Predictive Performance Research Articles

Improving embedding-based link prediction performance using clustering

Incomplete knowledge graphs are common problem that can impair task accuracy. As knowledge graphs grow extensively, the probability of incompleteness increases. Link prediction addresses this problem, but accurate and efficient link prediction methods are needed to handle incomplete and extensive knowledge graphs. This study proposed modifications to the embedding-based link prediction using clustering to improve performance. The proposed method involves four main processes: embedding, clustering, determining clusters, and scoring. Embedding converts entities and relations into vectors while clustering groups these vectors. Selected clusters are determined based on the shortest distance between the centroid and the incomplete knowledge graph. Scoring measures relation rankings, and link prediction result is selected based on highest scores. The link prediction performance is evaluated using Hits@1, Mean Rank, Mean Reciprocal Rank and prediction time on three knowledge graph datasets: WN11, WN18RR, and FB13. The link prediction methods used are TransE and ComplEx, with BIRCH as the clustering technique and Mahalanobis for short-distance measurement. The proposed method significantly improves link prediction performance, achieving accuracy up to 98% and reducing prediction time by 99%. This study provides effective and efficient solution for improving link prediction, demonstrating high accuracy and efficiency in handling incomplete and extensive knowledge graphs.

Prediction of fresh herbage yield using data mining techniques with limited plant quality parameters

The purpose of this study was to ascertain the fresh herbage yield, fertilizer dosage, and plant characteristics of the Sorghum-Sudangrass hybrid grown in arid and semi-arid regions, as well as their interrelationships. For this reason, data from the Sorghum-Sudangrass hybrid were used to assess the predictive performance of several data mining techniques, including CHAID, CART, MARS, and Bagging MARS. Plant traits were measured in Konya and Sanliurfa during 2021 and 2022. The descriptive statistical values were calculated as follows: plant height 306.27 cm, stem diameter 9.47 mm, fresh herbage yield 10852.51 kg/da, crude protein ratio 9.66%, acid detergent fiber 33.39%, neutral detergent fiber 51.85%, acid detergent lignin 9.76%, dry matter digestibility 62.88%, dry matter intake 2.34%, and relative feed value 114.68 (average values). The predictive capacities of the fitted models were assessed using model fit statistics such as the coefficient of determination (R²), adjusted R², root mean square error (RMSE), mean absolute percentage error (MAPE), standard deviation ratio (SD ratio), and Akaike Information Criterion (AIC). With the lowest values for RMSE, MAPE, SD ratio, and AIC (246, 1.926, 0.085, and 845, respectively), and the highest R² value (0.993) and adjusted R² value (0.989), the MARS algorithm was determined to be the best model for characterizing fresh herbage yield. As a solid alternative to other data mining techniques, the MARS algorithm was shown to be the most appropriate model for forecasting fresh herbage production.

A transfer learning based patent transfer prediction model: Discussion about the condition of China and U.S.

In the same patent classification, there are significant differences in the patent transfer quantity for different countries. For the countries with small patent transfer quantities, the performance of the patent transfer prediction is worse than that of high-transfer countries. To improve the performance of the patent prediction model learned for the low-transfer countries, we propose a transfer learning based patent transfer prediction model, for which the high-transfer countries are the source domain and the low-transfer countries are the target domain. In the model, the textual feature of each patent text is embedded using BERT-CNN with the parameters fine-tuned on the target domain from the source domain. The textual features are optimized to minimize the domain loss of Maximum Mean Discrepancy between source and target domain. What is more, we extract graph feature vectors by embedding the patent node in a three-layer heterogeneous network which consists of patent nodes, International Patent Classification nodes, consumer nodes, consumer location nodes and their relations. Finally, the patent text vectors, structural feature vectors, and graph feature vectors are spliced together to predict target domain patents transfer or not. The performance of the model is verified by some comparative experiments. The experimental results show that the proposed model outperforms significantly the baseline models in the values of Precision, Recall, and F1 of patent data improve by 6.47%, 6.54%, and 6.58% on average, respectively.

