关键词： 癌症基因组学   生存分析   Cox比例风险模型   随机生存森林模型   LASSO回归  

摘要： 乳腺癌、宫颈鳞状细胞癌、子宫内膜癌、卵巢癌是女性常见的癌症.由于癌症的恶性发展并缺乏有效的早期诊疗手段,这些癌症已成为当今世界女性患者的头号杀手.为了探索高通量组学数据能否促进癌症患者的预后,本研究利用美国癌症基因组图谱项目中四种女性癌症的1861个样本的临床数据和多维组学数据(包括DNA甲基化、mRNA表达、miRNA表达和基于芯片的蛋白表达组学数据),建立了Cox比例风险模型和随机生存森林模型用来回顾性地预测患者的生存率.本研究发现,在宫颈鳞状细胞癌中,通过整合临床与DNA甲基化及miRNA表达组学数据建立的模型,生存预测能力显著高于仅使用临床数据的模型(一致性指数c-index中位数提高了8.73%~15.03%).本研究虽然验证了有些组学数据对特定癌症生存模型的预测能力有提升作用,但也存在着相对于临床数据,组学数据对模型的预测能力无显著提升的情况.这些结果为系统地开展基于癌症基因组学的生存预测研究及提升临床生存分析的预测准确性提供了有用经验.

关键词： 骨强度   QCT   机器学习模型   机器学习方法  

摘要： 当今人口老龄化现象日益显著,骨质疏松、骨质增生等骨退变疾病是中老年群体的常见病、多发病。2018年10月国家卫健委发布的调查结果显示,我国65岁以上人群骨质疏松症患病率高达32.0%;50岁以上人群低骨量率为46.4%。骨质疏松最严重的后果是骨折。骨折的原因是骨的强度/刚度降低,不足以承担外部载荷。临床上常用的评价骨强度的方法是双能X线骨密度仪测量骨密度,然而,测量得到的是二维面积密度,不能准确反映骨强度和相应的骨折风险。骨是具有多层次结构的活的组织,其整体强度由骨的几何形态(包括宏观形态,即宏观形状和大小等,和内部骨小梁结构的排列)和骨材料/骨质的强度(包括矿化程度、胶原特性、显微损伤情况等)决定。基于临床医学影像的生物力学建模仿真方法(又称生物力学CT),是当今技术上最先进的无创评价骨强度的方法。然而,这种生物力学建模仿真方法对于力学要求高,临床医学工作者不容易掌握。因此,迫切需要建立起更为简单方便操作的基于临床医学影像评价骨强度和骨折风险的方法。本研究建立了基于临床QCT影像的机器学习模型,首先根据临床QCT影像建立个体化非线性有限元模型(QCT/FEA)预测骨强度,作为机器学习模型的输出,然后选取临床QCT影像中可以提取的与骨强度高度相关的信息作为输入,并采用参数筛选和参数降维的方法对输入参数进行处理,机器学习模型采用广义回归神经网络(GRNN)和支持向量机(SVM)。具体流程如下:(1)建立基于临床QCT影像的非线性有限元模型计算骨强度:QCT扫描获得骨影像,在Mimics软件里建立三维几何模型,并生成体网格,赋予横观各向同性、非均匀、弹塑性材料属性,在ABAQUS软件中施加载荷和边界条件,给出合适的失效准则,进行非线性显微有限元分析,计算得到骨强度,作为机器学习模型的输出量。(2)机器学习模型的输入量:包括松质骨的材料分布(灰度分布在1、2、5、8、10、15、20、25、30、40、45、50、60、70和75%位点处的灰度值)、骨内区域灰度值(整体、皮质骨和松质骨分区域灰度值)、以及综合特征参数。利用训练集内骨结构的输入量和非线性有限元分析计算的骨强度作为输出量,建立广义回归神经网络和支持向量机两种机器学习模型,并用测试集内骨结构的输入量预测骨强度,用QCT/FEA计算的骨强度来验证机器学习模型的预测效果。即将测试集样本的QCT/FEA计算强度和机器学习模型预测强度绘制Bland-Altman图,分析QCT/FEA模型和机器学习模型所预测椎体强度之间的差异。结果表明,用参数筛选方法和参数降维方法获得一些参数的组合,两种机器学习模型对于骨强度的预测效果很好,且无显著性差异。为了验证训练好的机器学习模型对新样本强度的预测能力,从社区医院随机选取志愿者的QCT影像,运用QCT/FEA计算骨强度,机器学习模型预测骨强度,从相对误差可见机器学习模型预测骨强度准确性较高(相对误差小于8%)。本研究表明,采用基于临床QCT影像的机器学习模型可以有效地预测骨强度,为个体化骨折风险的评估提供了理论方法,便于临床制定相关治疗和干预策略。基于QCT影像预测骨强度不仅可以保证精度,而且对医护人员的操作技术要求较低,因此该方法具有广阔的应用前景。机器学习模型在股骨、脊柱、桡骨和胫骨等部位的强度预测、骨折风险评估方面具有很大的应用潜力。

