Survivors of childhood cancer are at risk for obesity, a condition potentially modifiable if dietary intake and physical activity are optimized. These health behaviors are likely influenced by neighborhood of residence, a determinant of access to healthy, affordable food and safe and easy exercise opportunities. We examined associations between neighborhood level factors and obesity among survivors in the St. Jude Lifetime cohort and community comparison group members. Persons with residential addresses available for geocoding were eligible for analysis (n = 2,265, mean age 32.5 [SD 9.1] years, 46% female, 85% white). Survivors completed questionnaires regarding individual behaviors; percent body fat was assessed via dual X-ray absorptiometry (obesity: ≥25% males; ≥35% females); neighborhood effect was characterized using census tract of residence (e.g., neighborhood socioeconomic status [SES], rurality). Structural equation modeling was used to determine associations between neighborhood effect, physical activity, diet, smoking, treatment exposures and obesity. Obese survivors (n = 1,420, 62.7%) were more likely to live in neighborhoods with lower SES (RR: 1.23, 95% CI: 1.10–1.38) and rural areas (RR: 1.22, 95% CI: 1.07–1.39) compared to survivors with normal percent body fat. Resource-poor neighborhoods (standardized effect: 0.06, p < 0.001) and cranial radiation (0.16, p < 0.001) had direct effects on percent body fat. Associations between neighborhood of residence and percent body fat were increased (0.01, p = 0.04) among individuals with a poor diet. Neighborhoods where survivors reside as an adult is associated with obesity. Interventions targeting survivors should incorporate strategies that address environmental influences on obesity. 相似文献
In the area of large-scale graph data representation and semi-supervised learning, deep graph-based convolutional neural networks have been widely applied. However, typical graph convolutional network (GCN) aggregates information of neighbor nodes based on binary neighborhood similarity (adjacency matrix). It treats all neighbor nodes of one node equally, which does not suppress the influence of dissimilar neighbor nodes. In this paper, we investigate GCN based on similarity matrix instead of adjacency matrix of graph nodes. Gaussian heat kernel similarity in Euclidean space is first adopted, which is named EGCN. Then biologically inspired manifold similarity is trained in reproducing kernel Hilbert space (RKHS), based on which a manifold GCN (named MGCN) is proposed for graph data representation and semi-supervised learning with four different kernel types. The proposed method is evaluated with extensive experiments on four benchmark document citation network datasets. The objective function of manifold similarity learning converges very quickly on different datasets using various kernel functions. Compared with state-of-the-art methods, our method is very competitive in terms of graph node recognition accuracy. In particular, the recognition rates of MGCN (Gaussian kernel) and MGCN (Polynomial Kernel) outperform that of typical GCN about 3.8% on Cora dataset, 3.5% on Citeseer dataset, 1.3% on Pubmed dataset and 4% on Cora_ML dataset, respectively. Although the proposed MGCN is relatively simple and easy to implement, it can discover local manifold structure by manifold similarity learning and suppress the influence of dissimilar neighbor nodes, which shows the effectiveness of the proposed MGCN.