Performance Of Prediction Models Research Articles

Machine learning-based screening of fuel properties for SI and CI engines using a hybrid group extraction method

The design of high-performance fuels is crucial for achieving clean and efficient combustion of engines. In the current study, a framework for machine learning (ML) model-based fuel design is presented to identify compounds with desired properties for internal combustion engine (ICE) applications. Descriptors computed from newly proposed structural and positional-based group extraction method in this study are used as input in ML models due to the simplicity and computational-efficiency. The ML models were trained and validated using publically available experimental data for up to 1135 compounds. The results demonstrated a high level of predictive accuracy, with R2 values generally exceeding 0.99 for 11 physicochemical properties including melting point, boiling point, enthalpy of vaporization, surface tension, dynamic viscosity, low heating value, liquid density, yield sooting index, cetane number, research octane number and motor octane number. Furthermore, by employing the developed ML models to predict new data, a Fuel Property Database containing 1135 fuels with 12 fuel properties is established, enhanced by an interface for a user-friendly Fuel Physicochemical Properties Prediction Tool that facilitates swift property predictions and database expansion. The Pearson correlation coefficient (PCC) approach is subsequently employed to assess the correlation between descriptors and predicted or experimental properties. The strong agreement between the correlations affirms the effective predictive performance of the ML models within the calibrated data range. Additionally, a PCC analysis is conducted to reveal the individual effects of influential factors and their interactions, highlighting significant features like molecular weight size, degree of branching, and functional group content that influence physical and chemical properties, thereby providing valuable insights for fuel design and optimization. Finally, a initial comprehensive data-driven screening was carried out to identify potential fuel candidates that meet the key property limits for combustion applications in both spark-ignition (SI) and compression-ignition (CI) engines.

A novel deep learning method based on 2-D CNNs and GRUs for permeability prediction of tight sandstone

The accurate calculation of tight sandstone reservoir permeability is crucial for optimizing production and maximizing natural gas recovery. Traditional physical model-based permeability calculation methods heavily rely on core experimental parameters. However, petrophysical experiments are costly, prompting the development of fast and reliable reservoir permeability prediction methods. In this paper, a novel deep learning model that combines two-dimensional convolutional neural network (2-D CNN) model and gated recurrent neural network is proposed to predict tight sandstone reservoir permeability. The 1-D conventional well logs data is converted into the 2-D geological feature map to fit the model's input dimension. The TimeDistributed layer wrapping technology is used to couple the 2-D CNN and GRU to extract comprehensive features of geological feature maps. By training the model, a depth nonlinear mapping between the spatial and temporal features in the well logs data and permeability is established. The model is divided into two subnetworks, 2-D CNNs and GRUs, which are used to compare and analyze the functions of different modules in the proposed model. Additionally, two traditional machine learning algorithms, SVM and XGBoost, are also introduced as comparison models. Four regression evaluation metrics are used to quantify the predictive performance of different models. The test results in a blind well demonstrate that the proposed method outperforms the comparison models in predicting permeability, with a coefficient of determination (R2) of 0.9305, a root mean square error (RMSE) of 0.0895, a mean squared error (MSE) of 0.0080, and a mean absolute error (MAE) of 0.0725. This model can provide insight into the petrophysical characteristics of reaction reservoirs based on conventional well logs data and can accurately estimate reservoir permeability under limited core data, which has great application potential in improving the exploration accuracy of unconventional oil and gas resources.