关键词： Evapotranspiration   Latent heat exchange   Machine learning   Remote sensing   GLASS   MODIS   Regression tree   Neural network   Random kernel   Bootstrap aggregation tree   Flux tower   Surface radiation   Vegetation index   Leaf area index   FPAR   Albedo   Nadir adjusted reflectance   Regularized linear regression   Computational efficiency   Surface energy balance  

摘要： Remote sensing retrieval of evapotranspiration (ET), or surface latent heat exchange (LE), is of great utility for many applications. Machine learning (ML) methods have been extensively used in many disciplines, but so far little work has been performed systematically comparing ML methods for ET retrieval. This paper provides an evaluation of ten ML methods for estimating daily ET based on daily Global LAnd Surface Satellite (GLASS) radiation data and high-level Moderate-Resolution Imaging Spectroradiometer (MOD'S) data products and ground measured ET data from 184 flux tower sites. Measurements of accuracy (RMSE, R-2, and bias) and run time were made for each of ten ML methods with a smaller training data set (n = 7910 data points) and a larger training data set (n = 69,752 data points). Inclusion of more input variables improved algorithm performance but had little effect on run time. The best results were obtained with the larger training data set using the bootstrap aggregation (bagging) regression tree (validation RMSE = 19.91 W/m(2)) and three hidden layer neural network (validation RMSE = 20.94 W/m(2)), although the less computationally demanding random kernel (RKS) algorithm also produced good results (validation RMSE = 22.22 W/m(2)). Comparison of results from sites with different ecosystem types showed the best results for evergreen, shrub, and grassland sites, and the weakest results for wetland sites. Generally, performance was not improved by training with data from only the same ecosystem type.

关键词： Reduced rank KPCA   Nonlinear process monitoring   Fault detection   Tabu search algorithm   Air quality monitoring   Fault isolation  

摘要： Data-driven techniques have been receiving considerable attention in the industrial process monitoring field due to their major advantages of easy implementation and less requirement for the prior knowledge and process mechanism. Principal component analysis (PCA) method is known as a popular method for monitoring and fault detection in industrial systems but as it is basically a linear method. However, most practical systems are nonlinear. To make the extension to nonlinear systems, kernel PCA (KPCA) method has been proposed for process modeling and monitoring. We present in this paper an online reduced rank optimized KPCA (RR-KPCA) technique for fault detection in order to extend the advantages of the KPCA models to online processes. Following the fault detection, the identification of the variables correlated to the fault occurred is of great importance. For this purpose, it is proposed to extend the approaches of localization by partial PCA and by elimination in the linear case to the nonlinear case, by exploiting the solution of reduction of the dimension of the kernel matrix in the feature space. The partial RR-KPCA and the elimination sensor identification (ESI-RRKPCA) are generated based on the static RR-KPCA and the online RR-KPCA methods. The idea of these approaches is to generate partial RR-KPCA models with reduced sets of variables. In other words, their goal is to generate indices of fault detection sensitive to certain faults and insensitive to others. The proposed fault isolation methods are applied for monitoring an air quality monitoring network (AIRLOR) data.

关键词： Causal inference   positivity   doubly robust   IPTW   TMLE   super learner  

摘要： Data-adaptive methods have been proposed to estimate nuisance parameters when using doubly robust semiparametric methods for estimating marginal causal effects. However, in the presence of near practical positivity violations, these methods can produce a separation of the two exposure groups in terms of propensity score densities which can lead to biased estimates of the treatment effect. To motivate the problem, we evaluated the Targeted Minimum Loss-based Estimation procedure using a simulation scenario to estimate the average treatment effect. We highlight the divergence in estimates obtained when using parametric and data-adaptive methods to estimate the propensity score. We then adapted an existing diagnostic tool based on a bootstrap resampling of the subjects and simulation of the outcome data in order to show that the estimation using data-adaptive methods for the propensity score in this study may lead to large bias and poor coverage. The adapted bootstrap procedure is able to identify this instability and can be used as a diagnostic tool.

关键词： Learning-to-rank   machine learning   supervised learning   artificial intelligence  

摘要： Learning-to-rank is one of the learning frameworks in machine learning and it aims to organize the objects in a particular order according to their preference, relevance or ranking. In this paper, we give a comprehensive survey for learning-to-rank. First, we discuss the different approaches along with different machine learning methods such as regression, SVM, neural network-based, evolutionary, boosting method. In order to compare different approaches: we discuss the characteristics of each approach. In addition to that, learning-to-rank algorithms combine with other machine learning paradigms such as semi-supervised learning, active learning, reinforcement learning and deep learning. The learning-to-rank models employ with parallel or big data analytics to review computational and storage advantage. Many real-time applications use learning-to-rank for preference learning. In regard to this, we introduce some representative works. Finally, we highlighted future directions to investigate learning-to-rank methods.

