Search

Ovarian cancer (OC) is characterised by heterogeneity that complicates the prediction of patient survival and treatment outcomes. Here, we conducted analyses to predict the prognosis of patients from the Genomic Data Commons database and validated the predictions by fivefold cross-validation and by using an independent dataset in the International Cancer Genome Consortium database. We analysed the somatic DNA mutation, mRNA expression, DNA methylation, and microRNA expression data of 1203 samples from 599 serous ovarian cancer (SOC) patients. We found that principal component transformation (PCT) improved the predictive performance of the survival and therapeutic models. Deep learning algorithms also showed better predictive power than the decision tree (DT) and random forest (RF). Furthermore, we identified a series of molecular features and pathways that are associated with patient survival and treatment outcomes. Our study provides perspective on building reliable prognostic and therapeutic strategies and further illuminates the molecular mechanisms of SOC. What is already known on this subject? Recent studies have focussed on predicting cancer outcomes based on omics data. But the limitation is the performance of single-platform genomic analyses or the small numbers of genomic analyses. What do the results of this study add? We analysed multi-omics data, found that principal component transformation (PCT) significantly improved the predictive performance of the survival and therapeutic models. Deep learning algorithms also showed better predictive power than did decision tree (DT) and random forest (RF). Furthermore, we identified a series of molecular features and pathways that are associated with patient survival and treatment outcomes. What are the implications of these findings for clinical practice and/or further research? Our study provides perspective on how to build reliable prognostic and therapeutic strategies and further illuminates the molecular mechanisms of SOC for future studies. [ABSTRACT FROM AUTHOR]

Background: Hepatic encephalopathy (HE) is associated with marked increases in morbidity and mortality for cirrhosis patients. This study aimed to develop and validate machine learning (ML) models to predict 28-day mortality for patients with HE. Methods: A retrospective cohort study was conducted in the Medical Information Mart for Intensive Care (MIMIC)-IV database. Patients from MIMIC-IV were randomized into training and validation cohorts in a ratio of 7:3. Training cohort was used for establishing the model while validation cohort was used for validation. The outcome was defined as 28-day mortality. Predictors were identified by recursive feature elimination (RFE) within 24 h of intensive care unit (ICU) admission. The area under the curve (AUC) and calibration curve were used to determine the predictive performance of different ML models. Results: In the MIMIC-IV database, 601 patients were eventually diagnosed with HE. Of these, 112 (18.64%) experienced death within 28 days. Acute physiology score III (APSIII), sepsis related organ failure assessment (SOFA), international normalized ratio (INR), total bilirubin (TBIL), albumin, blood urea nitrogen (BUN), acute kidney injury (AKI) and mechanical ventilation were identified as independent risk factors. Validation set indicated that the artificial neural network (NNET) model had the highest AUC of 0.837 (95% CI:0.774–0.901). Furthermore, in the calibration curve, the NNET model was also well-calibrated (P = 0.323), which means that it can better predict the 28-day mortality in HE patients. Additionally, the performance of the NNET is superior to existing scores, including Model for End-Stage Liver Disease (MELD) and Model for End-Stage Liver Disease-Sodium (MELD-Na). Conclusions: In this study, the NNET model demonstrated better discrimination in predicting 28-day mortality as compared to other models. This developed model could potentially improve the early detection of HE with high mortality, subsequently improving clinical outcomes in these patients with HE, but further external prospective validation is still required. [ABSTRACT FROM AUTHOR]