Ultra-short-term wind power prediction model based on fixed scale dual mode decomposition and deep learning networks

In recent years, decomposition-based combination models have been widely used in wind power prediction. This type of method decomposes the highly volatile wind power into some relatively smooth subsequences, which reduces the difficulty of modeling. However, this might use information from future data in advance, creating the illusion of high prediction accuracy. Therefore, this paper proposes a wind power ultra-short-term prediction model based on fixed scale dual mode decomposition (FSDMD) and deep learning networks. First, the wind power series after fixed scale blocking is decomposed using ensemble empirical mode decomposition (EEMD), and use the improved variational mode decomposition (VMD) based on Spearman rank order correlation coefficient (SROCC) to decompose the obtained high-frequency components twice. Then, the appropriate mode components were selected by calculating the SROCC and experimental analysis, and combined with the convolutional neural network (CNN) and the bidirectional long short-term memory (BiLSTM) network to train the model. Finally, the historical data of wind turbines in a wind farm in Northwest China is used for example verification, and the comparison with other models in the two scenarios of sufficient and insufficient features. The results show that the proposed FSDMD–CNN–BiLSTM model has high prediction accuracy in both scenarios. Especially in the scenario of insufficient features, compared with CNN-BiLSTM model, RMSE, MAE and MAPE are reduced by 8.20,14.24 and 0.15, respectively. In addition, this paper verifies that mode decomposition can improve the performance of prediction model without using future features, which provides ideas for solving similar problems.

Validation of the first-trimester machine learning model for predicting pre-eclampsia in an Asian population.

To evaluate the performance of an artificial intelligence (AI) and machine learning (ML) model for first-trimester screening for pre-eclampsia in a large Asian population. This was a secondary analysis of a multicenter prospective cohort study in 10 935 participants with singleton pregnancies attending for routine pregnancy care at 11-13+6 weeks of gestation in seven regions in Asia between December 2016 and June 2018. We applied the AI+ML model for the first-trimester prediction of preterm pre-eclampsia (<37 weeks), term pre-eclampsia (≥37 weeks), and any pre-eclampsia, which was derived and tested in a cohort of pregnant participants in the UK (Model 1). This model comprises maternal factors with measurements of mean arterial pressure, uterine artery pulsatility index, and serum placental growth factor (PlGF). The model was further retrained with adjustments for analyzers used for biochemical testing (Model 2). Discrimination was assessed by area under the receiver operating characteristic curve (AUC). The Delong test was used to compare the AUC of Model 1, Model 2, and the Fetal Medicine Foundation (FMF) competing risk model. The predictive performance of Model 1 was significantly lower than that of the FMF competing risk model in the prediction of preterm pre-eclampsia (0.82, 95% confidence interval [CI] 0.77-0.87 vs. 0.86, 95% CI 0.811-0.91, P = 0.019), term pre-eclampsia (0.75, 95% CI 0.71-0.80 vs. 0.79, 95% CI 0.75-0.83, P = 0.006), and any pre-eclampsia (0.78, 95% CI 0.74-0.81 vs. 0.82, 95% CI 0.79-0.84, P < 0.001). Following the retraining of the data with adjustments for the PlGF analyzers, the performance of Model 2 for predicting preterm pre-eclampsia, term pre-eclampsia, and any pre-eclampsia was improved with the AUC values increased to 0.84 (95% CI 0.80-0.89), 0.77 (95% CI 0.73-0.81), and 0.80 (95% CI 0.76-0.83), respectively. There were no differences in AUCs between Model 2 and the FMF competing risk model in the prediction of preterm pre-eclampsia (P = 0.135) and term pre-eclampsia (P = 0.084). However, Model 2 was inferior to the FMF competing risk model in predicting any pre-eclampsia (P = 0.024). This study has demonstrated that following adjustment for the biochemical marker analyzers, the predictive performance of the AI+ML prediction model for pre-eclampsia in the first trimester was comparable to that of the FMF competing risk model in an Asian population.

