On the Estimation of Heritability from Unbalanced Data

The distribution and moments, of ANOVA estimator of heritability are given under unbalanced random model. These expressions are used to investigate the effect of unbalancedness on the bias and variance/MSE of the estimator and also the validity of certain approximations for its variance, numerically. The computed results reveal that the unbalancedness increases both the bias and variance/MSE of the estimator and the Smith-approximation for the variance of the estimator provides better accuracy.     

Influence of net charge on the aggregation and solubility behaviour of amphotericin B and its derivatives in aqueous media

The poor solubility of polyene antibiotics in aqueous media limits their application in the therapy of systemic fungal infections. In the present paper we have demonstrated that the ionic state (net electrical charge) of the antibiotic molecule is an important factor in determining the aggregation and solubility properties of amphotericin B and its derivatives. A multi-step model of polyene self-association in aqueous media has been proposed as an explanation for the fact that some major differences are observed when aggregation is monitored by different techniques. Offprint requests to: J. Mazerski     

基于指标自动筛选的新疆开孔河流域生态健康评价

生态健康评价对了解区域生态健康状况和促进区域可持续发展具有重要意义,如何自动筛选出能反映生态系统特性的重要指标,是生态健康定量评估的关键问题。基于压力-状态-响应(PSR,Press-State-Response)框架和生态等级网络框架(EHN,Ecological Hierarchy Network),通过文献调研和因果分析建立要素层与指标层之间的交叉联系,构建了生态健康评价"网状"指标体系;在保证指标体系完备性基础上,通过结合主成分分析和熵权法的候选指标权重的客观计算,基于目标优化理论构建了评价指标的自动筛选模型,并基于中选指标计算了新疆开孔河流域2001—2017年生态健康指数(EHCI,Ecological Health Comprehensive Indexes),分析其空间分异和时间变化特征。结果表明:利用所建立的评价指标自动筛选模型,开孔河流域生态健康评价指标由31个候选指标自动筛选出了17个中选指标,用54.8%的指标表达了85.98%的信息,中选的17个指标在干旱/半干旱区域有关文献中应用较多,使用频次比例都在20%以上,其中归一化植被指数(NDVI,Normalized Difference Vegetation Index)、年降水量和植被覆盖度(FVC,Fractional Vegetation Coverage)3个指标的使用频次百分比均超过了50%,说明指标自动筛选模型的合理性;开孔河流域空间分布差异显著,总体上西北高、东南低,东南部和中部绿洲区外围生态健康状况较差,西北部河谷地带和中部两大绿洲区生态健康状况较好;17年来,流域生态质量整体趋于改善,显著改善区域占10.26%,远高于显著退化的1.61%,显著改善区域以孔雀河绿洲最为明显。开孔河流域生态健康的总体好转趋势说明区域生态综合治理取得一定成效。     

Geographic Information System (GIS) modeling approach to determine the fastest delivery routes

This study involves the adoption of the Geographic Information System (GIS) modeling approach to determine the quickest routes for fresh vegetable delivery. During transport, fresh vegetables mainly deteriorate on account of temperature and delivery time. Nonetheless, little attention has been directed to transportation issues in most areas within Kuala Lumpur. In addition, perishable food normally has a short shelf life, thus timely delivery significantly affects delivery costs. Therefore, selecting efficient routes would consequently reduce the total transportation costs. The regression model is applied in this study to determine the parameters that affect route selection with respect to the fastest delivery of fresh vegetables. For the purpose of this research, ArcGIS software with network analyst extension is adopted to solve the problem of complex networks. The final output of this research is a map of quickest routes with the best delivery times based on all variables. The variables tested from regression analysis are the most effective parameters to make the flow of road networks slower. The objective is to improve the delivery services by achieving the least drive time. The main findings of this research are that Land use such as residential area and population as variables are the effective parameters on drive time.     

免疫细胞浸润在非小细胞肺癌诊断与预后中的应用

免疫细胞浸润对癌症的诊断与预后有着重要意义。文中收集TCGA数据库已收录的非小细胞肺癌肿瘤与正常组织基因表达数据,利用CIBERSORT工具得到22种免疫细胞占比来评估免疫细胞浸润情况。以22种免疫细胞占比为特征,用机器学习方法构建了非小细胞肺癌肿瘤与正常组织的分类模型,其中随机森林方法构建的模型分类效果AUC=0.987、敏感性0.98及特异性0.84。并且用随机森林方法构建的肺腺癌和肺鳞癌肿瘤组织分类模型效果AUC=0.827、敏感性0.75及特异性0.77。用LASSO回归筛选22种免疫细胞特征,保留8种强相关特征组成的免疫细胞评分结合临床特征构建了非小细胞肺癌预后模型。经评估及验证,预后模型C-index=0.71并且3年和5年的校准曲线拟合良好,可以对预后风险度进行准确预测。本研究基于免疫细胞浸润所构建的分类模型与预后模型,旨在对非小细胞肺癌的诊断与预后研究提供新的策略。     

A single-spring model predicts the majority of variance in impact force during a fall onto the outstretched hand

The objective of this study was to validate a single-spring model in predicting measured impact forces during an outstretched arm falling scenario. Using an integrated force plate, impact forces were assessed from 10 young adults (5 males; 5 females), falling from planted knees onto outstretched arms, from a random order of drop heights: 3, 5, 7, 10, 15, 20, and 25 cm. A single-spring model incorporating body mass, drop height plus the estimated linear stiffness of the upper extremity (hand, wrist and arm) was used to predict impact force on the hand. We used an analysis of variance linearity test to test the validity of using a linear stiffness coefficient in the model. We used linear regression to assess variance (R²) in experimental impact force predicted by the single-spring model. We derived optimum linear stiffness coefficients for male, female and sex-combined. Our results indicated that the association between experimental and predicted impact forces was linear (P < 0.05). Explain variance in experimental impact force was R² = 0.82 for sex-combined, R² = 0.88 for males and R² = 0.84 for females. Optimum stiffness coefficients were 7436 N/m for sex-combined, 8989 N/m for males and 4527 N/m for females. In conclusion, a linear spring coefficient used in the single-spring model proved valid for predicting impact forces from fall heights up to 25 cm. Results also suggest the use of sex-specific spring coefficients when estimating impact force using the single-spring model. This model may improve impact force to bone strength ratios (factor-of-risk) and prediction of forearm and wrist fracture.