Ovarian cancer (OC) is characterised by heterogeneity that complicates the prediction of patient survival and treatment outcomes. Here, we conducted analyses to predict the prognosis of patients from the Genomic Data Commons database and validated the predictions by fivefold cross-validation and by using an independent dataset in the International Cancer Genome Consortium database. We analysed the somatic DNA mutation, mRNA expression, DNA methylation, and microRNA expression data of 1203 samples from 599 serous ovarian cancer (SOC) patients. We found that principal component transformation (PCT) improved the predictive performance of the survival and therapeutic models. Deep learning algorithms also showed better predictive power than the decision tree (DT) and random forest (RF). Furthermore, we identified a series of molecular features and pathways that are associated with patient survival and treatment outcomes. Our study provides perspective on building reliable prognostic and therapeutic strategies and further illuminates the molecular mechanisms of SOC. What is already known on this subject? Recent studies have focussed on predicting cancer outcomes based on omics data. But the limitation is the performance of single-platform genomic analyses or the small numbers of genomic analyses. What do the results of this study add? We analysed multi-omics data, found that principal component transformation (PCT) significantly improved the predictive performance of the survival and therapeutic models. Deep learning algorithms also showed better predictive power than did decision tree (DT) and random forest (RF). Furthermore, we identified a series of molecular features and pathways that are associated with patient survival and treatment outcomes. What are the implications of these findings for clinical practice and/or further research? Our study provides perspective on how to build reliable prognostic and therapeutic strategies and further illuminates the molecular mechanisms of SOC for future studies. [ABSTRACT FROM AUTHOR]

With the gradual popularization of self-driving, it is becoming increasingly important for vehicles to smartly make the right driving decisions and autonomously obey traffic rules by correctly recognizing traffic signs. However, for machine learning-based traffic sign recognition on the Internet of Vehicles (IoV), a large amount of traffic sign data from distributed vehicles is needed to be gathered in a centralized server for model training, which brings serious privacy leakage risk because of traffic sign data containing lots of location privacy information. To address this issue, we first exploit privacy-preserving federated learning to perform collaborative training for accurate recognition models without sharing raw traffic sign data. Nevertheless, due to the limited computing and energy resources of most devices, it is hard for vehicles to continuously undertake complex artificial intelligence tasks. Therefore, we introduce powerful Spike Neural Networks (SNNs) into traffic sign recognition for energy-efficient and fast model training, which is the next generation of neural networks and is practical and well-fitted to IoV scenarios. Furthermore, we design a novel encoding scheme for SNNs based on neuron receptive fields to extract information from the pixel and spatial dimensions of traffic signs to achieve high-accuracy training. Numerical results indicate that the proposed federated SNN outperforms traditional federated convolutional neural networks in terms of accuracy, noise immunity, and energy efficiency as well. [ABSTRACT FROM AUTHOR]

Endometrial carcinoma (EC), a common female reproductive system malignant tumor, affects thousands of people with high morbidity and mortality worldwide. This study was aimed at developing a prediction model for the diagnosis of EC in the general population. First, we obtained datasets GSE63678, GSE106191, and GSE115810 from the Gene Expression Omnibus (GEO) database, dataset GSE17025 from the GEO database, and the RNA sequence of EC from The Cancer Genome Atlas (TCGA) database to constitute the training, test, and validation groups, respectively. Subsequently, the 96 most significantly differentially expressed genes (DEGs) were identified and analyzed for function and pathway enrichment in the training group. Next, we acquired the disease-specific genes by random forest and established an artificial neural network for the diagnosis. Receiver operating characteristic (ROC) curves were utilized to identify the signature across the three groups. Finally, immune infiltration was analyzed to reveal tumor-immune microenvironment (TIME) alterations in EC. The top 96 DEGs (77 down-regulated and 19 up-regulated genes) were primarily enriched in the interleukin-17 signaling pathway, protein digestion and absorption, and transcriptional misregulation in cancer. Subsequently, 14 characterizing genes of EC were identified by random forest. In the training, test, and validation groups, the artificial neural network was constructed with high diagnostic accuracies of 0.882, 0.864, and 0.839, respectively, and areas under the ROC curve (AUCs) of 0.928, 0.921, and 0.782, respectively. Finally, resting and activated mast cells were found to have increased in TIME. We constructed an artificial diagnostic model with excellent reliability for EC and uncovered variations in the immunological ecosystem of EC through integrated bioinformatics approaches, which might be potential diagnostic targets for EC. [ABSTRACT FROM AUTHOR]

