High Predictive Capability Research Articles

A Churn Prediction System for Telecommunication Company Using Random Forest and Convolution Neural Network Algorithms

INTRODUCTION: Customer churn is a severe problem of migrating from one service provider to another. Due to the direct influence on the company's sales, companies are attempting to promote strategies to identify the churn of prospective consumers. Hence it is necessary to examine issues that influence customer churn to yield effective solutions to minimize churn. OBJECTIVES: The major purpose of this work is to create a model of churn prediction that assists telecom operatives to envisage clients that are more probably to be prone to churn. METHODS: The experimental strategy for this study leverages the machine learning techniques on the telecom churn dataset, employing an improved Relief-F feature selection algorithm to extract related features from the enormous dataset. RESULTS: The result demonstrates that CNN has a high prediction capability of 94 percent compared to the 91 percent Random Forest classifier. CONCLUSION: The results are of enormous relevance to the telecommunication business in improving churners and loyal clients.

Estimation of fatigue crack initiation and growth in engineering product development using a digital twin approach

A digital twin is a computational model in cyberspace that is used to support engineering decisions. Maintaining high predictive capability of a digital twin model is of great concern to the engineers who make design decisions at the early stages of product development. In the work described in this paper, the predictive capability of the digital twin approach is improved by considering uncertainties in manufacturing and test conditions. The proposed digital twin approach can be used in a variety of product development settings. The proposed idea takes advantage of hybrid digital twin approaches, using both data-driven and physics-based approaches. The proposed approach is based on two techniques; (i) statistical model calibration and (ii) probabilistic element updating. In statistical model calibration, statistical parameters of input variables are estimated. Further, probabilistic analysis using estimated statistical parameters can predict possible critical elements. In probabilistic element updating procedures, the possible crack initiation and growth element is updated. The validity of the proposed method is demonstrated using a case study of an automotive sub-frame fatigue test. From the results, we conclude that the proposed digital twin approach can accurately estimate crack initiation and growth of an automotive structure under uncertain loading conditions and material properties.

Geospatial Artificial Intelligence (GeoAI) in the Integrated Hydrological and Fluvial Systems Modeling: Review of Current Applications and Trends

This paper reviews the current GeoAI and machine learning applications in hydrological and hydraulic modeling, hydrological optimization problems, water quality modeling, and fluvial geomorphic and morphodynamic mapping. GeoAI effectively harnesses the vast amount of spatial and non-spatial data collected with the new automatic technologies. The fast development of GeoAI provides multiple methods and techniques, although it also makes comparisons between different methods challenging. Overall, selecting a particular GeoAI method depends on the application’s objective, data availability, and user expertise. GeoAI has shown advantages in non-linear modeling, computational efficiency, integration of multiple data sources, high accurate prediction capability, and the unraveling of new hydrological patterns and processes. A major drawback in most GeoAI models is the adequate model setting and low physical interpretability, explainability, and model generalization. The most recent research on hydrological GeoAI has focused on integrating the physical-based models’ principles with the GeoAI methods and on the progress towards autonomous prediction and forecasting systems.

Alternative method for determining basis weight in papermaking by using an interactive soft sensor based on an artificial neural network model

Abstract Currently, there are two procedures to determine the basis weight in papermaking processes: the measurements made by the quality control laboratory or the measurements made by the quality control system. This research presents an alternative to estimating basis weight-based artificial neural network (ANN) modeling. The NN architecture was constructed by trial and error, obtaining the best results using two hidden layers with 48 and 12 neurons, respectively, in addition to the input and output layers. Mean absolute error and mean absolute percentage error was used for the loss and metric functions, respectively. Python was used in the training, validation, and testing process. The results indicate that the model can reasonably determine the basis weight given the independent variables analyzed here. The R 2 {R^{2}} reached by the model was 94 %, and MAE was 12.40 grams/m2. Using the same dataset, the fine tree regression model showed an R 2 {R^{2}} of 99 % and an MAE of 3.35 grams/m2. Additionally, a dataset not included in the building process was used to validate the method’s performance. The results showed that ANN-based modeling has a higher predictive capability than the regression tree model. Therefore, this model was embedded in a graphic user interface that was developed in Python.