Development, evaluation and comparison of machine learning algorithms for predicting in-hospital patient charges for congestive heart failure exacerbations, chronic obstructive pulmonary disease exacerbations and diabetic ketoacidosis

BackgroundHospitalizations for exacerbations of congestive heart failure (CHF), chronic obstructive pulmonary disease (COPD) and diabetic ketoacidosis (DKA) are costly in the United States. The purpose of this study was to predict in-hospital charges for each condition using machine learning (ML) models.ResultsWe conducted a retrospective cohort study on national discharge records of hospitalized adult patients from January 1st, 2016, to December 31st, 2019. We constructed six ML models (linear regression, ridge regression, support vector machine, random forest, gradient boosting and extreme gradient boosting) to predict total in-hospital cost for admission for each condition. Our models had good predictive performance, with testing R-squared values of 0.701-0.750 (mean of 0.713) for CHF; 0.694-0.724 (mean 0.709) for COPD; and 0.615-0.729 (mean 0.694) for DKA. We identified important key features driving costs, including patient age, length of stay, number of procedures, and elective/nonelective admission.ConclusionsML methods may be used to accurately predict costs and identify drivers of high cost for COPD exacerbations, CHF exacerbations and DKA. Overall, our findings may inform future studies that seek to decrease the underlying high patient costs for these conditions.

Adaptive incentive mechanism with predictors for on-time attended home delivery problem

The widespread use of the Internet and smart devices has led to a fast growth in online shopping, offering new chances for online retailers to boost profits. However, this expansion has also brought various challenges, such as the heavy workload faced by delivery riders. To meet customers’ delivery time preferences and increase earnings, riders often work long hours, especially during busy periods. This study explores how historical delivery data can be used to balance workload in on-time attended home delivery. Drawing on the actual delivery operations and data of an online shopping platform, we propose a framework that combines delivery demand and customer behavior predictors with an adaptive incentive system to balance rider workload. Specifically focusing on same-day attended home delivery, we introduce a method to forecast future delivery demand, an algorithm to estimate customer choice behavior using a simple model, and an adaptive incentive system to influence customer decisions and achieve workload balance. We show that as order volume increases, the proposed incentive system achieves the pre-determined workload target. Using real data, we conduct numerical experiments which not only underscore the superior predictive performance of our models but also affirm the efficacy of the proposed incentive structure.

Biomarkers of COVID-19 short-term worsening: a multiparameter analysis within the prospective multicenter COVIDeF cohort.

During a pandemic like COVID-19, hospital resources are constrained and accurate severity triage of the patients is required. The objective of this study is to estimate the predictive performances of candidate biomarkers for short-term worsening (STW) of COVID-19. Prospective, multicenter (20 hospitals in Paris) cohort study of consecutive COVID-19 patients with systematic biobanking at admission, during the first waves of COVID-19 in France in 2020 (COVIDeF cohort). Consecutive COVID-19 patients were screened for inclusion. They were excluded in presence of severity criteria defined by either an ICU admission, mechanical ventilation (including noninvasive ventilation), acute respiratory distress, or in-hospital death before sampling. Routine blood tests measured during usual care and centralized systematic measurement of creatine kinase, C-reactive protein (CRP), procalcitonin, soluble urokinase plasminogen activator receptor (suPAR), high-sensitive troponin T (TnT-hs), N terminal pro-B natriuretic peptide (NT-proBNP), calprotectin, platelet factor 4, mid-regional pro-adrenomedullin (MR-proADM), and proendothelin were performed. The primary outcome was STW, defined by a severity criteria within 7 days. A backward stepwise logistic regression model and a 'best subset' approach were used to identify independent association, and the area under the receiving operator characteristics (AUROC) was computed. Five hundred and eleven patients were analyzed, of whom 60 (11.7%) experienced STW. Median time to occurrence of a severity criteria was 3 days. At admission, lower values of eosinophils, lymphocytes, platelets, alanine aminotransferase, and higher values of neutrophils, creatinine, urea, CRP, TnT-hs, suPAR, NT-proBNP, calprotectin, procalcitonin, MR-proADM, and proendothelin were predictive of worsening. Stepwise logistic regression identified three biomarkers significantly associated with worsening: CRP [adjusted odds ratio (aOR): 1.10, 95% confidence interval (95% CI): 1.06-1.15 for a 10-unit increase, AUROC: 0.73 (0.66-0.79)], procalcitonin [aOR: 0.42, 95% CI: 0.22-0.81, AUROC: 0.69 (0.64-0.88)], and MR-proADM [aOR: 2.85, 95% CI: 1.74-4.69, AUROC: 0.75 (0.69-0.81)]. These biomarkers outperformed clinical variables except diabetes and cancer comorbidities. In this multicenter prospective study that assessed a large panel of biomarkers for COVID-19 patients, CRP, procalcitonin, and MR-proADM were independently associated with the risk of STW. ClinicalTrials.gov NCT04352348.

