and charge transport characteristics. The key innovation of this work lies in establishing a direct correlation between CuO-induced lattice expansion and the evolution of charge transport mechanisms, revealing a tunable microstructural-dielectric coupling. X-ray diffraction confirmed successful composite formation, revealing a significant lattice expansion and an optimized microstructure with increased crystallite size and reduced micro-strain. Dielectric spectroscopy demonstrated a remarkable enhancement in the dielectric constant and revealed a distinct interfacial polarization peak. The analysis of AC conductivity identified Overlapping Large-Polaron Tunneling as the dominant charge transport mechanism, a finding further supported by the calculated trends in hopping distance. Complex impedance analysis confirmed non-Debye relaxation behavior and visualized the critical role of interfacial effects, which transition from blocking to conductive with increasing temperature. The PANI/CuO-1 mol composite emerged as the optimal candidate, achieving an ideal balance between high charge storage and efficient transport. This work not only advances the fundamental understanding of charge dynamics in hybrid systems but also underscores the potential of these tailored nanocomposites for high-performance capacitive and optoelectronic applications.

Modeling, simulation, and optimization behavior of pharmaceutical compound removal from water in SR-AOPs technique for wastewater treatment

Deep learning based thyroid prediction with opposition learning based red panda optimization feature selection

Explainable machine learning model and gene expression programming for predicting reinforced concrete beams moment capacity exposed to fire

Abstract In this study, a new formulation for the moment capacity ( M r ) of Reinforced Concrete (RC) beams under fire conditions is estimated using Gene Expression Programming (GEP). In addition, the use of Machine Learning (ML) methods such as XGBoost, AdaBoost, and LightGBM is investigated for estimating the M r of RC beams in fire. The database for predicting the M r of RC beams includes 280 samples. In this paper, the cross-section width b w , cross-section depth d , distance from the beam edge to the center of steel reinforcement d eff , area of steel reinforcement A st , time duration of fire t , compressive strength of concrete f c , and moment capacity of the beam under fire M r are considered as the parameters of ML models. Several statistical metrics were employed to assess the performance of the models, including the mean absolute error ( MAE ), mean square error ( MSE ), root mean square error ( RMSE ), coefficient of determination ( R 2 ), and gradients of regression lines ( k and k ′). In this study, Shapley Additive exPlanations (SHAP) analysis was used to interpret the predictions of the XGBoost model, which was selected for its high accuracy with the best R 2 and the lowest error rate. The results indicate that the methods used demonstrate high accuracy in estimating the M r of RC beams.

Heavy metal in cosmetics and its risk to future generation in remote area of Azad Jammu and Kashmir Trarkhel District Sudhnoti

Enhanced DEWMA-type control chart for process mean monitoring utilizing auxiliary information

Abstract Statistical Process Control (SPC) improves product quality by monitoring process performance, with control charts being the primary tool to detect and manage variability. The effectiveness of a control chart can be enhanced by incorporating additional pertinent information regarding the study variable. This study revisits the DEWMA chart, which is designed to monitor variations in the process mean under the assumption that the underlying process follows a normal distribution. We propose a Regression-based DEWMA (M R DEWMA) control chart that utilizes an auxiliary variable through a regression estimation method to determine the process mean. The control limits for the proposed chart are established based on both time-varying and asymptotic conditions. The features of run length (RL), including Average Run Length (ARL), Standard Deviation of Run Length (SDRL), and Median Run Length (MRL), are evaluated using Monte Carlo simulations. A comparative analysis reveals that the proposed M R DEWMA chart outperforms the traditional DEWMA chart in detecting small to moderate shifts in the process mean. The efficacy of the proposed approach is demonstrated using both Monte Carlo simulated data and a real-life industrial case study.

Textbook outcomes after splenectomy in patients with portal hypertension

Effect of process parameters on surface integrity in laser powder bed fusion of Ti-6Al-4V alloy

Proteome-wide mendelian randomization reveals circulating proteins causally associated with childhood body mass index

Abstract Childhood obesity is a major public health problem, affecting one in 5 youths. We aimed to characterize biomarkers for pediatric obesity among circulating proteins using Mendelian randomization (MR). We utilized genome-wide significant cis-protein quantitative trait loci (pQTL) from three large adult proteomic GWAS (N total>58,000) and a small childhood proteomic GWAS (N=2,147) as genetic instruments for circulating protein levels. Using two-sample Mendelian randomization, we estimated causal effects of the circulating proteins on childhood body mass index (BMI) in a European GWAS of 39,620 children. MR Wald ratios were calculated to estimate the causal effect of each protein on childhood BMI. Sensitivity analyses testing the MR assumptions included colocalization and phenome-wide association studies (PheWAS). Replication was conducted using independent GWAS datasets, complemented by reverse MR and tissue enrichment analyses. Among 535 tested proteins, three colocalized and demonstrated decreasing effects on BMI per standard deviation increase in their level: endoglin (ENG; MR beta: -0.07, 95% CI [-0.10, -0.04], P=4.4×10⁻ 5 ), fatty acid binding protein 4 (FABP4; MR beta: -0.33, 95% CI [-0.50, -0.16], P=1.3×10⁻ 4 ), and cell adhesion molecule 1 (CADMI1; MR beta: -0.26, 95% CI [-0.37, -0.15], P=5.45×10⁻ 5 ). All three proteins showed evidence of colocalization (posterior probability >75%) and were identified using adult proteomic GWAS, given a limited statistical power using the pediatric proteomic GWAS data. Reverse causation was identified for FABP4, suggesting a compensatory mechanism. In conclusion, we identified three circulating proteins as potential blood biomarkers or drug targets for pediatric obesity, warranting further functional validation to elucidate biological mechanisms and assess therapeutic potential.

Placental MRI texture features in prenatal exposure to opioids and tobacco