To demonstrate the feasibility of automating UED operation and diagnosing the machine performance in real time, a two-stage machine learning (ML) model based on self-consistent start-to-end simulations has been implemented. This model will not only provide the machine parameters with adequate precision, toward the full automation of the UED instrument, but also make real-time electron beam information available as single-shot nondestructive diagnostics. Furthermore, based on a deep understanding of the root connection between the electron beam properties and the features of Bragg-diffraction patterns, we have applied the hidden symmetry as model constraints, successfully improving the accuracy of energy spread prediction by a factor of five and making the beam divergence prediction two times faster. The capability enabled by the global optimization via ML provides us with better opportunities for discoveries using near-parallel, bright, and ultrafast electron beams for single-shot imaging. It also enables directly visualizing the dynamics of defects and nanostructured materials, which is impossible using present electron-beam technologies. [ABSTRACT FROM AUTHOR]

To harness the full potential of the ultrafast electron diffraction (UED) and microscopy (UEM), we must know accurately the electron beam properties, such as emittance, energy spread, spatial-pointing jitter, and shot-to-shot energy fluctuation. Owing to the inherent fluctuations in UED/UEM instruments, obtaining such detailed knowledge requires real-time characterization of the beam properties for each electron bunch. While diagnostics of these properties exist, they are often invasive, and many of them cannot operate at a high repetition rate. Here, we present a technique to overcome such limitations. Employing a machine learning (ML) strategy, we can accurately predict electron beam properties for every shot using only parameters that are easily recorded at high repetition rate by the detector while the experiments are ongoing, by training a model on a small set of fully diagnosed bunches. Applying ML as real-time noninvasive diagnostics could enable some new capabilities, e.g., online optimization of the long-term stability and fine single-shot quality of the electron beam, filtering the events and making online corrections of the data for time-resolved UED, otherwise impossible. This opens the possibility of fully realizing the potential of high repetition rate UED and UEM for life science and condensed matter physics applications. [ABSTRACT FROM AUTHOR]

In this article, we investigate the nanoscale soil-water retention mechanism of unsaturated clay through molecular dynamics and machine learning. Pyrophyllite was chosen due to its stable structure and as the precursor of other 2:1 clay minerals. A series of molecular dynamics simulations of clay at low degrees of saturation were conducted. Soil water was represented by a point cloud through the center-of-mass method. Water-air interface area was measured numerically by the alpha-shape method. The soil-water retention mechanism at the nanoscale was analyzed by distinguishing adsorptive pressure and capillary pressure at different mass water contents and considering the apparent capillary interface area (i.e., water-air interface area per unit water volume). The water number density profile was used to quantify the adsorption effect. A neural-network based machine learning technique was utilized to construct functional relationships among matric suction, the mass water content, and the apparent water-air interface area. Our numerical results have demonstrated from a nanoscale perspective that the adsorption effect is dominated by the van der Waals force and hydroxyl hydration between the clay surface and water. As the mass water content increases, the adsorption pressure decreases, and capillarity plays a prominent role in the soil-water retention mechanism at the nanoscale. [ABSTRACT FROM AUTHOR]

