Types Of Data Inputs Research Articles

Multi-View Integrative Approach For Imputing Short-Chain Fatty Acids and Identifying Key factors predicting Blood SCFA.

Short-chain fatty acids (SCFAs) are the main metabolites produced by bacterial fermentation of dietary fiber within gastrointestinal tract. SCFAs produced by gut microbiotas (GMs) are absorbed by host, reach bloodstream, and are distributed to different organs, thus influencing host physiology. However, due to the limited budget or the poor sensitivity of instruments, most studies on GMs have incomplete blood SCFA data, limiting our understanding of the metabolic processes within the host. To address this gap, we developed an innovative multi-task multi-view integrative approach (M2AE, Multi-task Multi-View Attentive Encoders), to impute blood SCFA levels using gut metagenomic sequencing (MGS) data, while taking into account the intricate interplay among the gut microbiome, dietary features, and host characteristics, as well as the nuanced nature of SCFA dynamics within the body. Here, each view represents a distinct type of data input (i.e., gut microbiome compositions, dietary features, or host characteristics). Our method jointly explores both view-specific representations and cross-view correlations for effective predictions of SCFAs. We applied M2AE to two in-house datasets, which both include MGS and blood SCFAs profiles, host characteristics, and dietary features from 964 subjects and 171 subjects, respectively. Results from both of two datasets demonstrated that M2AE outperforms traditional regression-based and neural-network based approaches in imputing blood SCFAs. Furthermore, a series of gut bacterial species (e.g., Bacteroides thetaiotaomicron and Clostridium asparagiforme), host characteristics (e.g., race, gender), as well as dietary features (e.g., intake of fruits, pickles) were shown to contribute greatly to imputation of blood SCFAs. These findings demonstrated that GMs, dietary features and host characteristics might contribute to the complex biological processes involved in blood SCFA productions. These might pave the way for a deeper and more nuanced comprehension of how these factors impact human health.

Comparative analysis of hydrodynamic flowrate sources as drivers of water quality models for nitrogenous compounds in complex ungauged South African rivers

ABSTRACT Water quality modelling is a critical tool for managing the health of river ecosystems, particularly in regions impacted by point source pollution activities. This study investigates the influence of different hydrodynamic data sources on the performance of two river water quality models, the Basic Model (BM) and the Water Quality Analysis Simulation Programme (WASP) for modelling nitrogenous compounds in a complex river system including wastewater treatment plant effluent discharges. Four diverse hydrodynamic data input types were considered. These included measured station data, altered station data, rainfall-generated flow, and the WRSM/Pitman model estimate. Findings revealed trends, analysis of variance (ANOVA), and t-test analyses consistently demonstrated significant disparities between model predictions and measured data in specific river segments, indicating a need for segment-specific modelling approaches. An increase in Root Mean Square Error (RMSE) and Mean Square Error (MSE) values in certain segments pointed to a decline in model accuracy when confronted with distinct hydrodynamic conditions. Additionally, application of four diverse hydrodynamic data input sources yielded similar performance for BM and WASP against measured data. The research findings indicated a complex interplay between river hydrodynamics and water quality modelling, resulting in a recommendation for tailored modelling strategies that account for unique characteristics of river segments.

CrystalMind: A surrogate model for predicting 3D models with recrystallization in open-die hot forging including an optimization framework

This work introduces CrystalMind, a surrogate model developed to accurately reconstruct 3D models along with recrystallization induced by the open-die hot forging process. CrystalMind employs two types of data inputs, a 3D Model depicting pre-stroke workpiece geometry, and a forging vector outlining the proposed forging strategy. This forging strategy consists of bite infeed, bite offset, number of offsets and initial position. Moreover, a machine learning architecture with two different units, MLP-based (Multi-Layer Perceptron) and PointNET++-based, was implemented and compared, demonstrating similar performance in terms of recrystallization and deformation errors. However, MLP proved 36 times faster in computational time reaching an average computational time of 5ms pro run. Furthermore, CrystalMind adheres to the volume conservation condition and limits recrystallization error to less than 2% and deformation error to less than 0.1 mm or 0.9% for the test data. Finally, CrystalMind is employed in conjunction with an optimization algorithm, leading to remarkable enhancements in time efficiency. The optimization framework can effectively optimize a forging vector(s) for one, two, or three strokes. For instance, in the case of three strokes, CrystalMind enables the optimization process to be completed within an average of 7 min, a stark contrast to the approximately 2.3 years required when utilizing FEM simulation. Overall, CrystalMind provides fast and accurate predictions of deformed 3D models with corresponding recrystallization per point caused by a forging process, something not to be found in the state of the art so far.

