Risk factors and SCN5A-H558R polymorphism for atrial fibrillation in Tibetans living at different altitudes.

Several studies have found associations of genes with atrial fibrillation (AF), including SCN5A-H558R. However, there are limited data of these associations among populations living at different altitudes. We investigated the relationship between the SCN5A-H558R polymorphism and AF in Tibetans living at different altitudes in Qinghai, China. General clinical and genotype data were obtained from 72 patients with AF and 109 non-AF (NAF) individuals at middle altitudes, and from 102 patients with AF and 143 NAF individuals at high altitudes. Multifactor logistic regression was performed to determine associations and AF risk factors. SCN5A-H558R genotypes differed significantly between the AF and NAF groups (P < .0125) and the G allele was an independent AF risk factor (P < .05) at both altitudes, with no significant differences according to altitude (P > .0125). At middle altitudes, age, red blood cell distribution width (RDW-SD), left atrial internal diameter (LAD), and G allele were independent AF risk factors. At high altitudes, age, smoking, hypertension, RDW-SD, free triiodothyronine, LAD, and G allele were independent AF risk factors (P < .05). The G allele of SCN5A-H558R might be an independent risk factor of AF both high and middle altitude, but there are some differences in other clinical risk factors of AF.

Instance Segmentation Based on Improved Self-Adaptive Normalization.

The single batch normalization (BN) method is commonly used in the instance segmentation algorithms. The batch size is concerned with some drawbacks. A too small sample batch size leads to a sharp drop in accuracy, but a too large batch may result in the memory overflow of graphic processing units (GPU). These problems make BN not feasible to some instance segmentation tasks with inappropriate batch sizes. The self-adaptive normalization (SN) method, with an adaptive weight loss layer, shows good performance in instance segmentation algorithms, such as the YOLACT. However, the parameter averaging mechanism in the SN method is prone to problems in the weight learning and assignment process. In response to such a problem, the paper proposes to replace the single BN with an adaptive weight loss layer in SN models, based on which a weight learning method is developed. The proposed method increases the input feature expression ability of the subsequent layers. By building a Pytorch deep learning framework, the proposed method is validated in the MS-COCO data set and Autonomous Driving Cityscapes data set. The experimental results prove that the proposed method is effective in processing samples independent from the batch size. The stable accuracy for all kinds of target segmentation is achieved, and the overall loss value is significantly reduced at the same time. The convergence speed of the network is also improved.