Data-Driven Prediction of Quartz Dissolution Rates at Near-Neutral and Alkaline Environments

Dissolution of silicate-based materials is important to many natural processes and engineering applications, including cement and concrete production. Here, we present a data-driven study to predict the dissolution rates of crystalline silica (i.e., quartz) in near-neutral and alkaline environments. We present a quartz dissolution database containing both dissolution rates and five major dissolution conditions (i.e., temperature, pressure, pH at the experimental temperature T (pHT), and the sodium and alumina content in the solution) via data mining from the literature. We supplement the database with experimental data of quartz dissolution rate in sodium hydroxide solutions (0–5 M) at different target temperatures (25–90°C), which are significantly less covered by the existing literature. We build two data-driven models (i.e., random forest (RF) and artificial neural network (ANN)) to predict the dissolution rate of quartz (i.e., output target) as a function of dissolution conditions (i.e., input features). The results show that both RF and ANN models exhibit high predictive capability, with R2 values of 0.97–0.98, MAPEs of 2.95–4.24% and RMSEs of ∼0.31–0.44 log (mole/m2/s) for the test set. These prediction errors are much smaller than linear regression models (RMSE of ∼1.25 log) also presented here and comparable with those achieved in previous studies using reaction models based on a smaller and less complex dataset (RMSE of ∼0.35–0.44 log). We further evaluate the interpretability and performance of the data-driven models, and the results show that the model predictions are generally consistent with literature observations, including the different impacts of input features on dissolution rate. In particular, the ANN model appears to exhibit a certain level of ability to extrapolate, i.e., making predictions in feature space not covered in the database.

Interaction of rigid shallow foundation with dip-slip normal fault rupture outcrop: effective parameters and retrofitting strategies

Devastating influences of fault rupture outcrop on various geo-structures during past earthquakes have revealed the necessity of more detailed and comprehensive studies on the fault-foundation interaction. In this study, the interaction between the normal fault and shallow surface/embedded strip foundations is rigorously examined through a well-established upper bound limit analysis formulation in conjunction with the finite element discretization and linear programming technique. The adopted upper bound FELA method is shown to be efficient and cost-effective with high predictive capability, which can be easily implemented in any common computational platforms, like MATLAB. The accuracy of the adopted numerical method is first verified against some findings in the literature. Using this validated numerical approach, a comprehensive parametric study is performed to examine the interaction of the normal fault rupture outcrop with the surface and embedded shallow foundations. Accordingly, the influences of various parameters including the foundation breadth, the service footing pressure, the internal friction angle of the soil deposit, the foundation embedment depth, the fault throw and the relative distance between the fault rupture and shallow foundation, on the normal fault-footing interaction are systematically investigated and discussed. In addition, as an efficient retrofitting strategy to protect the shallow foundation against the fault outcrop, rigid mitigating barrier walls with different filling materials are proposed so as to divert the mainstream of the fault rupture and thus reduce its destructive effects on the overlying superstructures. In this regard, another parametric survey is carried out to explore the effects of wall-related geometric parameters, such as its depth, width, filling material and relative position, on the reduction of the destructive consequences of the normal fault rupture emergence at the ground surface. The results generally show that applying a rigid concrete wall with suitable height, width and distance from the foundation leads to the significant reduction in the amount of surface footing rotation subjected to normal fault rupture outcrop.