Observing Temporally Varying Synoptic‐Scale Total Alkalinity and Dissolved Inorganic Carbon in the Arctic Ocean

AbstractThe long‐term absorption by the oceans of atmospheric carbon dioxide is leading to the slow decline of ocean pH, a process termed ocean acidification (OA). The Arctic is a challenging region to gather enough data to examine the changes in carbonate chemistry over sufficient scales. However, algorithms that calculate carbonate chemistry parameters from more frequently measured parameters, such as temperature and salinity, can be used to fill in data gaps. Here, these published algorithms were evaluated against in situ measurements using different data input types (data from satellites or in situ re‐analysis climatologies) across the Arctic Ocean. With the lowest uncertainties in the Atlantic influenced Seas (AiS), where re‐analysis inputs achieved total alkalinity estimates with Root Mean Squared Deviation (RMSD) of 21 μmol kg−1 and a bias of 2 μmol kg−1 (n = 162) and dissolved inorganic carbon RMSD of 24 μmol kg−1 and bias of −14 μmol kg−1 (n = 262). AiS results using satellite observation inputs show similar bias but larger RMSD, although due to the shorter time span of available satellite observations, more contemporary in situ data would provide further assessment and improvement. Synoptic‐scale observations of surface water carbonate conditions in the Arctic are now possible to monitor OA, but targeted in situ data collection is needed to enable the full exploitation of satellite observation‐based approaches.

Applying artificial intelligence in healthcare: lessons from the COVID-19 pandemic

ABSTRACT The COVID-19 pandemic exposed vulnerabilities in global healthcare systems and highlighted the need for innovative, technology-driven solutions like Artificial Intelligence (AI). However, previous research on the topic has been limited and fragmented, leading to an incomplete understanding of the ‘what’, ‘where’ and ‘how’ of its application, as well as its associated benefits and challenges. This study proposes a comprehensive AI framework for healthcare and assesses its effectiveness within the UAE's healthcare sector. It provides valuable insights into AI applications for healthcare stakeholders that range from the molecular to the population level. The study covers the different computational techniques employed, from machine learning to computer vision, and the various types of data inputs fed into these techniques, including clinical, epidemiological, locational, behavioural and genomic data. Additionally, the research highlights AI's capacity to enhance healthcare's operational, quality-related and social outcomes, and recognises regulatory policies, technological infrastructure, stakeholder cooperation and innovation readiness as key facilitators of AI adoption. Lastly, we stress the importance of addressing challenges such as data privacy, security, generalisability and algorithmic bias. Our findings are relevant beyond the pandemic in facilitating the development of AI-related policy interventions and support mechanisms for building resilient healthcare sector that can withstand future challenges.

U-net-based semantic classification for flood extent extraction using SAR imagery and GEE platform: A case study for 2019 central US flooding

Data-driven models for water body extraction have experienced accelerated growth in recent years, thanks to advances in processing techniques and computational resources, as well as improved data availability. In this study, we modified the standard U-Net, a convolutional neural network (CNN) method, to extract water bodies from scenes captured from Sentinel-1 satellites of selected areas during the 2019 Central US flooding. We compared the results to several benchmark models, including the standard U-Net and ResNet50, an advanced thresholding method, Bmax Otsu, and a recently introduced flood inundation map archive. Then, we looked at how data input types, input resolution, and using pre-trained weights affect the model performance. We adopted a three-category classification frame to test whether and how permanent water and flood pixels behave differently. Most of the data in this study were gathered and pre-processed utilizing the open access Google Earth Engine (GEE) cloud platform. According to the results, the adjusted U-Net outperformed all other benchmark models and datasets. Adding a slope layer enhances model performance with the 30 m input data compared to training the model on only VV and VH bands of SAR images. Adding DEM and Height Above Nearest Drainage (HAND) model data layer improved performance for models trained on 10 m datasets. The results also suggested that CNN-based semantic segmentation may fail to correctly classify pixels around narrow river channels. Furthermore, our findings revealed that it is necessary to differentiate permanent water and flood pixels because they behave differently. Finally, the results indicated that using pre-trained weights from a coarse dataset can significantly minimize initial training loss on finer datasets and speed up convergence.

