关键词： Cornish-Fisher expansion   feature selection   nearest shrunken centroid   sparse PCA  

摘要： Type 2 diabetes mellitus (T2DM) affects millions of people through its life-altering complications. Worldwide, 3.4 million people die of diabetes annually. Studying the effect of genetic polymorphism on T2DM has been plagued by the available sample size. A 2016 Nature Reviews article summarized that the accuracy of predicting future type 2 diabetes from genetic polymorphism is very low at the population level. Innumerable associations between genes, environmental factors, and type 2 diabetes remain to be discovered. This research presents a method to identify subtle effects of genetic variants using whole genome sequencing data and improve prediction accuracy of T2DM at the population level. To achieve this, a new feature selection procedure and a classifier are proposed. The method involves (a) first applying sparse principal component analysis to genotype data to obtain orthogonal features;(b) building a new classifier using single nucleotide polymorphism (SNP)-specific regularization parameters to reduce the false positive rate of feature selection;(c) verifying feature relevance through penalized logistic regression. After application to a dataset containing 625 597 SNPs and 23 environmental variables from each of 3326 humans, the method identified 271 genetic variants with subtle effects on T2DM prediction. These variants led to greatly improved prediction accuracy for new patients at the population level. The proposed method also has the advantage of computational efficiency, over 15 times faster than random forest and extreme gradient boosting (XGBoost) classifiers, and thus provides a promising tool for large-scale genome-wide association studies.

关键词： Amphibole formula   Amphibole nomenclature   Machine learning   Principal components regression   Electron microprobe analysis  

摘要： Wepresent a new method for calculating amphibole formula from routine electron microprobe analysis (EMPA) data by applying a principal components regression (PCR)-based machine learning algorithm on reference amphibole data. The reference amphibole data collected from literature are grouped in two datasets, for Li-free and Li-bearing amphiboles respectively, which include Fe2+, Fe3+, OH contents and the ion site assignments determined by single crystal structure refinement. We established two PCR models, for Li-free and Li-bearing amphiboles respectively, by the 10-fold cross validation of training datasets and evaluated by independent test datasets. The results show that our models can successfully reproduce the reference data for most ions with an error less than +/- 0.01 atom per formula unit (apfu), for Fe3+ within an error less than +/- 0.2 apfu and for (OH)-O-W and O-W(2-) with errors less than +/- 0.3 apfu. The error in estimated Fe3+/Sigma Fe ratio shows a rough negative dependence on FeOT content (total iron expressed as FeO), ranging within +/- 0.3 for amphiboles with FeOT >= 5 wt% and within +/- 0.2 for amphiboles with FeOT >= 10 wt%. Our models are applicable to both (W)(OH, F, Cl)-dominant and O-W-dominant amphiboles. It is notable that this method is not suitable for calculating mineral formula of amphiboles that have been affected by deprotonation as a result of secondary oxidation, but it could offer an estimation of initial (OH)-O-W prior to the post-formation oxidation. A user-friendly Excel worksheet is provided with two independent PCR models for calculating the formula of Li-free amphibole and Li-bearing amphibole, respectively. An automatic nomenclature function is also provided according to the nomenclature criteria of the 2012 International Mineralogical Association (IMA) report. (C) 2020 Elsevier B.V. All rights reserved.

关键词： Throttling process   Artificial neural network   Multilayer perceptron   Group method of data handling   Natural gas temperature drop   Natural gas compositions effects  

摘要： It is vital for the designers of the throttling facilities to predict natural gas temperature drop along a throttling valve exactly. Generally, direct prediction of the temperature drop is not possible even by employing equations of states. In this work, artificial neural network method, specifically multilayer perceptron, is utilized to predict the physical properties of natural gas. Then, the method is employed for direct calculation of the temperature drop along a throttling process. To train, validate and test the network, a large database of natural gas fields of Iran plus some experimental data (30,000 random datasets) are gathered from the literature. In addition, according to complexity of the multilayer perceptron model, a group method of data handling approach is used to simplify the major trained network. For the first time, an equation is developed for calculating natural gas temperature drop as a function of molecular weight as well as pressure drop. The results show that the multilayer perceptron and group method of data handling methods have the error R-2 = 0.998 and R-2 = 0.997, respectively. In addition, the results indicate that both developed machine learning methods present a high accuracy in the calculations over a wide range of gas mixtures and input properties ranges.