Abstract 5865: Analysis of RNA sequencing data to advance our understanding of colorectal cancer health disparity in Native Hawaiians

Abstract Background Colorectal cancer (CRC) is the third deadliest cancer among Americans. Native Hawaiians (NH) as a unique ethnic group in the US, exhibit many disproportionate health issues. In particular, they suffer from both elevated CRC incidence and mortality rates. This study aimed to investigate if genetic differences could contribute to disparities observed between NH and White CRC patients by defining the association between unique NH-genetic differences and CRC outcomes. Methods Paired tumor and adjacent normal biospecimens of NH CRC patients were collected from the Hawaii Tumor Registry (HTR), and gene expression and mutation were examined by RNA sequencing. Genomic data of CRC Whites was extracted from The Cancer Genome Atlas (TCGA) database as a comparative cohort. Differential expressed genes (DEGs) were identified for both NH and TCGA cohorts via DESeq2 with FDR q-value < 0.05 and a cutoff of 2-fold change. MutTecT2 and GATK were used to identify significant somatic mutations. Least Absolute Shrinkage and Selection Operator (LASSO) logistic regression with 10-fold cross validation was performed to predict genetic risk factors for early detection, and Ingenuity Pathway Analysis (IPA) was processed for canonical pathways and network discovery. Univariate Cox proportional regression was performed to identify DEGs related to patient survival; multivariable Cox regression model with stepwise fitting generated a prognostic index (PI), and the prediction value was examined by the Area Under the Receiver Operating Characteristic Curve (AUC). A nomogram was developed by integrating PI and clinicopathologic factors; calibration curves were provided to internally validate the performance, and discriminative ability was appraised by concordance index. Results In total, 2096 DEGs were identified between NH tumor and normal groups, 1740 transcripts and 78 carrying mutations were unique to NH compared to the TCGA Whites cohort. A set of 23 genes including 10 NH specific DEGs were identified as genetic risk factors for detecting NH with CRC, and the AUC was 99.8%. A 9 gene-signature prognostic model including 5 NH specific DEGs was built with high survival prediction capability (AUC=0.99), and Kaplan-Meier curve showed that the low PI group had a better survival than the high PI group in NH with CRC (Logrank P=3.6E-05). After adjustment by age, gender, and tumor grade, the prognostic 9 gene-signature was still significant (P=0.0059). By integrating the above signatures with prognostic clinicopathologic features, a nomogram was constructed to stratified patients with overall survival rates for 3, 10, and 20 years. Conclusion This study demonstrated that CRC tumors from NH patients differ from White patients in their gene expression patterns and mutations, particularly those associated with clinical outcomes, which reinforced the necessity of NH race-specific biomedical research. Citation Format: Yuanyuan Fu, Devin Takahashi, Vedbar Khadka, Masaki Nasu, Mayumi Jijiwa, Heather Borgard, Peiwen Fei, Youping Deng. Analysis of RNA sequencing data to advance our understanding of colorectal cancer health disparity in Native Hawaiians [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5865.

Predicting electrophoretic mobility of TiO2, ZnO, and CeO2 nanoparticles in natural waters: The importance of environment descriptors in nanoinformatics models

Natural and engineered nanoparticles (NPs) entering the environment are influenced by many physicochemical processes and show various behavior in different systems (e.g., natural waters showing different characteristics). Determining the physicochemical characteristics and predicting the behavior of nanoparticles ending up in the natural aquatic environment are key aspects of their risk assessment. Here, we show that the quantitative structure-property relationship modeling method used in nanoinformatics (nano-QSPR) can be successfully applied to predict environmental fate-relevant properties (electrophoretic mobility) of TiO2, ZnO, and CeO2 nanoparticles. However, in contrast to the previous works, we postulate to use, in parallel: (i) the nanoparticles' structure descriptors (S-descriptors) and (ii) the environment descriptors (E-descriptors) as the input variables. Thus, the method should be abbreviated more precisely as nano-QSEPR (“E” stands for the “environment”). As a proof-of-the-concept, we have developed a group of models (including MLR, GA-PLS, PCR, and Meta-Consensus models) with high predictive capabilities (QEXT2 = 0.931 for the GA-PLS model), where the S-descriptors are represented by the core-shell model descriptor and the E-descriptors – by different ambient water features (including ions concentration and the ionic strength). The newly proposed nano-QSEPR modeling scheme can be efficiently used to design safe and sustainable nanomaterials.

Mechanistic Analysis of Chemically Diverse Bromodomain-4 Inhibitors Using Balanced QSAR Analysis and Supported by X-ray Resolved Crystal Structures.