Background: Global myopia prevalence poses a substantial public health burden with vision-threatening complications, necessitating effective prevention and control strategies. Precise prediction of spherical equivalent (SE), myopia, and high myopia onset is vital for proactive clinical interventions. Methods: We reviewed electronic medical records of pediatric and adolescent patients who underwent cycloplegic refraction measurements at the Eye & Ear, Nose, and Throat Hospital of Fudan University between January 2005 and December 2019. Patients aged 3–18 years who met the inclusion criteria were enrolled in this study. To predict the SE and onset of myopia and high myopia in a specific year, two distinct models, random forest (RF) and the gradient boosted tree algorithm (XGBoost), were trained and validated based on variables such as age at baseline, and SE at various intervals. Outputs included SE, the onset of myopia, and high myopia up to 15 years post-initial examination. Age-stratified analyses and feature importance assessments were conducted to augment the clinical significance of the models. Results: The study enrolled 88,250 individuals with 408,255 refraction records. The XGBoost-based SE prediction model consistently demonstrated robust and better performance than RF over 15 years, maintaining an R2 exceeding 0.729, and a Mean Absolute Error ranging from 0.078 to 1.802 in the test set. Myopia onset prediction exhibited strong area under the curve (AUC) values between 0.845 and 0.953 over 15 years, and high myopia onset prediction showed robust AUC values (0.807–0.997 over 13 years, with the 14th year at 0.765), emphasizing the models' effectiveness across age groups and temporal dimensions on the test set. Additionally, our classification models exhibited excellent calibration, as evidenced by consistently low brier score values, all falling below 0.25. Moreover, our findings underscore the importance of commencing regular examinations at an early age to predict high myopia. Conclusions: The XGBoost predictive models exhibited high accuracy in predicting SE, onset of myopia, and high myopia among children and adolescents aged 3–18 years. Our findings emphasize the importance of early and regular examinations at a young age for predicting high myopia, thereby providing valuable insights for clinical practice. [ABSTRACT FROM AUTHOR]

Optically transparent and structurally flexible metamaterial absorbers (MMAs) are widely used in many practical applications. However, the realization of such MMAs requires both thickness reduction and bandwidth enhancement, which remains a challenge task. As a multi-objective optimization problem, it is a time-consuming and resource-demanding process. Guided by prior-knowledge, two degrees of freedom are taken into account in the design, namely the structure of unit-cells and their spatial arrangement. In this way, the multi-objective optimization problem is simplified into a two-step problem. A general procedure based on machine-learning (ML) method is proposed to solve this problem. First, an artificial neural network (ANN) is used to map the configuration parameters of a type of meta-atom onto its reflection coefficients; then, a dynamic graph convolutional neural network (DGCNN) is trained to realize the synthesis of quasi-periodic distributed MMA array; finally, combined with the differential evolution (DE) algorithm, the optimized configuration of the meta-atoms and their optimum distribution are obtained. Based on the proposed procedure, an ultrathin optically transparent and flexible MMA is designed, fabricated and further verified by experiments. The designed MMA realizes 87% absorption bandwidth covering 6.22 GHz–19.42 GHz (103% relative bandwidth) with the thickness only 3. 3 mm (0. 068 λ L) , which approaches the theoretical limit 1 / 17 λ L . And its averaged transparency is about 63% at the wavelength range 500 – 800 mm. The whole design process is achieved 50 times faster than the conventional full-wave simulation, which convincingly demonstrates the superiority of the proposed method. [ABSTRACT FROM AUTHOR]

With rapid urbanization and industrialization, the number of light-duty gasoline vehicles (LDGVs) in China has continued to grow rapidly, leading to a significant increase in traffic pollution. Therefore, it is essential to accurately calculate the emission of LDGVs for air quality monitoring and management. Fortunately, Motor Vehicle Emission Simulator (MOVES) is a sophisticated model for estimating mobile source emissions with good prediction accuracy. However, the parameters of MOVES are based on the field tests in the US, which is worth exploring whether MOVES can be applied to other countries. Hence, in this paper, we used the portable emission measurement system (PEMS) to conduct real driving emission (RDE) tests of LDGVs, aiming to explore the possibility of the MOVES application in China. Based on the field tests, we modified basic parameters in the MOVES model, but unsatisfactory prediction performance was obtained. Existing research on improving MOVES performance mainly involved new binning of operating modes, but these methods had limited improvements. Though studies have also used machine learning methods for predicting LDGV emissions, they lacked comparison and integration with the MOVES model. To further improve the prediction accuracy, we proposed a novel road vehicle emission model that integrated the machine learning method and the MOVES to predict the road-level emission rates of NO and CO 2 emissions of LDGVs. In addition, we employed the Boruta algorithm to capture the key influencing factors and promote prediction performance. The enhanced model outperformed MOVES and achieved higher R2 values. On average, the improvement for CO 2 was 0.132, and for NO, it was 0.261. This work will provide references for MOVES improvements in practical scenarios and better predict pollutant emissions for LDGVs using limited resources of field tests in cities outside the US. • Integrating engine characteristics with the MOVES model can improve model performance. • Integrating machine learning with MOVES performs well in prediction of vehicle CO 2 and NO emission. • The proposed emission model provides a possibility for calculation rather than costly an experiment in countries outside the USA. [ABSTRACT FROM AUTHOR]