On the impact of soil texture on local scale organic carbon quantification: From airborne to spaceborne sensing domains

Soil organic carbon (SOC) distribution and interaction with light is influenced by soil texture parameters (clay, silt and sand), which makes SOC prediction complicated, especially in samples with considerable pedological variability. Hence, understanding the relationship between SOC and soil texture is important within the context of SOC prediction using remote sensing data. The main objective of this study was to find the impact of soil texture on the performance of local SOC prediction models that were developed on Sentinel-2 (S2) multispectral and CASI/SASI (CS) hyperspectral airborne data as the main predictor variables. One approach to that objective was to lowering the texture variance by stratification of the samples. Therefore, soil samples collected from four agricultural sites in the Czech Republic were segregated based on the i) site-based and ii) texture-based stratification strategies. Random forest (RF) models were then developed on all stratified classes with and without considering the soil texture parameters as predictor variables and results were compared with those obtained by the RF models developed on the non-stratified (NS) samples. Both stratification strategies provided more homogeneous classes, which enhanced the accuracy of SOC prediction, compared to using the NS samples. In addition, the texture-based RF models yielded higher accuracy predictions than the site-based ones. Except sand, adding texture parameters to the main predictors improved accuracy of the models, so that the highest prediction performance was obtained by a texture-based model developed on clay added CS data. Overall, texture-based stratification could significantly enhance the SOC prediction, when the texture parameters were added to the S2 and CS data as the main predictor variables.

Nature-inspired optimization algorithms based feature selection: Application in credit scoring

Nature-inspired algorithms as problem-solving methodologies are extremely effective in discovery of optimized solutions in multi-dimensional and multi-modal problems. Because of qualities like “self-optimization”, “flexibility” and etc., nature-inspired algorithms for problem solving are effectively optimal. Feature selection is an approach to find approximate optimal subset of the features which are more relevant towards the particular outcome. In this study, we focused on how feature selection may improve the credit scoring model’s performance for prediction. Nature-inspired algorithms are applied for feature selection to improve the predictive performance of the credit scoring model. Additionally, four benchmark credit scoring datasets collected from the UCI repository are used to test feature selection by several Nature-inspired algorithms aggregated with “Random Forest (RF)”, “Logistic Regression (LR),” and “Multi-layer Perceptron (MLP)” for classification and results are compared in terms of classification accuracy and G-measures.

Exploring combinations of dimensionality reduction, transfer learning, and regularization methods for predicting binary phenotypes with transcriptomic data

BackgroundNumerous transcriptomic-based models have been developed to predict or understand the fundamental mechanisms driving biological phenotypes. However, few models have successfully transitioned into clinical practice due to challenges associated with generalizability and interpretability. To address these issues, researchers have turned to dimensionality reduction methods and have begun implementing transfer learning approaches.MethodsIn this study, we aimed to determine the optimal combination of dimensionality reduction and regularization methods for predictive modeling. We applied seven dimensionality reduction methods to various datasets, including two supervised methods (linear optimal low-rank projection and low-rank canonical correlation analysis), two unsupervised methods [principal component analysis and consensus independent component analysis (c-ICA)], and three methods [autoencoder (AE), adversarial variational autoencoder, and c-ICA] within a transfer learning framework, trained on > 140,000 transcriptomic profiles. To assess the performance of the different combinations, we used a cross-validation setup encapsulated within a permutation testing framework, analyzing 30 different transcriptomic datasets with binary phenotypes. Furthermore, we included datasets with small sample sizes and phenotypes of varying degrees of predictability, and we employed independent datasets for validation.ResultsOur findings revealed that regularized models without dimensionality reduction achieved the highest predictive performance, challenging the necessity of dimensionality reduction when the primary goal is to achieve optimal predictive performance. However, models using AE and c-ICA with transfer learning for dimensionality reduction showed comparable performance, with enhanced interpretability and robustness of predictors, compared to models using non-dimensionality-reduced data.ConclusionThese findings offer valuable insights into the optimal combination of strategies for enhancing the predictive performance, interpretability, and generalizability of transcriptomic-based models.

Improved training framework in a neural network model for disruption prediction and its application on EXL-50

A neural network model with a classical annotation method has been used on the EXL-50 tokamak to predict impending disruption. However, the results revealed issues of overfitting and overconfidence in predictions caused by inaccurate labeling. To mitigate these issues, an improved training framework has been proposed. In this approach, soft labels from previous training serve as teachers to supervise the further learning process; this has lead to a significant improvement in predictive model performance. Notably, this enhancement is primarily attributed to the coupling effect of the soft labels and correction mechanism. This improved training framework introduces an instance-specific label smoothing method, which reflects a more nuanced model assessment on the likelihood of a disruption. It presents a possible solution to effectively address the challenges associated with accurate labeling across different machines.