Bromodomain-4 (BRD-4) is a key enzyme in post-translational modifications, transcriptional activation, and many other cellular processes. Its inhibitors find their therapeutic usage in cancer, acute heart failure, and inflammation to name a few. In the present study, a dataset of 980 molecules with a significant diversity of structural scaffolds and composition was selected to develop a balanced QSAR model possessing high predictive capability and mechanistic interpretation. The model was built as per the OECD (Organisation for Economic Co-operation and Development) guidelines and fulfills the endorsed threshold values for different validation parameters (R2tr = 0.76, Q2LMO = 0.76, and R2ex = 0.76). The present QSAR analysis identified that anti-BRD-4 activity is associated with structural characters such as the presence of saturated carbocyclic rings, the occurrence of carbon atoms near the center of mass of a molecule, and a specific combination of planer or aromatic nitrogen with ring carbon, donor, and acceptor atoms. The outcomes of the present analysis are also supported by X-ray-resolved crystal structures of compounds with BRD-4. Thus, the QSAR model effectively captured salient as well as unreported hidden pharmacophoric features. Therefore, the present study successfully identified valuable novel pharmacophoric features, which could be beneficial for the future optimization of lead/hit compounds for anti-BRD-4 activity.

Spectrum Anomaly Detection Based on Spatio-Temporal Network Prediction

With the miniaturization of communication devices, the number of distributed electromagnetic devices is increasing. In order to achieve effective management of the electromagnetic spectrum, prediction and anomaly detection of the spectrum has become increasingly critical. This paper proposes an algorithmic framework for detecting spectrum anomalies using deep learning techniques. More specifically, the framework includes spectrum prediction and anomaly detection. We use the sliding window method to divide the time series, construct multi-timescale historical data, and train the model with normal data to have high accuracy spectrum prediction capability. We analyze and determine the discriminant function to distinguish the spectral anomalies by calculating the differences between the predicted and real data. The experimental results show that the proposed method outperforms existing baseline algorithms based on real-world spectrum measurement data and simulated anomaly data.

Implementing machine learning to optimize the cost-benefit of urban water clarifier geometrics

Clarification basins are ubiquitous water treatment units applied across urban water systems. Diverse applications include stormwater systems, stabilization lagoons, equalization, storage and green infrastructure. Residence time (RT), surface overflow rate (SOR) and the Storm Water Management Model (SWMM) are readily implemented but are not formulated to optimize basin geometrics because transport dynamics remain unresolved. As a result, basin design yields high costs from hundreds of thousands to tens of million USD. Basin optimization and retrofits can benefit from more robust and efficient tools. More advanced methods such as computational fluid dynamics (CFD), while demonstrating benefits for resolving transport, can be complex and computationally expensive for routine applications. To provide stakeholders with an efficient and robust tool, this study develops a novel optimization framework for basin geometrics with machine learning (ML). This framework (1) leverages high-performance computing (HPC) and the predictive capability of CFD to provide artificial neural network (ANN) development and (2) integrates a trained ANN model with a hybrid evolutionary-gradient-based optimization algorithm through the ANN automatic differentiation (AD) functionality. ANN model results for particulate matter (PM) clarification demonstrate high predictive capability with a coefficient of determination (R2) of 0.998 on the test dataset. The ANN model for total PM clarification of three (3) heterodisperse particle size distributions (PSDs) also illustrates good performance (R2>0.986). The proposed framework was implemented for a basin and watershed loading conditions in Florida (USA), the ML basin designs yield substantially improved cost-effectiveness compared to common designs (square and circular basins) and RT-based design for all PSDs tested. To meet a presumptive regulatory criteria of 80% PM separation (widely adopted in the USA), the ML framework yields 4.7X to 8X lower cost than the common basin designs tested. Compared to the RT-based design, the ML design yields 5.6X to 83.5X cost reduction as a function of the finer, medium, and coarser PSDs. Furthermore, the proposed framework benefits from ANN’s high computational efficiency. Optimization of basin geometrics is performed in minutes on a laptop using the framework. The framework is a promising adjuvant tool for cost-effective and sustainable basin implementation across urban water systems.