Predictive performance of urinalysis for urine culture results according to causative microorganisms: an integrated analysis with artificial intelligence.

Urinary tract infections (UTIs) are pervasive and prevalent in both community and hospital settings. Recent trends in the changes of the causative microorganisms in these infections could affect the effectiveness of urinalysis (UA). We aimed to evaluate the predictive performance of UA for urinary culture test results according to the causative microorganisms. In addition, UA results were integrated with artificial intelligence (AI) methods to improve the predictive power. A total of 360,376 suspected UTI patients were enrolled from two university hospitals and one commercial laboratory. To ensure broad model applicability, only a limited range of clinical data available from commercial laboratories was used in the analyses. Overall, 53,408 (14.8%) patients were identified as having a positive urine culture. Among the UA tests, the combination of leukocyte esterase and nitrite tests showed the highest area under the curve (AUROC, 0.766; 95% CI, 0.764-0.768) for predicting urine culture positivity but performed poorly for Gram-positive bacteriuria (0.642; 0.637-0.647). The application of an AI model improved the predictive power of the model for urine culture results to an AUROC of 0.872 (0.870-0.875), and the model showed superior performance metrics not only for Gram-negative bacteriuria (0.901; 0.899-0.902) but also for Gram-positive bacteriuria (0.745; 0.740-0.749) and funguria (0.872; 0.865-0.879). As the prevalence of non-Escherichia coli-caused UTIs increases, the performance of UA in predicting UTIs could be compromised. The addition of AI technologies has shown potential for improving the predictive performance of UA for urine culture results.IMPORTANCEUA had good performance in predicting urine culture results caused by Gram-negative bacteria, especially for Escherichia coli and Pseudomonas aeruginosa bacteriuria, but had limitations in predicting urine culture results caused by Gram-positive bacteria, including Streptococcus agalactiae and Enterococcus faecalis. We developed and externally validated an AI model incorporating minimal demographic information of patients (age and sex) and laboratory data for UA, complete blood count, and serum creatinine concentrations. The AI model exhibited improved performance in predicting urine culture results across all the causative microorganisms, including Gram-positive bacteria, Gram-negative bacteria, and fungi.

Empirical Bayesian Kriging, a Robust Method for Spatial Data Interpolation of a Large Groundwater Quality Dataset from the Western Netherlands

No single spatial interpolation method reigns supreme for modelling the precise spatial distribution of groundwater quality data. This study addresses this challenge by evaluating and comparing several commonly used geostatistical methods: Local Polynomial Interpolation (LPI), Ordinary Kriging (OK), Simple Kriging (SK), Universal Kriging (UK), and Empirical Bayesian Kriging (EBK). We applied these methods to a vast dataset of 3033 groundwater records encompassing a substantial area (11,100 km2) in the coastal lowlands of the western Netherlands. To our knowledge, no prior research has investigated these interpolation methods in this specific hydrogeological setting, exhibiting a range of groundwater qualities, from fresh to saline, often anoxic, with high natural concentrations of PO4 and NH4. The prediction performance of the interpolation methods was assessed through statistical indicators such as root means square error. The findings indicated that EBK outperforms the other geostatistical methods in forecasting groundwater quality for the five variables considered: Cl, SO4, Fe, PO4, and NH4. In contrast, SK performed worst for the species except for SO4. We recommend not using SK to interpolate groundwater quality species unless the data exhibit low spatial variation, high sample density, or evenly distributed sampling.