Hybrid Machine Learning Approach for Early Diagnosis of Polycystic Ovary Syndrome with Stable Features

Polycystic Ovary Syndrome (PCOS) is a hormonal condition that typically affects female during the time of their reproduction. It is identified by the disruptions in hormonal balance, particularly an increase in levels of androgen (male hormone) in the female body. PCOS can lead to various symptoms and health complications including irregular menstrual cycles, ovarian cysts, fertility issues, insulin resistance, weight gain, acne, and excess hair growth. The real-world PCOS detection is a challenging task whilst PCOS specific cause is unknown and its symptoms are unclear. Thus, accurate and timely diagnosis of PCOS is crucial for effective management and prevention of long-term complications. In such cases, Machine learning based PCOS prediction model support diagnostic process, address potential errors and time constraints. Machine learning algorithms can analyze large set of patient data, including medical history, hormonal profiles, and imaging results, to assist in the diagnosis of PCOS. In particular, the performance of data analysis chore and prediction model is improved by ensemble feature selection strategies. These methods concentrate on selecting a subset of pertinent features from a broader range of features. The unstable nature of the outcome of feature selection algorithm is a frequent issue in practical applications, when it is applied multiple times on similar dataset or with slight modifications in the data. Thus, evaluating the robustness of feature selection algorithm is most important. To address these issues and quantify the robustness, this study uses Jenson-Shannon divergence, an information theoretic approach with ensemble feature selection method to handle the various findings, such as complete ranking, half ranking and top-k lists (without ranking). Furthermore, this article proposes a hybrid machine learning classifier with SMOTE – SVM for the prompt detection of PCOS and the performance of the model is compared with a number of other individual classifiers including KNN (K-Nearest Neighbour), Support Vector Machine (SVM), AdaBoost, LR –Logistic Regression, NB –Nave Bayes, RF –Random Forest, Decision Tree. The proposed SWISS-AdaBoost classifier surpassed other models with 97.81% of accuracy and AUC of 99.08%.

Progression from Prediabetes to Diabetes in a Diverse U.S. Population: A Machine Learning Model.

Objective: To date, there are no widely implemented machine learning (ML) models that predict progression from prediabetes to diabetes. Addressing this knowledge gap would aid in identifying at-risk patients within this heterogeneous population who may benefit from targeted treatment and management in order to preserve glucose metabolism and prevent adverse outcomes. The objective of this study was to utilize readily available laboratory data to train and test the performance of ML-based predictive risk models for progression from prediabetes to diabetes. Methods: The study population was composed of laboratory information services data procured from a large U.S. outpatient laboratory network. The retrospective dataset was composed of 15,029 adults over a 5-year period with initial hemoglobin A1C (A1C) values between 5.0% and 6.4%. ML models were developed using random forest survival methods. The ground truth outcome was progression to A1C values indicative of diabetes (i.e., ≥6.5%) within 5 years. Results: The prediabetes risk classifier model accurately predicted A1C ≥6.5% within 5 years and achieved an area under the receiver-operator characteristic curve of 0.87. The most important predictors of progression from prediabetes to diabetes were initial A1C, initial serum glucose, A1C slope, serum glucose slope, initial HDL, HDL slope, age, and sex. Conclusions: Leveraging readily obtainable laboratory data, our ML risk classifier accurately predicts elevation in A1C associated with progression from prediabetes to diabetes. Although prospective studies are warranted, the results support the clinical utility of the model to improve timely recognition, risk stratification, and optimal management for patients with prediabetes.