Breast Cancer Classification by Using Multi-Headed Convolutional Neural Network Modeling.

Breast cancer is one of the most widely recognized diseases after skin cancer. Though it can occur in all kinds of people, it is undeniably more common in women. Several analytical techniques, such as Breast MRI, X-ray, Thermography, Mammograms, Ultrasound, etc., are utilized to identify it. In this study, artificial intelligence was used to rapidly detect breast cancer by analyzing ultrasound images from the Breast Ultrasound Images Dataset (BUSI), which consists of three categories: Benign, Malignant, and Normal. The relevant dataset comprises grayscale and masked ultrasound images of diagnosed patients. Validation tests were accomplished for quantitative outcomes utilizing the exhibition measures for each procedure. The proposed framework is discovered to be effective, substantiating outcomes with only raw image evaluation giving a 78.97% test accuracy and masked image evaluation giving 81.02% test precision, which could decrease human errors in the determination cycle. Additionally, our described framework accomplishes higher accuracy after using multi-headed CNN with two processed datasets based on masked and original images, where the accuracy hopped up to 92.31% (±2) with a Mean Squared Error (MSE) loss of 0.05. This work primarily contributes to identifying the usefulness of multi-headed CNN when working with two different types of data inputs. Finally, a web interface has been made to make this model usable for non-technical personals.

Potential evaluation of visible-thermal UAV image fusion for individual tree detection based on convolutional neural network

• One of the advantages of the U-Net convolution network is having a large receptive field. • The U-Net network has no fully connected layer which reduces the need for training data. • Using the fusing visible and thermal images, trees in shady areas can be identified with high accuracy. • Isolation of trees from tall boxwoods and also detection of small trees using the normalized digital surface model. • The proposed algorithm can accurately estimate tree crown parameters with high accuracy. Unmanned aerial vehicles (UAVs) outfitted with thermal and visible sensors are already a popular platform in precision agriculture thanks to recent advances in remote sensing. Many researchers have studied integrating data from sensors with different spectral characteristics to achieve higher-level properties and, consequently, detect the trees accurately. In this research, visible and thermal images, as well as normalized digital surface models resulting from UAVs with high spatial resolution, are employed to accurately extract trees from two studied urban areas with complex backgrounds. In the thermal image, trees can be detected in hidden areas based on their brightness temperature difference compared to other features. In contrast, the visible image has a higher spatial resolution, and fusing this data with thermal images can resolve the complexity of the problem. In the proposed method, first, a deep learning network based on visible-thermal data is evaluated in terms of detecting trees with various data approaches. These evaluations include comparison tests on four types of data input to the convolutional network of the visible images, thermal images, fusing visible-thermal images, and also fusing visible-thermal- normalized digital surface model images. Results of evaluation parameters indicate maximum precision in the fourth approach (intersection-over-union = 91.72, F-score = 95.67). Then, the output binary map with the highest accuracy approach and Canny edge detection operator is utilized to accurately identify tree boundaries, count, and estimate the area and diameter of the tree canopy. Finally, the findings revealed the root mean square error (RMSE) first and second areas are 0.21 m 2 , 0.08 m and 0.24 m 2 , 0.11 m respectively for the area and diameter of the tree crown.

FindIT2: an R/Bioconductor package to identify influential transcription factor and targets based on multi-omics data

BackgroundTranscription factors (TFs) play central roles in regulating gene expression. With the rapid growth in the use of high-throughput sequencing methods, there is a need to develop a comprehensive data processing and analyzing framework for inferring influential TFs based on ChIP-seq/ATAC-seq datasets.ResultsHere, we introduce FindIT2 (Find Influential TFs and Targets), an R/Bioconductor package for annotating and processing high-throughput multi-omics data. FindIT2 supports a complete framework for annotating ChIP-seq/ATAC-seq peaks, identifying TF targets by the combination of ChIP-seq and RNA-seq datasets, and inferring influential TFs based on different types of data input. Moreover, benefited from the annotation framework based on Bioconductor, FindIT2 can be applied to any species with genomic annotations, which is particularly useful for the non-model species that are less well-studied.ConclusionFindIT2 provides a user-friendly and flexible framework to generate results at different levels according to the richness of the annotation information of user’s species. FindIT2 is compatible with all the operating systems and is released under Artistic-2.0 License. The source code and documents are freely available through Bioconductor (https://bioconductor.org/packages/devel/bioc/html/FindIT2.html).