RadMLBench: A dataset collection for benchmarking in radiomics

BackgroundNew machine learning methods and techniques are frequently introduced in radiomics, but they are often tested on a single dataset, which makes it challenging to assess their true benefit. Currently, there is a lack of a larger, publicly accessible dataset collection on which such assessments could be performed. In this study, a collection of radiomics datasets with binary outcomes in tabular form was curated to allow benchmarking of machine learning methods and techniques. MethodsA variety of journals and online sources were searched to identify tabular radiomics data with binary outcomes, which were then compiled into a homogeneous data collection that is easily accessible via Python. To illustrate the utility of the dataset collection, it was applied to investigate whether feature decorrelation prior to feature selection could improve predictive performance in a radiomics pipeline. ResultsA total of 50 radiomic datasets were collected, with sample sizes ranging from 51 to 969 and 101 to 11165 features. Using this data, it was observed that decorrelating features did not yield any significant improvement on average. ConclusionsA large collection of datasets, easily accessible via Python, suitable for benchmarking and evaluating new machine learning techniques and methods was curated. Its utility was exemplified by demonstrating that feature decorrelation prior to feature selection does not, on average, lead to significant performance gains and could be omitted, thereby increasing the robustness and reliability of the radiomics pipeline.

Multimodal radiomics-based methods using deep learning for prediction of brain metastasis in non-small cell lung cancer with 18F-FDG PET/CT images

Objective. Approximately 57% of non-small cell lung cancer (NSCLC) patients face a 20% risk of brain metastases (BMs). The delivery of drugs to the central nervous system is challenging because of the blood–brain barrier, leading to a relatively poor prognosis for patients with BMs. Therefore, early detection and treatment of BMs are highly important for improving patient prognosis. This study aimed to investigate the feasibility of a multimodal radiomics-based method using 3D neural networks trained on 18F-FDG PET/CT images to predict BMs in NSCLC patients. Approach. We included 226 NSCLC patients who underwent 18F-FDG PET/CT scans of areas, including the lung and brain, prior to EGFR-TKI therapy. Moreover, clinical data (age, sex, stage, etc) were collected and analyzed. Shallow lung features and deep lung-brain features were extracted using PyRadiomics and 3D neural networks, respectively. A support vector machine (SVM) was used to predict BMs. The receiver operating characteristic (ROC) curve and F1 score were used to assess BM prediction performance. Main result. The combination of shallow lung and shallow-deep lung-brain features demonstrated superior predictive performance (AUC = 0.96 ± 0.01). Shallow-deep lung-brain features exhibited strong significance (P < 0.001) and potential predictive performance (coefficient > 0.8). Moreover, BM prediction by age was significant (P < 0.05). Significance. Our approach enables the quantitative assessment of medical images and a deeper understanding of both superficial and deep tumor characteristics. This noninvasive method has the potential to identify BM-related features with statistical significance, thereby aiding in the development of targeted treatment plans for NSCLC patients.

Identification of Compounds That Contribute to Consumer Aroma Liking of Roasted American-European Hybrid Hazelnuts.

New interspecific hybrid hazelnut crosses between American (Corylus americana) and European (Corylus avellana) hazelnuts are being developed to support a commercial industry in the Midwest region of the United States. In this study, volatile compounds that impact consumer aroma liking of roasted hybrid hazelnuts (C. americana × C. avellana) were investigated by targeted and nontargeted GC/MS flavoromics. Chemical profiles from 10 roasted hybrid hazelnut samples were modeled with consumer aroma liking scores by orthogonal partial least-squares with good fit and predictive performance (R2 ≥ 0.92, Q2 ≥ 0.82, RMSECV = 0.2). Top ranked predictors positively correlated with liking included 12 aroma compounds and 4 profiled volatiles for the targeted and nontargeted methods, respectively. Sensory recombination testing of hazelnut samples with addition of the 12 predictive odorants was preferred by consumers (p < 0.001, Δ aroma liking = 2.2 on 9-point scale) and perceived as more roasty, nutty, and sweet compared to the control (p < 0.05). Addition of the 4 predictive volatiles at subthreshold levels also was preferred (p = 0.02) and perceived as less earthy and mushroom like than the control (p < 0.05).