关键词： machine learning   microbial risk assessment   support vector machines   whole genome sequencing  

摘要： Next‐generation sequencing (NGS) data present an untapped potential to improve microbial risk assessment (MRA) through increased specificity and redefinition of the hazard. Most of the MRA models do not account for differences in survivability and virulence among strains. The potential of machine learning algorithms for predicting the risk/health burden at the population level while inputting large and complex NGS data was explored with Listeria monocytogenes as a case study. Listeria data consisted of a percentage similarity matrix from genome assemblies of 38 and 207 strains of clinical and food origin, respectively. Basic Local Alignment (BLAST) was used to align the assemblies against a database of 136 virulence and stress resistance genes. The outcome variable was frequency of illness, which is the percentage of reported cases associated with each strain. These frequency data were discretized into seven ordinal outcome categories and used for supervised machine learning and model selection from five ensemble algorithms. There was no significant difference in accuracy between the models, and support vector machine with linear kernel was chosen for further inference (accuracy of 89%[95% CI: 68%, 97%]). The virulence genes FAM002725, FAM002728, FAM002729, InlF, InlJ, Inlk, IisY, IisD, IisX, IisH, IisB, lmo2026, and FAM003296 were important predictors of higher frequency of illness. InlF was uniquely truncated in the sequence type 121 strains. Most important risk predictor genes occurred at highest prevalence among strains from ready‐to‐eat, dairy, and composite foods. We foresee that the findings and approaches described offer the potential for rethinking the current approaches in MRA. [ABSTRACT FROM AUTHOR]

关键词： Vehicle ownership   Phnom Penh   Features ranking   Multinomial logit model   Neural networks   Random forests  

摘要： Vehicle ownership patterns and their determinants play an important role in transportation policy-making. This issue has been paid even greater attention in developing countries that aspire to reach sustainable transportation development goals in the era of urbanization and globalization. In this study, the multinomial logit model, neural networks and random forest were applied to examine the features' impact level and to also predict vehicle ownership patterns in Phnom Penh city. The empirical results indicate that household income is the most powerful variable affecting motorization in Phnom Penh. Supplementation of individual trip characteristics such as total number of trips made, number of trips made for work purposes and overall travel distance all make effective contributions as classifiers. Furthermore, it is acknowledged that the machine-learning approach outperformed not only in terms of predicting accuracy, but also in dealing with unbalanced categories when compared with the statistical approach. This detection supplies the advantages of applying machine learning techniques in terms of, but not limited to, the field of vehicle ownership.

关键词： methods: data analysis   techniques: photometric   catalogues   galaxies: distances and redshifts   galaxies: photometry  

摘要： Star formation rates (SFRs) are crucial to constrain theories of galaxy formation and evolution. SFRs are usually estimated via spectroscopic observations requiring large amounts of telescope time. We explore an alternative approach based on the photometric estimation of global SFRs for large samples of galaxies, by using methods such as automatic parameter space optimisation, and supervised machine learning models. We demonstrate that, with such approach, accurate multiband photometry allows to estimate reliable SFRs. We also investigate how the use of photometric rather than spectroscopic redshifts, affects the accuracy of derived global SFRs. Finally, we provide a publicly available catalogue of SFRs for more than 27 million galaxies extracted from the Sloan Digital Sky Survey Data Release 7. The catalogue will be made available through the Vizier facility.

关键词： Nonparametric regression   Local approximate Gaussian Process   Structure-property linkages   High-contrast composites   Gaussian Process regression  

摘要： Reduced-order structure-property (S-P) linkages play a pivotal role in the tailored design of materials for advanced engineering components. There is a critical need to distill these from the simulation datasets aggregated using sophisticated, computationally expensive, physics-based numerical tools (e.g., finite element methods). The recent emergence of materials data science approaches has opened new avenues for addressing this challenge. In this paper, we critically compare the relative merits of the application of four distinct machine learning approaches for their efficacy in extracting microstructure-property linkages from the finite element simulation data aggregated on high-contrast elastic composites with different microstructures. The machine learning approaches selected for the study have included different combinations of local/global and parametric/nonparametric approaches. Furthermore, the nonparametric approaches selected for this study are based on Gaussian Process (GP) models that allow for a formal treatment of uncertainty quantification in the predicted values. The predictive performances of these different approaches have been compared against each other using rigorous cross-validation error metrics. Furthermore, their sensitivity to both the dataset size and dimensionality has been investigated.