A novel sensitivity analysis of commercial building hybrid energy-structure performance

The Architecture, Engineering, and Construction community is highly focused on designing buildings that can improve energy efficiency while maintaining adequate structural behavior (termed herein as energy-structure performance). Sensitivity analysis is a fundamental tool to identify the influential factors under this multidisciplinary setting which can be used for model calibration, model reduction, and design optimization. To enhance the study of energy-structure performance, a novel framework of sensitivity analysis, including initial studies and global sensitivity analysis, was developed based on a general and automated platform of multi-physics simulation. This platform meets the requirements of object-orientation and provides compatibility with parametrical representation, including geometric/non-geometric parameters and detailed modeling. A systematic methodology of sensitivity analysis considering sensitivity auditing and a Scaled Morris Method is developed within the context of hybrid building energy-structure performance. The refined global screening sensitivity analysis (Scaled Morris Method) is proposed to deal with different types of data inputs and multiple-output sensitivity index comparison problems. A building case study was part of the research, which included two types of façades totaling 15 input factors. More than 2000 simulations were conducted producing the yearly/peak energy performance indicators over four climate zones along with static/dynamic structural performance indicators. The results from the initial studies and global sensitivity analysis provided varying/sensitivity effects of building topology/layout, structural/nonstructural components, and inherited features, which were compared quantitatively. For the curtain wall building , the absolute mean (sensitivity indices) input factors were up to 2.3 for energy indicators and 3.0 for structural indicators. The precast concrete panel wall building exhibited the absolute mean of input factors up to 2.8 for energy indicators and 2.4 for structural indicators. Overall, the results indicate this new framework, methodology, and sensitivity analysis can be used for commercial hybrid building energy-structure performance analysis. • Introduced a novel sensitivity framework based on a multi-physics simulation. • Proposed a systematic sensitivity analysis for energy-structure performance. • Proposed a Scaled Morris Method for global sensitivity analysis. • Performed a case study to quantify the building energy-structure indicators.

Imaging and artificial intelligence for progression of age-related macular degeneration.

Age-related macular degeneration (AMD) is a leading cause of severe vision loss. With our aging population, it may affect 288 million people globally by the year 2040. AMD progresses from an early and intermediate dry form to an advanced one, which manifests as choroidal neovascularization and geographic atrophy. Conversion to AMD-related exudation is known as progression to neovascular AMD, and presence of geographic atrophy is known as progression to advanced dry AMD. AMD progression predictions could enable timely monitoring, earlier detection and treatment, improving vision outcomes. Machine learning approaches, a subset of artificial intelligence applications, applied on imaging data are showing promising results in predicting progression. Extracted biomarkers, specifically from optical coherence tomography scans, are informative in predicting progression events. The purpose of this mini review is to provide an overview about current machine learning applications in artificial intelligence for predicting AMD progression, and describe the various methods, data-input types, and imaging modalities used to identify high-risk patients. With advances in computational capabilities, artificial intelligence applications are likely to transform patient care and management in AMD. External validation studies that improve generalizability to populations and devices, as well as evaluating systems in real-world clinical settings are needed to improve the clinical translations of artificial intelligence AMD applications.

Deep learning of topological phase transitions from the point of view of entanglement for two-dimensional chiral p -wave superconductors

Applying deep learning to investigate topological phase transitions (TPTs) becomes a useful method due to not only its ability to recognize patterns but also its statistical excellency to examine the amount of information carried by different types of data inputs. Among possible data types, entanglement-related quantities, such as Majorana correlation matrices (MCMs), one-particle entanglement spectra (OPES), and entanglement eigenvectors (OPEEs), have been proved effective, however, are to date mostly restricted to one dimension. Here, we propose practical input data forms based on those quantities to study TPTs and to compare the efficiency of each form on classic two-dimensional chiral $p$-wave superconductors via the deep learning approach. First, we find that different input forms, either matrices or tensors both originated from real MCMs, can affect the precise locations of the predicted transition points. Next, due to the complex nature of OPEEs, we extract three spatially dependent quantities from OPEEs, one related to the "intensity", and the other two related to "phases" of particle and hole components. We show that similar to taking OPES directly as inputs, solely using "intensity" quantity can only distinguish topological phases from trivial ones, whereas using either whole MCMs or complete OPEE-extracted quantities can provide sufficient information for deep learning to distinguish between phases of matter with different $U(1)$ gauges or Chern numbers. Finally, we discuss certain characteristic features in the deep learning approach and, in particular, they reveal that our trained models indeed learn physically meaningful features, which confirms the potential use even at high dimensions.