Forecasting crude oil returns in different degrees of ambiguity: Why machine learn better?

Numerous studies have demonstrated the strong out-of-sample predictive ability of machine learning models, particularly in variable selection and dimension reduction, on crude oil price returns. We find significant disparities in out-of-sample predictive performance between these two methods under varying degrees of ambiguity, a fuzziness measure proposed by Izhakian (2020), independent of outcomes, risks, and attitudes. Variable selection methods exhibit strong out-of-sample predictive power in low ambiguity environments, but weaker performance in high ambiguity environments, whereas dimension reduction methods show the opposite pattern. Furthermore, the optimal penalty coefficient selected by variable selection methods during in-sample model fitting is highly correlated with ambiguity, indicating that the predictive ability of variable selection stems from its ability to accurately identify the correct predictors.

Using machine learning methods to investigate the impact of age on the causes of death in patients with early intrahepatic cholangiocarcinoma who underwent surgery.

The impact of age on the causes of death (CODs) in patients with early-stage intrahepatic cholangiocarcinoma (ICC) who had undergone surgery was analyzed in this study. A total of 1555 patients (885 in the older group and 670 in the younger group) were included in this study. Before and after applying inverse probability of treatment weighting (IPTW), the different CODs in the 2 groups were further investigated. Additionally, 7 different machine learning models were used as predictive tools to identify key variables, aiming to evaluate the therapeutic outcome in early ICC patients undergoing surgery. Before (5.92 vs. 4.08years, P < 0.001) and after (6.00 vs. 4.08years, P < 0.001) IPTW, the younger group consistently showed longer overall survival (OS) compared with the older group. Before IPTW, there were no significant differences in cholangiocarcinoma-related deaths (CRDs, P = 0.7) and secondary malignant neoplasms (SMNs, P = 0.78) between the 2 groups. However, the younger group had a lower cumulative incidence of cardiovascular disease (CVD, P = 0.006) and other causes (P < 0.001) compared with the older group. After IPTW, there were no differences between the 2 groups in CRDs (P = 0.2), SMNs (P = 0.7), and CVD (P = 0.1). However, the younger group had a lower cumulative incidence of other CODs compared with the older group (P < 0.001). The random forest (RF) model showed the highest C-index of 0.703. Time-dependent variable importance bar plots showed that age was the most important factor affecting the 2-, 4-, and 6-year survival, followed by stage and size. Our study confirmed that younger patients have longer OS compared with older patients. Further analysis of the CODs indicated that older patients are more likely to die from CVDs. The RF model demonstrated the best predictive performance and identified age as the most important factor affecting OS in early ICC patients undergoing surgery.

Effect of sintering temperature on feature resolution and flexural strength of ceramics fabricated through vat photopolymerization additive manufacturing

PurposeAlthough ceramic additive manufacturing (AM) could be used to fabricate complex, high-resolution parts for diverse, functional applications, one ongoing challenge is optimizing the post-process, particularly sintering, conditions to consistently produce geometrically accurate and mechanically robust parts. This study aims to investigate how sintering temperature affects feature resolution and flexural properties of silica-based parts formed by vat photopolymerization (VPP) AM.Design/methodology/approachTest artifacts were designed to evaluate features of different sizes, shapes and orientations, and three-point bend specimens printed in multiple orientations were used to evaluate mechanical properties. Sintering temperatures were varied between 1000°C and 1300°C.FindingsDeviations from designed dimensions often increased with higher sintering temperatures and/or larger features. Higher sintering temperatures yielded parts with higher strength and lower strain at break. Many features exhibited defects, often dependent on geometry and sintering temperature, highlighting the need for further analysis of debinding and sintering parameters.Originality/valueTo the best of the authors’ knowledge, this is the first time test artifacts have been designed for ceramic VPP. This work also offers insights into the effect of sintering temperature and print orientation on flexural properties. These results provide design guidelines for a particular material, while the methodology outlined for assessing feature resolution and flexural strength is broadly applicable to other ceramics, enabling more predictable part performance when considering the future design and manufacture of complex ceramic parts.