Compressive Strength Estimation of Fly Ash/Slag Based Green Concrete by Deploying Artificial Intelligence Models.

Cement production is one of the major sources of decomposition of carbonates leading to the emission of carbon dioxide. Researchers have proven that incorporating industrial wastes is of paramount significance for producing green concrete due to the benefits of reducing cement production. The compressive strength of concrete is an imperative parameter to consider when designing concrete structures. Considering high prediction capabilities, artificial intelligence models are widely used to estimate the compressive strength of concrete mixtures. A variety of artificial intelligence models have been developed in the literature; however, evaluation of the modeling procedure and accuracy of the existing models suggests developing such models that manifest the detailed evaluation of setting parameters on the performance of models and enhance the accuracy compared to the existing models. In this study, the computational capabilities of the adaptive neurofuzzy inference system (ANFIS), gene expression programming (GEP), and gradient boosting tree (GBT) were employed to investigate the optimum ratio of ground-granulated blast furnace slag (GGBFS) and fly ash (FA) to the binder content. The training process of GEP modeling revealed 200 chromosomes, 5 genes, and 12 head sizes as the best hyperparameters. Similarly, ANFIS hybrid subclustering modeling with aspect ratios of 0.5, 0.1, 7, and 150; learning rate; maximal depth; and number of trees yielded the best performance in the GBT model. The accuracy of the developed models suggests that the GBT model is superior to the GEP, ANFIS, and other models that exist in the literature. The trained models were validated using 40% of the experimental data along with parametric and sensitivity analysis as second level validation. The GBT model yielded correlation coefficient (R), mean absolute error (MAE), and root mean square error (RMSE), equaling 0.95, 3.07 MPa, and 4.80 MPa for training, whereas, for validation, these values were recorded as 0.95, 3.16 MPa, and 4.85 MPa, respectively. The sensitivity analysis revealed that the aging of the concrete was the most influential parameter, followed by the addition of GGBFS. The effect of the contributing parameters was observed, as corroborated in the literature.

Machine Learning Short-Term Energy Consumption Forecasting for Microgrids in a Manufacturing Plant

Energy production and supply are important challenges for civilisation. Renewable energy sources present an increased share of the energy supply. Under these circumstances, small-scale grids operating in small areas as fully functioning energy systems are becoming an interesting solution. One crucial element to the success of micro-grid structures is the accurate forecasting of energy consumption by large customers, such as factories. This study aimed to develop a universal forecasting tool for energy consumption by end-use consumers. The tool estimates energy use based on real energy-consumption data obtained from a factory or a production machine. This model allows the end-users to be equipped with an energy demand prediction, enabling them to participate more effectively in the smart grid energy market. A single, long short-term memory (LSTM)-layer-based artificial neural network model for short-term energy demand prediction was developed. The model was based on a manufacturing plant’s energy consumption data. The model is characterised by high prediction capability, and it predicted energy consumption, with a mean absolute error value of 0.0464. The developed model was compared with two other methodologies.

Fine-grained highway autonomous vehicle lane-changing trajectory prediction based on a heuristic attention-aided encoder-decoder model

Accurate prediction of lane-changing (LC) trajectories can improve traffic efficiency and reduce the probability of accidents, as well as provide driving strategies for intelligent connected vehicles (ICVs) and connected autonomous vehicles (CAVs). Aiming at solving the problems of low prediction accuracy, difficulty in long-term prediction, and an inability of a fine-grained level description of conventional models, a prediction model based on an attention-aided encoder-decoder structure and deep neural network (DNN) is proposed. This study analyzes the LC process and proposes an LC segmentation and sampling method for dividing LC into four stages. The optimal attention-aided encoder-decoder model is tested by trajectory data in the four LC stages and then a heuristic network model is designed. In addition, the proposed heuristic network is connected with the DNN for predicting vehicle kinematics data while predicting the vehicle trajectory. Finally, the heuristic network and DNN are tested in cascade to form a joint cascade prediction model that can perform a fine-grained LC description based on the prediction results. The experimental results show that the proposed cascade prediction model has high accuracy and long-term prediction capability of a vehicle’s trajectory, velocity, acceleration, and steering angle and is also capable of fine-grained LC description. The proposed prediction model can provide a useful theoretical basis for further research on ICVs and CAVs.

