关键词： 机器学习   基因交互   全基因组关联分析   单核苷酸多态性   上位性  

摘要： 复杂疾病是基因与基因、基因与环境交互作用的结果,高维基因交互作用的探测给计算带来了极大的挑战。在过去20年间,机器学习方法被用于探测基因-基因交互作用,并取得了一定的效果。本文综述了机器学习方法在基因交互作用探测中的研究进展,系统地介绍了神经网络(neural networks,NN)、随机森林(random forest,RF)、支持向量机(support vector machines,SVM)和多因子降维法(multifactor dimensionality reduction,MDR)等机器学习方法在全基因组关联研究(genome wide association study,GWAS)中探测基因交互作用的原理和局限性,并对未来的研究进行了展望。

关键词： GEP   逻辑回归   地质异常   分类预测   化探数据  

摘要： 高效精准的识别地质异常对成矿预测至关重要。本文将GEP引入到逻辑回归方法中,并对传统单一模型进行改进,克服了单一模型的局限性和缺点。将模型应用于东天山地区的化探数据处理,对斑岩型-矽卡岩型成矿具有指示意义的元素进行了识别应用,结果表明:GEP-逻辑回归法在识别异常信息方面效果良好;与单一模型(神经网络、Knn、逻辑回归)验算结果对比,GEP-逻辑回归法表现更优。

关键词： 数据挖掘   WEKA平台   慢性肾病   Random   Fores  

摘要： 目的:随着医院信息化建设的发展,如何从海量医院临床信息挖掘出隐藏的关联数据,为患者提供早期辅助诊断,是目前智能医疗诊断领域热门的研究方向。方法:在WEKA数据挖掘平台的基础上,对慢性肾脏疾病数据进行挖掘分析,并比较不同挖掘算法的分类准确性。结果:对比实验结果表明,较于其他分类器,Random Forest分类器对慢性肾脏疾病数据集具有较好的分类准确性。结论:基于WEKA数据挖掘平台的医学数据最适合肾脏疾病诊断和预测的算法,为后期医疗行业的大数据分析及挖掘提供新思路。

关键词： 无线传感网络   最小二乘法   时钟同步   时钟偏差   频率偏差   均方误差  

摘要： 提出一种无线传感网络同步的方法,结合发送者-接收者同步(SRS)和接收者-接收者同步(RRS)两种方法,使得位于特定区域的传感器节点通过监听一对超级传感器节点交换时钟消息,借助最小二乘法精确估计出时钟偏差以及频率偏差进行补偿,实现节点间同步。该时钟同步方案的设计与其他已提出的同步方案相比,显著降低了整个网络的能量消耗而不引起任何同步精度的损失。

摘要： Introduction Patients with major depressive disorder (MDD) are likely to experience stress and have a higher risk of developing aging-related diseases [1]. Globally, an estimated 350 million people of all ages suffer from a depression. At the same time, we are facing an aging society, in which manifestations of depression and aging are linked to decreased health, decreased quality of life, and increased healthcare costs. This underlines the importance of developing targeted therapeutic interventions in pursuance of a more healthy and happy society. More evidence is emerging showing that chronological age and biological age may be two different processes that can even diverge. MDD patients seem to feel older and age faster than healthy people of the same age. Current machine learning methods can accurately predict chronological age with correlations of 0.8-0.95 from epigenetics, transcriptomics, proteomics, metabolomics, brain, and other biological data [2]. By calculating the difference between one’s estimated biology-based age and chronological age, one can translate a complex multidimensional aging pattern across biology into one outcome [3]: “the age gap”. A positive age gap represents accelerated aging, whereas a negative brain age gap represents decelerated aging. Previous studies have shown biological age gap associations with physical and cognitive fitness, psychiatric and somatic diseases, health and lifestyle variables, and mortality [4], providing biologically meaningful *** Within the Netherlands Study of Depression and Anxiety (NESDA) [5] we also developed such a biology-based age, specifically one based on DNA methylation patterns (DNAmAge). We selected DNA methylation levels of all 28 million CpG sites in blood as features to predict chronological age in N=1130 subjects (n=811 MDD), using ridge regression. The predictive power was estimated using 10-fold cross-validation. The correlation between chronological age and DNA methylation age est

关键词： Zebrafish   Stress   Meta-analysis   PCA   PLS-DA   R-SVM  

摘要： In order to understand. the mechanisms underlying stress responses, meta-analysis of transcriptome is made to identify differentially expressed genes (DEGs) and their biological, molecular and cellular mechanisms in response to stressors. The present study is aimed at identifying the effect of abiotic and biotic stress factors, and it is found that several stress responsive genes are common for both abiotic and biotic stress factors in zebrafish. The meta-analysis of micro-array studies revealed that almost 4.7% i.e., 108 common DEGs are differentially regulated between abiotic and biotic stresses. This shows that there is a global coordination and fine-tuning of gene regulation in response to these two types of challenges. We also performed dimension reduction methods, principal component analysis, and partial least squares discriminant analysis which are able to segregate abiotic and biotic stresses into separate entities. The supervised machine learning model, recursive-support vector machine, could classify abiotic and biotic stresses with 100% accuracy using a subset of DEGs. Beside these methods, the random forests decision tree model classified five out of 8 stress conditions with high accuracy. Finally, Functional enrichment analysis revealed the different gene ontology terms, transcription factors and miRNAs factors in the regulation of stress responses.