A comprehensive review of Bayesian statistics in natural hazards engineering

This study conducts a comprehensive review of the promises and challenges of implementing Bayesian statistics in natural hazards engineering. The reviewed natural hazards include earthquakes, floods, extreme wind events, wildfires, and landslides and debris flows. An attributes matrix is developed under each hazard to analyze each study based on its associated scale of analysis, topic area, Bayesian method, and data resource. In particular, the state-of-the-art survey elaborates the level of involvement for three categories of Bayesian methods, such as Bayesian model updating, Bayesian network, and Bayesian neural network, in the topic areas of hazard analysis, risk assessment, and structural health monitoring. In general, the existing research in natural hazards engineering is benefited by leveraging Bayesian statistics to handle uncertainties explicitly and deal with large-scale problems that involve different types of data inputs. However, the substantial computational cost and the determination of prior probability distributions are two major challenges bottlenecking the future development of Bayesian statistics. Compared with machine learning, Bayesian approaches offer more transparent model inference and exhibit different abilities to avoid data over fitting. This reviewed work can serve as a sound reference for interested practitioners and researchers to practice, develop, and promote broader and more in-depth Bayesian advances in solving grand challenges in natural hazards engineering.

Composite Material Failure Model Updating Approach Leveraging Nondestructive Evaluation Data

Abstract A novel failure model updating methodology is presented in this paper for composite materials. The innovation in the approach presented is found in both the experimental and computational methods used. Specifically, a dominant bottleneck in data-driven failure model development relates to the types of data inputs that could be used for model calibration or updating. To address this issue, nondestructive evaluation data obtained while performing mechanical testing at the laboratory scale are used in this paper to form a damage metric based on a series of processing steps that leverage raw sensing inputs and provide progressive failure curves that are then used to calibrate the damage initiation point computed by full-field three-dimensional finite element simulations of fiber-reinforced composite material that take into account both intra- and interlayer damage. Such curves defined based on nondestructive evaluation data are found to effectively monitor the progressive failure process, and therefore, they could be used as a way to form modeling inputs at different length scales.

Deep learning with wearable based heart rate variability for prediction of mental and general health

The ubiquity and commoditisation of wearable biosensors (fitness bands) has led to a deluge of personal healthcare data, but with limited analytics typically fed back to the user. The feasibility of feeding back more complex, seemingly unrelated measures to users was investigated, by assessing whether increased levels of stress, anxiety and depression (factors known to affect cardiac function) and general health measures could be accurately predicted using heart rate variability (HRV) data from wrist wearables alone. Levels of stress, anxiety, depression and general health were evaluated from subjective questionnaires completed on a weekly or twice-weekly basis by 652 participants. These scores were then converted into binary levels (either above or below a set threshold) for each health measure and used as tags to train Deep Neural Networks (LSTMs) to classify each health measure using HRV data alone. Three data input types were investigated: time domain, frequency domain and typical HRV measures. For mental health measures, classification accuracies of up to 83% and 73% were achieved, with five and two minute HRV data streams respectively, showing improved predictive capability and potential future wearable use for tracking stress and well-being.