Development and validation of machine-learning models of diet management for hyperphenylalaninemia: a multicenter retrospective study

BackgroundAssessing dietary phenylalanine (Phe) tolerance is crucial for managing hyperphenylalaninemia (HPA) in children. However, traditionally, adjusting the diet requires significant time from clinicians and parents. This study aims to investigate the development of a machine-learning model that predicts a range of dietary Phe intake tolerance for children with HPA over 10 years following diagnosis.MethodsIn this multicenter retrospective observational study, we collected the genotypes of phenylalanine hydroxylase (PAH), metabolic profiles at screening and diagnosis, and blood Phe concentrations corresponding to dietary Phe intake from over 10 years of follow-up data for 204 children with HPA. To incorporate genetic information, allelic phenotype value (APV) was input for 2965 missense variants in the PAH gene using a predicted APV (pAPV) model. This model was trained on known pheno-genotype relationships from the BioPKU database, utilizing 31 features. Subsequently, a multiclass classification model was constructed and trained on a dataset featuring metabolic data, genetic data, and follow-up data from 3177 events. The final model was fine-tuned using tenfold validation and validated against three independent datasets.ResultsThe pAPV model achieved a good predictive performance with root mean squared error (RMSE) of 1.53 and 2.38 on the training and test datasets, respectively. The variants that cause amino acid changes in the region of 200–300 of PAH tend to exhibit lower pAPV. The final model achieved a sensitivity range of 0.77 to 0.91 and a specificity range of 0.8 to 1 across all validation datasets. Additional assessment metrics including positive predictive value (0.68–1), negative predictive values (0.8–0.98), F1 score (0.71–0.92), and balanced accuracy (0.8–0.92) demonstrated the robust performance of our model.ConclusionsOur model integrates metabolic and genetic information to accurately predict age-specific Phe tolerance, aiding in the precision management of patients with HPA. This study provides a potential framework that could be applied to other inborn errors of metabolism.

Accurately identifying positive and negative regulation of apoptosis using fusion features and machine learning methods

Apoptotic proteins play a crucial role in the apoptosis process, ensuring a balance between cell proliferation and death. Thus, further elucidating the regulatory mechanisms of apoptosis will enhance our understanding of their functions. However, the development of computational methods to accurately identify positive and negative regulation of apoptosis remains a significant challenge. This work proposes a machine learning model based on multi-feature fusion to effectively identify the roles of positive and negative regulation of apoptosis. Initially, we constructed a reliable benchmark dataset containing 200 positive regulation of apoptosis and 241 negative regulation of apoptosis proteins. Subsequently, we developed a classifier that combines the support vector machine (SVM) with pseudo composition of k-spaced amino acid pairs (PseCKSAAP), composition transition distribution (CTD), dipeptide deviation from expected mean (DDE), and PSSM-composition to identify these proteins. Analysis of variance (ANOVA) was employed to select optimized features that could yield the maximum prediction performance. Evaluating the proposed model on independent data revealed and achieved an accuracy of 0.781 with an AUROC of 0.837, demonstrating our model's potent capabilities.

Band configurations and seasonality influence the predictions of common boreal tree species using UAS image data