关键词： Power quality disturbances   2D-Riesz Transform   Multi-objective grey wolf optimizer   Machine learning methods   Classification  

摘要： In this study, a new method combined with two-dimensional Riesz Transform (RT) in feature extraction stage and Multi-Objective Grey Wolf Optimizer (MOGWO) with k-Nearest Neighbor (KNN) algorithm in the feature selection stage is introduced to classify Power Quality (PQ) disturbances. After determining the most suitable feature group, classification models are created by using machine learning approaches. Although one-dimensional (1D) signal processing methods by nature are used in the classification stage of PQ disturbances, it is observed that studies have developed in the literature including two-dimensional (2D) signal processing. 2D signal processing approach is used because it gives good feature diversity and leads to creation a good model. In this study, firstly PQ disturbances events data is collected synthetically and experimentally. 1D signals are converted to 2D signals to apply 2D-RT. In 2D-RT, it is obtained 12 sub bands matrices to find better features for one 2D matrix. 15 statistical and image-based features are calculated for each band. Totally 180 features are obtained for one sub bands matrix. At this point, with the MOGWO-KNN method, it is aimed to create a simple classification model with high performance by selecting the most suitable features obtained by 2D-RT. The models based on KNN, SVM, MLP and ensemble learner methods are created to investigate if there is a better classification accuracy or not. The simulation study is also done for data consists of noisy (40 dB, 30 dB, 20 dB noise levels) and multiple events. The model can classify 9 types of multiple disturbances in 18 PQ disturbances. At the same time, a robust model that classify even noisy situations is created. It is showed that the proposed PQ disturbances classification method gives high performance compared to the methods in the literature for both simulations and real data. (C) 2020 Elsevier Inc. All rights reserved.

关键词： machine learning   physical fitness test   successful aging  

摘要： Aim The purpose of this study was to develop a machine learning prediction model for successful aging (SA) based on physical fitness tests. Methods A total of 3657 community-dwelling adults aged >= 60 years from Nanchang city were recruited in this study. A 3-year follow-up test was carried out for all the participants to determine whether they turn to non-SA. Developed questionnaires and physical fitness tests were used to obtain overall health condition, balance, agility, speed, reactions and gait. Four machine learning models (logistic regression, deep learning, random forest and gradient boosting decision tree) were applied to develop the prediction models, the analyzed sample was 890. Results The baseline prevalence of successful aging was 26.99%, The average annual incidence rate of SA to non-SA was 11.04%. There were significant differences between the SA and non-SA groups for all physical fitness tests at baseline. The accuracy and area under the curve of all four machine learning models was >85%, the positive predictive value and sensitivity was >75%, and the specificity was >86% on the average. The deep learning model outperformed the other model, with area under the curve 90.00%, accuracy 89.3%, positive predictive value 85.8% and specificity 93.1%, respectively. Compared with other models, the logistic regression model performed best in sensitivity. Age, arm curl, 30-s sit-to-stand and reaction time were important predictors in all models. Conclusion The deep learning model is ideal in the prediction of SA maintenance, and the corresponding physical fitness interventions are essential to ensuring SA. Geriatr Gerontol Int 2020;center dot center dot: center dot center dot-center dot center dot.

关键词： Radiomics   Nasopharyngeal carcinoma   Machine learning   Diagnosis   PET   CT  

摘要： Purpose To identify optimal machine learning methods for radiomics-based differentiation of local recurrence versus inflammation from post-treatment nasopharyngeal positron emission tomography/X-ray computed tomography (PET/CT) images. Procedures Seventy-six nasopharyngeal carcinoma (NPC) patients were enrolled (41/35 local recurrence/inflammation as confirmed by pathology). Four hundred eighty-seven radiomics features were extracted from PET images for each patient. The diagnostic performance was investigated for 42 cross-combinations derived from 6 feature selection methods and 7 classifiers. Of the original cohort, 70 % was applied for feature selection and classifier development, and the remaining 30 % used as an independent validation set. The diagnostic performance was evaluated using area under the ROC curve (AUC), test error, sensitivity, and specificity. Furthermore, the performance of the radiomics signatures against routine features was statistically compared using DeLong's method. Results The cross-combination fisher score (FSCR) + k-nearest neighborhood (KNN), FSCR + support vector machines with radial basis function kernel (RBF-SVM), FSCR + random forest (RF), and minimum redundancy maximum relevance (MRMR) + RBF-SVM outperformed others in terms of accuracy (AUC 0.883, 0.867, 0.892, 0.883;sensitivity 0.833, 0.864, 0.831, 0.750;specificity 1, 1, 0.873, 1) and reliability (test error 0.091, 0.136, 0.150, 0.136). Compared with conventional metrics, the radiomics signatures showed higher AUC values (0.867-0.892 vs. 0.817), though the differences were not statistically significant (p = 0.462-0.560). Conclusion This study identified the most accurate and reliable machine learning methods, which could enhance the application of radiomics methods in the precision of diagnosis of NPC.