Effect of Data Augmentation Using Deep Learning on Predictive Models for Geopolymer Compressive Strength

In recent years, machine learning models have become a potential approach in accurately predicting the concrete compressive strength, which is essential for the real-world application of geopolymer concrete. However, the precursor system of geopolymer concrete is known to be more heterogeneous compared to Ordinary Portland Cement (OPC) concrete, adversely affecting the data generated and the performance of the models. To its advantage, data enrichment through deep learning can effectively enhance the performance of prediction models. Therefore, this study investigates the capability of tabular generative adversarial networks (TGANs) to generate data on mixtures and compressive strength of geopolymer concrete. It assesses the impact of using synthetic data with various models, including tree-based, support vector machines, and neural networks. For this purpose, 930 instances with 11 variables were collected from the open literature. In particular, 10 variables including content of fly ash, slag, sodium silicate, sodium hydroxide, superplasticizer, fine aggregate, coarse aggregate, added water, curing temperature, and specimen age are considered as inputs, while compressive strength is the output of the models. A TGAN was employed to generate an additional 1000 data points based on the original dataset for training new predictive models. These models were evaluated on real data test sets and compared with models trained on the original data. The results indicate that the developed models significantly improve performance, particularly neural networks, followed by tree-based models and support vector machines. Moreover, data characteristics greatly influence model performance, both before and after data augmentation.

Game theory-based mandatory lane change model in intelligent connected vehicles environment

In the environment of intelligent connected vehicles, drivers are capable of making wiser and safer decisions. However, the interaction between drivers and vehicle systems has undergone changes in the intelligent connected vehicles environment, leading to a decrease in the applicability of existing microscopic driving models, such as the mandatory lane change model, which requires reevaluation or improvement. Therefore, to investigate the influence of different intelligent connected vehicles environments on the microscopic mandatory lane-changing model, this study developed three interaction systems to characterize different intelligent connected vehicles environments: the baseline, warning group, and guidance group. The Baseline provides basic information, the warning group adds icons of preceding vehicles and real-time headway information, while the guidance group further includes speed and voice guidance functions. The baseline describes the traditional environment, while the other two groups describe the intelligent connected vehicles environment. Using a self-developed intelligent connected vehicle testing platform, we conducted driving simulation experiments with 43 participants at the interchange merging area of a highway. This study, grounded in game theory, establishes function models for participants, strategies, and payoff functions in the mandatory lane-changing process. Utilizing data from driving simulation experiments, the parameters of the dual-layered planning model are calibrated. Evaluation of the constructed model is conducted through confusion matrices and lane-changing spatiotemporal characteristic indicators. The results demonstrate satisfactory predictive performance of the baseline group model, warning group model, and guidance group model across different intelligent connected vehicles environments. Specifically, compared to existing literature, the baseline group model exhibits improvements of 7 % and 2 % respectively in overall lane-changing detection accuracy by drivers. The warning group model shows improvements of 2.9 % and 1.7 %, while the guidance group model exhibits improvements of 5.1 % and 4.3 %. Additionally, the baseline group model reduces the mean absolute error in predicting different game strategies by 16.7 % and 5.6 % respectively compared to existing literature. Concerning lane-changing position prediction, the warning and guidance group models demonstrate minimal errors, whereas the baseline group model exhibits good consistency in predicting lane-changing duration. Furthermore, both the warning and guidance group models show some delay in predicting lane-changing duration. While intelligent connected vehicles environments significantly influence the prediction of lane-changing positions, they do not significantly affect the prediction of lane-changing duration. However, game strategies significantly impact the prediction of lane-changing duration but do not significantly affect the prediction of lane-changing positions. The study findings offer valuable insights into micro lane-changing behaviors of drivers in intelligent connected vehicles environments, bearing crucial significance for the in-depth investigation of real-time control and guidance strategies for vehicles in merge areas of highways under intelligent connected vehicles conditions.