Key messageData acquisition of remote sensing products is an essential component of modern forest inventories. The quality and properties of optical remote sensing data are further emphasised in tree species-specific inventories, where the discrimination of different tree species is based on differences in their spectral properties. Furthermore, phenology affects the spectral properties of both evergreen and deciduous trees through seasons. These confounding factors in both sensor configuration and timing of data acquisition can result in unexpectedly complicated situations if not taken into consideration. This paper examines how the timing of data acquisition and sensor properties influence the prediction of tree species proportions and volumes in a boreal forest area dominated by Norway spruce and Scots pine, with a smaller presence of deciduous trees.ContextThe effectiveness of remote sensing for vegetation mapping depends on the properties of the survey area, mapping objectives and sensor configuration.AimsThe objective of this study was to investigate the plot-level relationship between seasonality and different optical band configurations and prediction performance of common boreal tree species. The study was conducted on a 40-ha study area with a systematically sampled circular field plots.MethodsTree species proportions (0–1) and volumes (m3 ha−1) were predicted with repeated remote sensing data collections in three stages of the growing season: prior (spring), during (summer) and end (autumn). Sensor band configurations included conventional RGB and multispectral (MS). The importance of different wavelengths (red, green, blue, near-infrared and red-edge) and predictive performance of the different band configurations were analysed using zero–one-inflated beta regression and Gaussian process regression.ResultsPrediction errors of broadleaves were most affected by band configuration, MS data resulting in lower prediction errors in all seasons. The MS data exhibited slightly lower prediction errors with summer data acquisition compared to other seasons, whereas this period was found to be less suitable for RGB data.ConclusionThe MS data was found to be much less affected by seasonality than the RGB data. Spring was found to be the least optimal season to collect MS and RGB data for tree species-specific predictions.

Short-term wind power prediction method based on multivariate signal decomposition and RIME optimization algorithm

Accurate characterization of the dynamics of wind power systems under highly unstable conditions is essential for short-term wind power forecasting. Multivariate factors, such as wind speed, wind direction and climatic factors at different heights, as well as the relationships between them, need to be fully considered. To this end, after the Pearson correlation coefficient is used to identify the main influencing variables, the empirical modal decomposition (MEMD) method is utilized to decompose the data in order to capture the deep coupling relationship and the spatio-temporal pattern of change among the variables. Combined with a bidirectional long and short-term memory network (BiLSTM) optimized through the RIME algorithm, a short-term wind power prediction model was constructed. Data analysis of two wind farms in Xinjiang and Gansu, China, shows that the model has higher prediction accuracy compared to the baseline model. The effectiveness of RIME and MEMD is verified through component ablation experiments and comparison experiments with other optimization algorithms and signal decomposition algorithms. To cope with large-scale data, dimensionality reduction of IMF features using Elastic Net (EN) is explored to reduce computational requirements and maintain good prediction performance. The method provides strong support for power system scheduling and efficient utilization of renewable energy.

Serum amyloid A for predicting prognosis in patients with newly diagnosed Crohn’s disease

ObjectiveSerum amyloid A (SAA) was found to be positively correlated with the activity of Crohn’s disease (CD); however, its prognostic value remains uncertain. Here, we examined its predictive ability in...

Modelling soil prokaryotic traits across environments with the trait sequence database ampliconTraits and the R package MicEnvMod

We present a comprehensive, customizable workflow for inferring prokaryotic phenotypic traits from marker gene sequences and modelling the relationships between these traits and environmental factors, thus overcoming the limited ecological interpretability of marker gene sequencing data. We created the trait sequence database ampliconTraits, constructed by cross-mapping species from a phenotypic trait database to the SILVA sequence database and formatted to enable seamless classification of environmental sequences using the SINAPS algorithm. The R package MicEnvMod enables modelling of trait – environment relationships, combining the strengths of different model types and integrating an approach to evaluate the models' predictive performance in a single framework. Traits could be accurately predicted even for sequences with low sequence identity (80 %) with the reference sequences, indicating that our approach is suitable to classify a wide range of environmental sequences. Validating our approach in a large trans-continental soil dataset, we showed that trait distributions were robust to classification settings such as the bootstrap cutoff for classification and the number of discrete intervals for continuous traits. Using functions from MicEnvMod, we revealed precipitation seasonality and land cover as the most important predictors of genome size. We found Pearson correlation coefficients between observed and predicted values up to 0.70 using repeated split sampling cross validation, corroborating the predictive ability of our models beyond the training data. Predicting genome size across the Iberian Peninsula, we found the largest genomes in the northern part. Potential limitations of our trait inference approach include dependence on the phylogenetic conservation of traits and limited database coverage of environmental prokaryotes. Overall, our approach enables robust inference of ecologically interpretable traits combined with environmental modelling allowing to harness traits as bioindicators of soil ecosystem functioning.