关键词： Lithological group   Pattern recognition   Multivariate data   Sedimentary rocks  

摘要： Specific computational tools assist geologists in identifying and sorting lithologies in well surveys and reducing operational costs and practical working time. This allows for the management of professional output, the efficient interpretation of data, and completion of scientific research on data collected in geologically distinct regions. Machine learning methods and applications integrate large sets of information with the goal of efficient pattern recognition and the capability of leveraging accurate decision making. The objective of this study is to apply machine learning methods to the supervised classification of lithologies using multivariate log parameter data from offshore wells from the International Ocean Discovery Program (IODP). According to the analysis of the lithologies proposed in the IODP Expeditions and for the application of our methods, the lithologies were divided into four groups. The IODP Expeditions were organized into four templates for better results in analyzing the set of expeditions and practical application of the methods. The templates were submitted to training, validation, and testing by multilayer perceptron (MLP), decision tree, random forest, and support vector machine (SVM) methods. The evaluation was randomly divided into training (70%), validation (10%), and testing (20%) using the classification methods as an evaluation of the results. In the results, it was observed that Template1 (IODP Expedition 362) obtained better results with the MLP method, Template2 (IODP Expeditions 354, 355, and 359) and Template3 (IODP Expeditions 354, 355, 359, and 362) obtained better results with the random forest method with greater than 80.00% accuracy. For cross-validation, the random forest method performed well in all scenarios. In the practical template, the G2 group obtained a better result with the MLP method with an average accuracy above 85.00%. It is expected that machine learning methods can help improve the study of geology with a

摘要： Waist circumference (cm)Arm circumference (cm)Thigh circumference (cm)Subscapular skin thickness (mm)Triceps skin (mm)Weight (in kg)BMI (kg/m2)Pulse rate (/min)Head fat (gm)Head lean mass(gm)Head fat %Left arm fat %Left leg fat %Right arm fat ( gm)Right leg fat %Thoracic BMD (bone mineral density)Lumbar BMD (gm/cm2)Pelvis BMD (gm/cm2)Truncal fat mass (gm)Truncal lean mass (gm)Total truncal mass (gm)Truncal fat %Subtotal lean mass (gm)Subtotal mass (gm)Subtotal fat %Total BMD (gm/cm2)Total fat mass (gm)Total fat %Total lean mass (gm)Age (in years)GenderThe findings show us that though age is the most important factor in the determination of BP, body composition metrics are superior to BMI and weight as features of importance. The results of our study are shown in Figure1. Stacking regression achieved the highest R2 value and lowest MAE (Mean Absolute Error) for predicting the average systolic and diastolic blood pressure. Our study did not include categorical variables like history of smoking, presence of other factors like anti‐hypertensive drug use, ethnicity, etc that might limit the overall significance. Certain important determinants of cardiovascular risk like pulse pressure could not be used, since they are derived from systolic and diastolic blood pressure. Our study, however, is able to show that DXA measured body composition is a relatively inexpensive tool, associated with a low level of radiation exposure, that might be useful in identifying population at risk for developing hypertension, especially when traditional markers like BMI and waist circumference might not be adequate in capturing the ***.

关键词： Business decision   Binary NPD strategy forecasting   Machine learning   Incremental NPD strategy   Radical NPD strategy  

摘要： The new product development (NPD) is crucial to firms' survival and success. Tough decisions must be made between the binary NPD strategy (i.e. incremental NPD strategy and radical NPD strategy) to ensure that scarce resources are allocated efficiently. The inappropriate NPD strategy that does not meet the internal and external conditions may lead to resources waste and performance decline. The binary NPD strategy forecasting is helpful to guide the firms when to improve existing products and when to develop 'really new' products. Therefore, the primary purposes of this study are to construct an evaluating indicator system and to find the appropriate method for the binary NPD strategy forecasting. Here we obtain 1088 valid sample datasets from Chinese automotive industry, covering the period 2001-2014. The empirical results indicate that RS-MultiBoosting as a kind of hybrid ensemble machine learning (HEML) method demonstrate an outstanding forecasting performance in dealing with the small datasets by comparison with the other four ensemble machine learning (EML) methods and three individual machine learning (IML) methods. The findings can help firms to make the right business decision between incremental and radical NPD strategies so that they can avoid resources waste and improve the overall NPD performance.

关键词： RGB camera   filter   low cost   near-infrared   NDVI   machine learning  

摘要： The normalized difference vegetation index (NDVI) is widely used in remote sensing to monitor plant growth and chlorophyll levels. Usually, a multispectral camera (MSC) or hyperspectral camera (HSC) is required to obtain the near-infrared (NIR) and red bands for calculating NDVI. However, these cameras are expensive, heavy, difficult to geo-reference, and require professional training in imaging and data processing. On the other hand, the RGBN camera (NIR sensitive RGB camera, simply modified from standard RGB cameras by removing the NIR rejection filter) have also been explored to measure NDVI, but the results did not exactly match the NDVI from the MSC or HSC solutions. This study demonstrates an improved NDVI estimation method with an RGBN camera-based imaging system (Ncam) and machine learning algorithms. The Ncam consisted of an RGBN camera, a filter, and a microcontroller with a total cost of only $70 similar to 85. This new NDVI estimation solution was compared with a high-end hyperspectral camera in an experiment with corn plants under different nitrogen and water treatments. The results showed that the Ncam with two-band-pass filter achieved high performance (R2 = 0.96, RMSE = 0.0079) at estimating NDVI with the machine learning model. Additional tests showed that besides NDVI, this low-cost Ncam was also capable of predicting corn plant nitrogen contents precisely. Thus, Ncam is a potential option for MSC and HSC in plant phenotyping projects.