A Nomogram to Predict the Risk of Acute Ischemic Stroke in Patients with Maintenance Hemodialysis: A Retrospective Cohort Study

Introduction: Acute ischemic stroke (AIS) stands as a leading cause of death and disability globally. This study aimed to investigate the risk factors and relevance linked with AIS in patients undergoing maintenance hemodialysis (MHD) and to create and validate nomogram models. Methods: We examined the medical records of 314 patients with stage 5 chronic kidney disease (CKD 5) undergoing MHD, who sought neurology outpatient department consultation for suspected AIS symptoms between January 2018 and December 2023. These 314 patients were randomly divided into the training cohort (n = 222) and validation cohort (n = 92). The least absolute shrinkage selection operator (LASSO) regression model was employed for optimal feature selection in the AIS risk model. Subsequently, multivariable logistic regression analysis was used to construct a predictive model incorporating the features selected through LASSO. This predictive model’s performance was assessed using the C-index and the area under the receiver operating characteristic curve (AUC). Additionally, calibration and clinical utility were evaluated through calibration plots and decision curve analysis (DCA). The model’s internal validation was conducted using the validation cohort. Results: Predictors integrated into the prediction nomogram encompassed cardiovascular disease (CVD) (odds ratio [OR] 7.95, 95% confidence interval [CI] 2.400–29.979), smoking (OR 5.7, 95% CI: 1.661–21.955), dialysis time (OR: 5.91, 95% CI: 5.866–29.979), low-density lipoprotein (OR: 2.99, 95% CI: 0.751–13.007), and fibrin degradation products (OR: 5.47, 95% CI: 1.563–23.162). The model exhibited robust discrimination, with a C-index of 0.877 and 0.915 in the internal training and validation cohorts, respectively. The AUC for the training set was 0.857, and a similar AUC of 0.905 was achieved in the validation cohort. DCA demonstrated a positive net benefit within a threshold risk range of 2–96%. Conclusion: The proposed nomogram effectively identifies MHD patients at high risk of AIS at an early stage. This model holds the potential to aid clinicians in making preventive recommendations.

A deep learning-based method for the prediction of temporal lobe injury in patients with nasopharyngeal carcinoma

PurposeTo establish a deep learning-based model to predict radiotherapy-induced temporal lobe injury (TLI). Materials and methodsSpatial features of dose distribution within the temporal lobe were extracted using both the three-dimensional convolution (C3D) network and the dosiomics method. The Minimal Redundancy-Maximal-Relevance (mRMR) method was employed to rank the extracted features and select the most relevant ones. Four machine learning (ML) classifiers, including logistic regression (LR), k-nearest neighbors (kNN), support vector machines (SVM) and random forest (RF), were used to establish prediction models. Nested sampling and hyperparameter tuning methods were applied to train and validate the prediction models. For comparison, a prediction model base on the conventional D0.5cc of the temporal lobe obtained from dose volume (DV) histogram was established. The area under the receiver operating characteristic (ROC) curve (AUC) was utilized to compare the predictive performance of the different models. ResultsA total of 127 nasopharyngeal carcinoma (NPC) patients were included in the study. In the model based on C3D deep learning features, the highest AUC value of 0.843 was achieved with 5 features. For the dosiomics features model, the highest AUC value of 0.715 was attained with 1 feature. Both of these models demonstrated superior performance compared to the prediction model based on DV parameters, which yielded an AUC of 0.695. ConclusionThe prediction model utilizing C3D deep learning features outperformed models based on dosiomics features or traditional parameters in predicting the onset of TLI. This approach holds promise for predicting radiation-induced toxicities and guide individualized radiotherapy.

MRI-based clinical-radiomics nomogram model for predicting microvascular invasion in hepatocellular carcinoma.