Identifying Timber Species by Diffuse Reflection Spectra in the Near-Infrared Region Using a Linear Discriminant Analysis

The article presents an algorithm of analysis of coniferous timber materials ( Pinus sylvestris L., Pinus sibirica Du Tour, Larix decidua Mill., Abies alba L. and Picea abies (L.) Karst.) using the spectroscopy method in the near-infrared region. Timber is a complex organic material consisting of cellulose, hemicellulose, lignin, and extractive substances. These compounds generate absorption bands in the near-infrared region which are mainly overtones and combination bands of the O-H, N-H and C-H functional groups. The authors suggest the best ways to correct scattering considering the specific features of the samples under examination. Based on the application of linear discriminant analysis, a method is proposed for the automatic identification of the timber species by diffuse reflection spectra in the near-infrared region, as well as key parameters sufficient for the classification procedure are determined. The obtained LDA-models showed high predictive capability. The overall average classification accuracy reached 97,6%. Based on the results obtained, it can be stated that the NIR spectroscopy is suitable for solving tasks related to the identification of the timber species; it has proven to be an effective tool for distinguishing wood species.

Prediction of the Total Base Number (TBN) of Engine Oil by Means of FTIR Spectroscopy

The objective of this study is to develop a statistical model to accurately estimate the total base number (TBN) value of diesel engine oils on the basis of the Fourier transform infrared spectroscopy (FTIR) analysis. The research sample consisted of oils used in the course of 14,820 km. The samples were collected after each 1000 km and both FTIR and TBN measurements were performed. By applying the measured absorbance values, five statistical models aimed at predicting TBN values were elaborated with the use of the following information: aggregated values of measured absorbance in defined spectral ranges, extremes at wavenumbers, or the surface area of spectral bands related to the vibrations of specific molecular structures. The obtained models may be considered a continuation and an extension of previous studies of this type described in the literature on the subject. The results of the study and the analysis of the obtained data have led to the development of two models with high predictive capabilities (R2 > 0.98, RMSE < 0.5). Another model, which had the smallest number of variables in comparison to other models, had markedly lower R2 value (0.9496) and the highest RMSE (0.5596). Yet another model, where the dimensionality of the pre-processed full spectra was reduced to four aggregates through averaging, turned out to be slightly worse than the best one (R2 = 0.9728). The study contributes to a more in-depth understanding of the FTIR-based TBN prediction tools that may be readily available to all interested parties.

Safety performance functions to predict separation minima infringements in en-route airspace

PurposeThe purpose of this study is to establish a systematic framework to characterise the safety of air routes, in terms of separation minima infringements (SMIs) between en-route aircraft, based on the definition of models known as safety performance functions.Design/methodology/approachTechniques with high predictive capability were selected that enable both expert knowledge and data to be harnessed: Bayesian networks. It was necessary to establish a conceptual framework that integrates the knowledge currently available on the causality and precursors of SMIs with the hindsight derived from the analysis of the type of data available. To translate the conceptual framework into a set of causal subnets, the concepts of air traffic management (ATM) barrier model and event trees have been incorporated.FindingsThe model combines analytics and insights, as well as predictive capability, to answer the question of how airspace separation infringements are produced and what their frequency of occurrence will be. The main outputs of the network are the predicted probability of success for the ATM barriers and the predicted probability distribution of the vertical and horizontal separation of an aircraft in its closest point of approach.Originality/valueThe main contribution of this work is that, by virtue of the calculation capacity obtained, the network can be used to draw conclusions about the impact that a modification of the airspace and of the traffic, or operational conditions, would have on the effectiveness of the barriers and on the final distributions of distance between aircraft in the CPA, thereby estimating the probability of SMI.