Machine Learning for Identification of Craniomaxillofacial Radiographic Lesions

Machine Learning for Identification of Craniomaxillofacial Radiographic Lesions

Siamese-Like Convolutional Neural Network for Fine-Grained Income Estimation of Developed Economies

Estimating the per-capita income and the household income at a fine-grained geographical scale is critical but challenging, even across the developed economies. In this article, a novel Siamese-like Convolutional Neural Network, integrating Ridge Regression and Gaussian Process Regression, has been developed for fine-grained estimation of income across different parts of New York City. Our model (the GP-Mixed-Siamese-like-Double-Ridge model) makes good use of the pairwise comparison of location-based house price information, daytime satellite image, street view and spatial location information as the inputs. Taking the per-capita income and the median household income in New York City as the ground truths, our model outperforms (R 2 = 0.72-0.86 for five-fold validation) other state-of-the-art income estimation models and achieves good performance in cross-district and cross-scale validation. We also find that models which partially share our model architecture, including the Spatial-Information-GP and the Mixed-Siamese-like model, perform well under certain spatial granularity and data availability. Since such models rely on less data input types and simpler architectures, they can be used to save resources on data collection and model training. Hence, using our model for fine-grained income estimation does not mean excluding these models that share similar architectures. Our fine-grained income estimation model can allow the per-capita and the household income data generated in fine-grained resolution to couple with other types of data, such as the air pollution or the epidemic data, of the same scale, to ensure that any location-specific socio-economic-related study and evidence-based decision-making at the fine-grained resolution can be conducted. Future research will focus on extending our model for fine-grained income estimation in developing metropolises, and for developing other socio-economic indicators.

Artificial Intelligence – State of Art Convolution Neural Network Architectures in a Nutshell

It is a well-known fact that all the Artificial Intelligence (AI)researches happening across multiple verticals such as Neuro Imaging, Computer Vision, Deep learning etc point to one master goal of modelling the human brain function by understanding how each part of the brain works. The Convolution neural network (CNN) is one of best deep architecture suitable to handle variety of inputs. In this paper we explore the different types of input data the CNN deep architecture can process and some of the CNN configuration changes that has proved good Accuracy. We have highlighted those specialized CNN architectures along with different types of data inputs they handle including the Functional Magnetic Resonance (fMRI) Neuro Image brain data input.

Standardised electronic algorithms for monitoring prophylaxis of postoperative nausea and vomiting.

IntroductionDespite comprehensive guidelines with high-grade evidence, postoperative nausea and vomiting (PONV) remains a frequent problem in anaesthesia care. Anaesthesia information management systems (AIMS) may aid clinicians in PONV prevention, but their benefit is critically dependent on the details of implementation into practice. This study aimed to examine strengths and weaknesses of the local AIMS-based algorithm in prevention of PONV.Material and methodsThis retrospective study was conducted in the post-anaesthesia care unit (PACU) of a university hospital and included 10 604 patients aged 18 or older who were followed up in the PACU (intracranial, obstetrical or cardiothoracic surgery excluded) from March 2013 until March 2014. The PONV incidence in PACU and AIMS data validity were analysed.ResultsAdherence to PONV guideline recommendations was considerably low, with only 5749 (54%) of the patients receiving correct PONV prophylaxis. Two thousand seven hundred sixty-six (26%) of the patients received an insufficient PONV prophylaxis, which was associated with an excess PONV incidence (11% vs. 4% with correct prophylaxis, p < 0.001) in the PACU. Two thousand four hundred forty-nine (23%) of all patients were discharged from the PACU with an insufficient PONV prophylaxis despite perioperative digital PONV prevention algorithms.ConclusionsAdherence to PONV prophylaxis guidelines in the era of AIMS software and decision support is still remarkably low. The AIMS data usefulness depends on the user, the type of data input and the configuration of the software. Adherence to correct PONV prophylaxis should be re-evaluated systematically before discharge from PACU.

Application of fluorescence spectroscopy and chemometric models for the detection of vegetable oil adulterants in Maltese virgin olive oils.

Fluorescence spectrometry, combined with principle component analysis, partial least-squares regression (PLSR) and artificial neural network (ANN), was applied for the analysis of Maltese extra virgin olive oil (EVOO) adulterated by blending with vegetable oil (corn oil, soybean oil, linseed oil, or sunflower oil). The novel results showed that adjusted PLSR models based on synchronised spectra for detecting the % amount of EVOO in vegetable oil blends had a lower root mean square error (0.02-6.27%) and higher R2 (0.983-1.000) value than those observed when using PLSR on the whole spectrum. This study also highlights the use of ANN as an alternative chemometric tool for the detection of olive oil adulteration. The performance of the model generated by the ANN is highly dependent both on the type of data input and the mode of cross validation; for spectral data which had a variable importance plot value > 0.8 the excluded row cross validation was more appropriate while for complete spectral analysis k-fold or CV-10 was more appropriate.