Preoperative microvascular invasion (MVI) of liver cancer is an effective method to reduce the recurrence rate of liver cancer. Hepatectomy with extended resection and additional adjuvant or targeted therapy can significantly improve the survival rate of MVI+ patients by eradicating micrometastasis. Preoperative prediction of MVI status is of great clinical significance for surgical decision-making and the selection of other adjuvant therapy strategies to improve the prognosis of patients. Established a radiomics machine learning model based on multimodal MRI and clinical data, and analyzed the preoperative prediction value of this model for microvascular invasion (MVI) of hepatocellular carcinoma (HCC). The preoperative liver MRI data and clinical information of 130 HCC patients who were pathologically confirmed to be pathologically confirmed were retrospectively studied. These patients were divided into MVI-positive group (MVI+) and MVI-negative group (MVI-) based on postoperative pathology. After a series of dimensionality reduction analysis, six radiomic features were finally selected. Then, linear support vector machine (linear SVM), support vector machine with rbf kernel function (rbf-SVM), logistic regression (LR), Random forest (RF) and XGBoost (XGB) algorithms were used to establish the MVI prediction model for preoperative HCC patients. Then, rbf-SVM with the best predictive performance was selected to construct the radiomics score (R-score). Finally, we combined R-score and clinical-pathology-image independent predictors to establish a combined nomogram model and corresponding individual models. The predictive performance of individual models and combined nomogram was evaluated and compared by receiver operating characteristic curve (ROC). Alpha-fetoprotein concentration, peritumor enhancement, maximum tumor diameter, smooth tumor margins, tumor growth pattern, presence of intratumor hemorrhage, and RVI were independent predictors of MVI. Compared with individual models, the final combined nomogram model (AUC: 0.968, 95% CI: 0.920-1.000) constructed by radiometry score (R-score) combined with clinicopathological parameters and apparent imaging features showed the optimal predictive performance. This multi-parameter combined nomogram model had a good performance in predicting MVI of HCC, and had certain auxiliary value for the formulation of surgical plan and evaluation of prognosis.

Degradation prediction of proton exchange membrane fuel cell based on mixed gated units under multiple operating conditions

The aging process of proton exchange membrane fuel cell (PEMFC) is affected by different operating factors, and its practical application scenarios involve multiple operation conditions. To improve the accuracy and controllability of aging prediction, a mixed recurrent neural network (RNN) model based on Gated Recurrent Unit (GRU) and Minimal Gated Unit (MGU) is proposed. The model dynamically adjusts the mixed weights of GRU and MGU throughout the entire training process to obtain the optimal prediction network structure throughout the entire lifecycle. Furthermore, an attention mechanism is incorporated into the mixed gated unit (MIXGU) model. The effectiveness of the MIXGU model is evaluated by utilizing experimental data from static condition, quasi-dynamic condition, and dynamic load cycling conditions. The predictive performance of MGU, GRU, MIXGU, and MIXGU model with attention mechanism (AT-MIXGU) are compared under different operating conditions. The validation results indicate that MIXGU model exhibits superior predictive performance compared to single gated unit, the attention mechanism enhances prediction accuracy. And the AT-MIXGU model demonstrates strong generalization capabilities, and well-suited for accurately predicting PEMFC aging under diverse operating conditions.

Supplementation of deep neural networks with simplified physics-based features to increase accuracy of plate fundamental frequency predictions

To improve predictive machine learning-based models limited by sparse data, supplemental physics-related features are introduced into a deep neural network (DNN). While some approaches inject physics through differential equations or numerical simulation, improvements are possible using simplified relationships from engineering references. To evaluate this hypothesis, thin rectangular plates were simulated to generate training datasets. With plate dimensions and material properties as input features and fundamental natural frequency as the output, predictive performance of a data-driven DNN-based model is compared with models using supplemental inputs, such as modulus of rigidity. To evaluate model accuracy improvements, these additional features are injected into various DNN layers, and the network is trained with four different dataset sizes. When evaluated against independent data of similar features to the training sets, supplementation provides no statistically-significant prediction error reduction. However, notable accuracy gains occur when independent test data is of material and dimensions different from the original training set. Furthermore, when physics-enhanced data is injected into multiple DNN layers, reductions in mean error from 33.2% to 19.6%, 34.9% to 19.9%, 35.8% to 22.4%, and 43.0% to 28.4% are achieved for dataset sizes of 261, 117, 60, and 30, respectively, demonstrating potential for generalizability using a data supplementation approach. Additionally, when compared with other methods—such as linear regression and support vector machine (SVM) approaches—the physics-enhanced DNN demonstrates an order of magnitude reduction in percentage error for dataset sizes of 261, 117, and 60 and a 30% reduction for a size of 30 when compared with a cubic SVM model independently tested with data divergent from the training and validation set.