Chronic nonbacterial osteomyelitis (CNO) is an autoinflammatory bone disorder. The origin and development of CNO involve many complex immune processes, resulting in delayed diagnosis and a lack of effective treatment. Although bioinformatics analysis has been utilized to seek key genes and pathways in CNO, only a few bioinformatics studies that focus on CNO pathogenesis and mechanisms have been reported. This study aimed to identify key biomarkers that could serve as early diagnostic or therapeutic markers for CNO. Two RNA-seq datasets (GSE133378 and GSE187429) were obtained from the Gene Expression Omnibus (GEO). Weighted gene coexpression network analysis (WGCNA) and differentially expressed gene (DEG) analysis were conducted to identify the genes associated with CNO. Then, the autoinflammatory genes most associated with CNO were identified based on the GeneCards database and a CNO prediction model, which was created by the LASSO machine learning algorithm. The accuracy of the model and effects of the autoinflammatory genes according to receiver operating characteristic (ROC) curves were verified in external datasets (GSE7014). Finally, we performed clustering analysis with ConsensusClusterPlus. In total, eighty CNO-related genes were identified and were significantly enriched in the biological processes regulation of actin filament organization, cell–cell junction organization and gamma-catenin binding. The main enriched pathways were adherens junctions, viral carcinogenesis and systemic lupus erythematosus. Two autoinflammatory genes with high expression in CNO samples were identified by combining an optimal machine learning algorithm (LASSO) with the GeneCards database. An external validation dataset (GSE187429) was utilized for ROC analysis of the prediction model and two genes, and the results indicated good efficiency. Then, based on consensus clustering analysis, we found that the expression of UTS2 and MPO differed between clusters. Finally, the ceRNA network of lncRNAs and the small molecule compounds targeting the two autoinflammatory genes were predicted. The identification of two autoinflammatory genes, the HCG18/has-mir-147a/UTS2/MPO axis and signalling pathways in this study can help us understand the molecular mechanism of CNO formation and provides candidate targets for the diagnosis and treatment of CNO. [ABSTRACT FROM AUTHOR]

Background: Functioning and non-functioning adrenocortical adenoma are two subtypes of benign adrenal adenoma, and their differential diagnosis is crucial. Current diagnostic procedures use an invasive method, adrenal venous sampling, for endocrinologic assessment. Methods: This study proposes establishing an accurate differential model for subtyping adrenal adenoma using computed tomography (CT) radiomic features and machine learning (ML) methods. Dataset 1 (289 patients with adrenal adenoma) was collected to develop the models, and Dataset 2 (54 patients) was utilized for external validation. Cuboids containing the lesion were cropped from the non-contrast, arterial, and venous phase CT images, and 1,967 features were extracted from each cuboid. Ten discriminative features were selected from each phase or the combined phases. Random forest, support vector machine, logistic regression (LR), Gradient Boosting Machine, and eXtreme Gradient Boosting were used to establish prediction models. Results: The highest accuracies were 72.7%, 72.7%, and 76.1% in the arterial, venous, and non-contrast phases, respectively, when using radiomic features alone with the ML classifier of LR. When features from the three CT phases were combined, the accuracy of LR reached 83.0%. After adding clinical information, the area under the receiver operating characteristic curve increased for all the machine learning methods except for LR. In Dataset 2, the accuracy of LR was the highest, reaching 77.8%. Conclusion: The radiomic features of the lesion in three-phase CT images can potentially suggest the functioning or non-functioning nature of adrenal adenoma. The resulting radiomic models can be a non-invasive, low-cost, and rapid method of minimizing unnecessary testing in asymptomatic patients with incidentally discovered adrenal adenoma. [ABSTRACT FROM AUTHOR]