Predicting Australia’s Domestic Airline Passenger Demand using an Anfis Approach

Abstract The forecasting of future airline passenger demand is critical task for airline management. The objective of the present study was to develop an adaptive neuro-fuzzy inference system (ANFIS) for predicting Australia’s domestic airline passenger demand. The ANFIS model was trained, tested, and validated in the study. Sugeno fuzzy rules were used in the ANFIS structure and Gaussian membership function, and linear membership functions were also developed. The hybrid learning algorithm and the subtractive clustering partition method were used to generate the optimum ANFIS models. The results found that the mean absolute percentage error (MAPE) for the overall data set of the ANFIS model was 3.25% demonstrating that the ANFIS model has high predictive capabilities. The ANFIS model could be used in other domestic air travel markets.

Bacteremia detection from complete blood count and differential leukocyte count with machine learning: complementary and competitive with C-reactive protein and procalcitonin tests

BackgroundBiomarkers, such as leukocyte count, C-reactive protein (CRP), and procalcitonin (PCT), have been commonly used to predict the occurrence of life-threatening bacteremia and provide prognostic information, given the need for prompt intervention. However, such diagnosis methods require much time and money. Therefore, we propose a method with a high prediction capability using machine learning (ML) models based on complete blood count (CBC) and differential leukocyte count (DC) and compare its performance with traditional CRP or PCT biomarker methods and those of models incorporating CRP or PCT biomarkers.MethodsWe collected 366,586 daily blood culture (BC) results, of which 350,775 (93.2%), 308,803 (82.1%), and 23,912 (6.4%) cases were issued CBC/DC (CBC/DC group), CRP with CBC/DC (CRP&CBC/DC group), and PCT with CBC/DC (PCT&CBC/DC group), respectively. For the ML methods, conventional logistic regression and random forest models were selected, trained, applied, and validated for each group. Fivefold validation and prediction capability were also evaluated and reported.ResultsOverall, the ML methods, such as the random forest model, demonstrated promising performances. When trained with CBC/DC data, it achieved an area under the ROC curve (AUC) of 0.802, which is superior to the prediction conventionally made with CRP/PCT levels (0.699/0.731). Upon evaluating the performance enhanced by incorporating CRP or PCT biomarkers, it reported no substantial AUC increase with the addition of either CRP or PCT to CBC/DC data, which suggests the predicting power and applicability of using only CBC/DC data. Moreover, it showed competitive prognostic capability compared to the PCT test with similar all-cause in-hospital mortality (45.10% vs. 47.40%) and overall median survival time (27 vs. 25 days).ConclusionsThe ML models using only CBC/DC data yielded more accurate bacteremia predictions compared to those by methods using CRP and PCT data and reached similar prognostic performance as by PCT data. Thus, such models are potentially complementary and competitive with traditional CRP and PCT biomarkers for conducting and guiding antibiotic usage.

Incorporating Approximate Dynamics Into Data-Driven Calibrator: A Representative Model for Ship Maneuvering Prediction

High-fidelity models capable of accurately predicting ship motion are critical for promoting innovation and efficiency in the maritime industry. However, creating an advanced model that comprehensively represents the system and its interaction with dynamic environments has always been challenging. Many models provide partial knowledge about a system. To handle the deficiency and improve model fidelity, in this artile, we propose a hybrid modeling methodology, in which prior knowledge describing the ship dynamic effects is incorporated into a data-driven calibrator, yielding a representative model with high predictive capability. Enabled by the integration of model estimated ship states into the calibrator, the informative information could be interpreted and carried forward. Simulation and full-scale experiments are conducted on the research vessel Gunnerus to exemplify the concept. A best available numerical model and a neural network are prepared to be the foundation and calibrator, respectively. Experiment results show that the cooperative model greatly improves the predictive capability of the research vessel. From the ship modeling perspective, this study provides new insights by bridging the gap between two separate domains: 1) model-based and 2) data-driven.