Prediction model for pheochromocytoma/paraganglioma using nCounter assay.

World Health Organization defined pheochromocytomas/paragangliomas (PPGL) as malignant tumors in 2017 because the existing classification system could not reflect locally aggressive behavior sufficiently. However, predicting the likelihood of metastasis remains a crucial part of the treatment strategy. From one tertiary care hospital and onesecondary hospital, 97 PPGL cases were selected. Medical records of PPGL cases with the presence of formalin-fixed and paraffin-embedded (FFPE) tissue of primary lesion were reviewed. For FFPE tissues, a nCounter assay was conducted to determine differently expressed genes between metastatic and non-metastatic PPGL groups. Performances of prediction models for the likelihood of metastasis were calculated. Of a total of 97 PPGL cases, 39, 20, and 38 were classified as benign, malignant, and validation, respectively. In the nCounter assay, CDK1, TYMS, and TOP2A genes showed significant differences in expression. Tumor size was positively correlated with CDK1 expression level. The Lasso regression model showed supreme performance of sensitivity 91.7% and specificity 95.5% when those significant factors were considered. Machine learning of multi-modal classifiers can be used to create a prediction model for metastasis of PPGL with high sensitivity and specificity using nCounter assay. Moreover, CDK1 inhibitors could be considered for developing drug treatment.

A prediction model for rumor user propagation behavior based on sparse representation and transfer learning

This paper introduces a prediction model rooted in sparse representation and transfer learning, with the primary objective of predicting user behavior during rumor propagation. Users' behavior is dynamic, and rumor, rumor-refuting, and rumor-promoting messages interact dynamically, according to the model. Firstly, this paper proposes to compensate for the low performance of propagation prediction models due to data sparsity at the beginning of rumor topics' lifes. Transfer learning is used to compensate for the data sparsity problem at the beginning of the topic. To map the rumor topic space effectively, a low-rank dense vectorization algorithm based on sparse representation is proposed. Finally, to mine the potential impact of multiple types of information on users, a model based on three-party game theory is constructed. It considers the complex interaction between rumor, rumor-refuting, and rumor-promoting information in the propagation of rumor. Additionally, this paper develops a model of dynamic user behavior prediction using Rumor-Attention-Mechanism-Graph-Attention (RAM-GAT) to predict rumor propagation. Experiments demonstrate that our model can effectively mine the interaction influence between multiple types of information. It can accurately predict user behavior when initial data is insufficient. In addition, rumor propagation patterns and trends are revealed.

Identifying multilevel predictors of behavioral outcomes like park use: A comparison of conditional and marginal modeling approaches.

This study compared marginal and conditional modeling approaches for identifying individual, park and neighborhood park use predictors. Data were derived from the ParkIndex study, which occurred in 128 block groups in Brooklyn (New York), Seattle (Washington), Raleigh (North Carolina), and Greenville (South Carolina). Survey respondents (n = 320) indicated parks within one half-mile of their block group used within the past month. Parks (n = 263) were audited using the Community Park Audit Tool. Measures were collected at the individual (park visitation, physical activity, sociodemographic characteristics), park (distance, quality, size), and block group (park count, population density, age structure, racial composition, walkability) levels. Generalized linear mixed models and generalized estimating equations were used. Ten-fold cross validation compared predictive performance of models. Conditional and marginal models identified common park use predictors: participant race, participant education, distance to parks, park quality, and population >65yrs. Additionally, the conditional mode identified park size as a park use predictor. The conditional model exhibited superior predictive value compared to the marginal model, and they exhibited similar generalizability. Future research should consider conditional and marginal approaches for analyzing health behavior data and employ cross-validation techniques to identify instances where marginal models display superior or comparable performance.