Aerosols are complex compounds that greatly affect the global radiation balance and climate system and even human health; in addition, aerosols are currently a large source of uncertainty in the numerical simulation process. The arid and semi-arid areas have fragile ecosystems with abundant dust but lack related high-accuracy aerosol data. To solve these problems, we use the bagging trees ensemble model, based on 1 km aerosol optical depth (AOD) data and multiple environmental covariates, to produce a monthly advanced-performance, full-coverage, and high-resolution (250 m) AOD product (named FEC AOD, fusing environmental covariates AOD) covering the arid and semi-arid areas. Then, based on the FEC AOD products, we analyzed the spatiotemporal AOD pattern and further discussed the interpretation of environmental covariates to AOD. The results show that the bagging trees ensemble model has a good performance, with its verification R2 values always remaining at 0.90 and the R2 being 0.79 for FEC AOD compared with AERONET AOD product. The high-AOD areas are located in the Taklimakan Desert and on the Loess Plateau, and the low-AOD areas are concentrated in southern Qinghai province. The higher the AOD, the stronger the interannual variability. Interestingly, the AOD reflected a dramatic decrease on the Loess Plateau and an evident increase in the south-eastern Taklimakan Desert, while the southern Qinghai province AODs showed almost no significant change between 2000 and 2019. The annual variation characteristics show that the AOD was largest in spring (0.267±0.200) and smallest in autumn (0.147±0.089); the annual AOD variation pattern showed different features, with two peaks in March and August over Gansu province but only one peak in April in other provinces/autonomous regions. Farmlands and construction lands have high AOD levels compared to other land cover types. Meteorological factors demonstrate the maximum interpretation ability of the AODs on all set temporal scales, followed by the terrain factors, while surface properties have the smallest explanatory abilities; the corresponding average contributions are 77.1 %, 59.1 %, and 50.4 %, respectively. The capability of the environmental covariates to explain the AOD varies seasonally in the following sequence: winter (86.6 %) > autumn (80.8 %) > spring (79.9 %) > summer (72.5 %). In this research, we provide a pathbreaking high spatial resolution (250 m) and long time series (2000–2019) FEC AOD dataset covering arid and semi-arid regions to support atmospheric and related studies in northwest China; the full dataset is available at 10.5281/zenodo.5727119 (Chen et al., 2021b). [ABSTRACT FROM AUTHOR]

There was a mistake in the value of I V i in the righthand column of Table 1. Table 1 Typical properties of manual optimization and machine-learning optimization. [Extracted from the article]

The structure integrity of reactor pressure vessels (RPV) plays an important role in the safety of nuclear power plants, which mainly concerns the risk of sealing failure. In the present work, a hybrid method consisting of an artificial neural network (ANN) and multi-objective particle swarm optimization (MOPSO) is proposed, aiming to find an optimized RPV structure with improved sealing performance. The ANN model provides a more accurate prediction of sealing performance than the radial basis function neural network (RBFNN) model. There is also a better generalization ability in the ANN model with R 2 above 0.96 and RMSE below 0.06 in the test set. With ANN as a surrogate model, multi-objective optimization is applied in RPV structure design. Compared with that of the original RPV structure, the sealing performance can be improved by even over 25% in the optimized structure. The application of ANN surrogate models can save more than 99% computation cost, compared with optimization based on finite element analysis (FEA). • The RPV sealing performance is predicted accurately by ANN surrogate model. • A MOPSO method with Pareto front is used for the optimization design. • By ANN surrogate model, the optimization time cost is reduced by over 99%. • The sealing performance is improved by over 25% in the optimized structure. [ABSTRACT FROM AUTHOR]