In several studies the sudden infant death syndrome (SIDS) has been significantly associated with sleeping in the prone position. It is not known how the prone position increases the risk of SIDS. We analyzed data from a case-control study (58 infants with SIDS and 120 control infants) and a prospective cohort study (22 infants with SIDS and 213 control infants) in Tasmania. Interactions were examined in matched analyses with a multiplicative model of interaction. In the case-control study, SIDS was significantly associated with sleeping in the prone position, as compared with other positions (unadjusted odds ratio, 4.5; 95 percent confidence interval, 2.1 to 9.6). The strength of this association was increased among infants who slept on natural-fiber mattresses (P = 0.05), infants who were swaddled (P = 0.09), infants who slept in heated rooms (P = 0.006), and infants who had had a recent illness (P = 0.02). These variables had no significant effect on infants who did not sleep in the prone position. A history of recent illness was significantly associated with SIDS among infants who slept prone (odds ratio, 5.7; 95 percent confidence interval, 1.8 to 19) but not among infants who slept in other positions (odds ratio, 0.83). In the cohort study, the risk of SIDS was greater among infants who slept prone on natural-fiber mattresses (odds ratio, 6.6; 95 percent confidence interval, 1.3 to 33) than among infants who slept prone on other types of mattresses (odds ratio, 1.8). When infants sleep prone, the elevated risk of SIDS is increased by each of four factors: the use of natural-fiber mattresses, swaddling, recent illness, and the use of heating in bedrooms.

The Three‐Georges Dam holds many records in the history of engineering. While the dam has produced benefits in terms of flood control, hydropower generation and increased navigation capacity of the Yangtze River, serious questions have been raised concerning its impact on both upstream and downstream ecosystems. It has been suggested that the dam operation intensifies the extremes of wet and dry conditions in the downstream Poyang Lake, and affects adversely important local wetlands. A floodgate has been proposed to maintain the lake water level by controlling the flow between the Poyang Lake and Yangtze River. Using extensive hydrological data and generalized linear statistical models, we demonstrated that the dam operation induces major changes in the downstream river discharge near the dam, including an average “water loss”. The analysis also revealed considerable effects on the Poyang Lake water level, particularly a reduced level over the dry period from late summer to autumn. However, the dam impact needs to be further assessed based on long‐term monitoring of the lake ecosystem, covering a wide range of parameters related to hydrological and hydraulic characteristics of the lake, water quality, geomorphological characteristics, aquatic biota and their habitat, wetland vegetation and associated fauna.

The analysis of large genomic data is hampered by issues such as a small number of observations and a large number of predictive variables (commonly known as “large P small N”), high dimensionality or highly correlated data structures. Machine learning methods are renowned for dealing with these problems. To date machine learning methods have been applied in Genome-Wide Association Studies for identification of candidate genes, epistasis detection, gene network pathway analyses and genomic prediction of phenotypic values. However, the utility of two machine learning methods, Gradient Boosting Machine (GBM) and Extreme Gradient Boosting Method (XgBoost), in identifying a subset of SNP makers for genomic prediction of breeding values has never been explored before. In this study, using 38,082 SNP markers and body weight phenotypes from 2,093 Brahman cattle (1,097 bulls as a discovery population and 996 cows as a validation population), we examined the efficiency of three machine learning methods, namely Random Forests (RF), GBM and XgBoost, in (a) the identification of top 400, 1,000, and 3,000 ranked SNPs; (b) using the subsets of SNPs to construct genomic relationship matrices (GRMs) for the estimation of genomic breeding values (GEBVs). For comparison purposes, we also calculated the GEBVs from (1) 400, 1,000, and 3,000 SNPs that were randomly selected and evenly spaced across the genome, and (2) from all the SNPs. We found that RF and especially GBM are efficient methods in identifying a subset of SNPs with direct links to candidate genes affecting the growth trait. In comparison to the estimate of prediction accuracy of GEBVs from using all SNPs (0.43), the 3,000 top SNPs identified by RF (0.42) and GBM (0.46) had similar values to those of the whole SNP panel. The performance of the subsets of SNPs from RF and GBM was substantially better than that of evenly spaced subsets across the genome (0.18–0.29). Of the three methods, RF and GBM consistently outperformed the XgBoost in genomic prediction accuracy.

The method of generalised estimating equations for regression modelling of clustered outcomes allows for specification of a working matrix that is intended to approximate the true correlation matrix of the observations. We investigate the asymptotic relative efficiency of the generalised estimating equation for the mean parameters when the correlation parameters are estimated by various methods. The asymptotic relative efficiency depends on three features of the analysis, namely (i) the discrepancy between the working correlation structure and the unobservable true correlation structure, (ii) the method by which the correlation parameters are estimated and (iii) the ‘design’, by which we refer to both the structures of the predictor matrices within clusters and distribution of cluster sizes. Analytical and numerical studies of realistic data‐analysis scenarios show that choice of working covariance model has a substantial impact on regression estimator efficiency. Protection against avoidable loss of efficiency associated with covariance misspecification is obtained when a ‘Gaussian estimation’ pseudolikelihood procedure is used with an AR(1) structure.