关键词： Landslide susceptibility modeling   Machine learning   Support vector machine (SVM)   Artificial neural network (ANN)   Logistic regression (LR)  

摘要： Landslide is a common natural hazard and responsible for extensive damage and losses in mountainous areas. In this study, Longju in the Three Gorges Reservoir area in China was taken as a case study for landslide susceptibility assessment in order to develop effective risk prevention and mitigation strategies. To begin, 202 landslides were identified, including 95 colluvial landslides and 107 rockfalls. Twelve landslide causal factor maps were prepared initially, and the relationship between these factors and each landslide type was analyzed using the information value model. Later, the unimportant factors were selected and eliminated using the information gain ratio technique. The landslide locations were randomly divided into two groups: 70% for training and 30% for verifying. Two machine learning models: the support vector machine (SVM) and artificial neural network (ANN), and a multivariate statistical model: the logistic regression (LR), were applied for landslide susceptibility modeling (LSM) for each type. The LSM index maps, obtained from combining the assessment results of the two landslide types, were classified into five levels. The performance of the LSMs was evaluated using the receiver operating characteristics curve and Friedman test. Results show that the elimination of noise-generating factors and the separated modeling of each landslide type have significantly increased the prediction accuracy. The machine learning models outperformed the multivariate statistical model and SVM model was found ideal for the case study area.

摘要： The use of retrospective health care claims datasets is frequently criticized for the lack of complete information on potential confounders. Utilizing patient's health status-related information from claims datasets as surrogates or proxies for mismeasured and unobserved confounders, the high-dimensional propensity score algorithm enables us to reduce bias. Using a previously published cohort study of postmyocardial infarction statin use (1998-2012), we compare the performance of the algorithm with a number of popular machine learning approaches for confounder selection in high-dimensional covariate spaces: random forest, least absolute shrinkage and selection operator, and elastic net. Our results suggest that, when the data analysis is done with epidemiologic principles in mind, machine learning methods perform as well as the high-dimensional propensity score algorithm. Using a plasmode framework that mimicked the empirical data, we also showed that a hybrid of machine learning and high-dimensional propensity score algorithms generally perform slightly better than both in terms of mean squared error, when a bias-based analysis is used.

关键词： 巨大儿   超声测量   机器学习  

摘要： 目的:应用机器学习方法,提高预测巨大儿的准确性。方法:查阅2015年1月至2016年12月我院产科巨大儿及正常体质量新生儿病历,获取新生儿出生体质量及超声测量数据。以新生儿实际出生体质量为金标准,评价并比较超声内置的Hadlock公式、多元线性回归以及k近邻、支持向量机、随机森林分类的机器学习方法对巨大儿预测的准确性。结果:内置的Hadlock公式预测巨大儿的灵敏度为40.86%,Youden指数为39.95%;多元线性回归预测巨大儿的灵敏度为60.22%,Youden指数为58.85%;机器学习法中k近邻预测巨大儿的灵敏度86.21%,Youden指数为75.10%;支持向量机预测巨大儿的灵敏度86.21%,Youden指数为73.51%;随机森林预测巨大儿的灵敏度81.03%,Youden指数为71.51%。多元线性回归模型方法的Youden指数大于超声内置公式方法,差异有统计学意义(u=3.64,P<0.001);机器学习法中k近邻、支持向量机、随机森林分类的Youden指数大于超声内置公式的Youden指数(P<0.001),k近邻、支持向量机的Youden指数大于多元线性回归模型法(P<0.05)。结论:机器学习方法预测巨大儿的准确性较高,具有参考应用价值。

关键词： Mobile Robotics   Localization   Omnidirectional Sonar   Machine Learning  

摘要： In the past few years, the mobile robotics has presented a growth in the research field, due to its high application area. The robot localization and its displacement in an unknown environment are fundamental to its operation. Furthemore, signal processing techniques and machine learning methods have been widely appled along with sensors into mobile robot mapping, localization and navigation. In this paper, a low cost and efficient approach for mobile robot localization using sensing by omnidirectional sonar along with signal processing techniques and machine learning methods is proposed. Regarding to classifiers, Bayes classifier, Multilayer Perceptron (MLP), k-Nearest Neighbors kNN, Support Vector Machines (SVM) and Optimum Path Forest (OPF) were analyzed. According to results, OPF using Canberra distance was the best method with accuracy of 98.13 percent and test time of 17.4us.