An EMD-Based Adaptive Client Selection Algorithm for Federated Learning in Heterogeneous Data Scenarios.

Federated learning is a distributed machine learning framework that enables distributed nodes with computation and storage capabilities to train a global model while keeping distributed-stored data locally. This process can promote the efficiency of modeling while preserving data privacy. Therefore, federated learning can be widely applied in distributed conjoint analysis scenarios, such as smart plant protection systems, in which widely networked IoT devices are used to monitor the critical data of plant production to improve crop production. However, the data collected by different IoT devices can be dependent and identically distributed (non-IID), causing the challenge of statistical heterogeneity. Studies have also shown that statistical heterogeneity can lead to a marked decline in the efficiency of federated learning, making it challenging to apply in practice. To promote the efficiency of federated learning in statistical heterogeneity scenarios, an adaptive client selection algorithm for federated learning in statistical heterogeneous scenarios called ACSFed is proposed in this paper. ACSFed can dynamically calculate the possibility of clients being selected to train the model for each communication round based on their local statistical heterogeneity and previous training performance instead of randomly selected clients, and clients with heavier statistical heterogeneity or bad training performance would be more likely selected to participate in the later training. This client selection strategy can enable the federated model to learn the global statistical knowledge faster and thereby promote the convergence of the federated model. Multiple experiments on public benchmark datasets demonstrate these improvements in the efficiency of the models in heterogeneous settings.

The interaction of chitosan and BMP-2 tuned by deacetylation degree and pH value.

Integrating proteins with chitosan (CS) can not only improve the biocompatibility of CS, but also enable the sustained release of proteins. The deacetylation degree (DD) and pH conditions are considered as the two most important factors affecting the interaction between CS and proteins. In this study, the effects of CS with different DDs and pH conditions on bone morphogenetic protein-2 (BMP-2) adsorption were evaluated using the molecular dynamics (MD) method. The results showed that the interaction energy first increased and then decreased, with the increase of the DD in the range of 0-100%; the strongest interaction was found between BMP-2 and 75% DD-CS surface. Furthermore, decreasing pH values favor the interaction between BMP-2 and CS surfaces. These results are useful to gain a better understanding of the interaction mechanism between polysaccharides and proteins and provide a meaningful reference for the application of CS with different DDs and pH conditions. © 2018 Wiley Periodicals, Inc. J Biomed Mater Res Part A: 107A: 769-779, 2019.

Effects of atomic-level nano-structured hydroxyapatite on adsorption of bone morphogenetic protein-7 and its derived peptide by computer simulation.

Hydroxyapatite (HA) is the principal inorganic component of bones and teeth and has been widely used as a bone repair material because of its good biocompatibility and bioactivity. Understanding the interactions between proteins and HA is crucial for designing biomaterials for bone regeneration. In this study, we evaluated the effects of atomic-level nano-structured HA (110) surfaces on the adsorption of bone morphogenetic protein-7 (BMP-7) and its derived peptide (KQLNALSVLYFDD) using molecular dynamics and density functional theory methods. The results indicated that the atomic-level morphology of HA significantly affected the interaction strength between proteins and HA substrates. The interactions of BMP-7 and its derived peptide with nano-concave and nano-pillar HA surfaces were stronger than those with flat or nano-groove HA surfaces. The results also revealed that if the groove size of nano-structured HA surfaces matched that of residues in the protein or peptide, these residues were likely to spread into the grooves of the nano-groove, nano-concave, and nano-pillar HA, further strengthening the interactions. These results are helpful in better understanding the adsorption behaviors of proteins onto nano-structured HA surfaces, and provide theoretical guidance for designing novel bioceramic materials for bone regeneration and tissue engineering.

Interaction Behaviors of Fibrinopeptide-A and Graphene with Different Functional Groups: A Molecular Dynamics Simulation Approach.

Graphene as a 2-dimentional material has been widely used in the field of biomedical applications. In this study, molecular dynamics simulations are carried out on the fibrinopeptide-A and graphene surfaces with N and O modifications. A new set of parameters for the CHARMM force field are developed to describe the behaviors of the surfaces. Our results indicate that the existence of most oxygen and nitrogen groups may enhance the interaction between the surfaces and the peptide, whereas the substitutional nitrogen on the graphene surface does not make a big difference. The improvement of interaction is not only because of the functional group on the surface, but also the defective morphology. The defective morphology also clears away the surface water layer. Our results suggest that the interactions between graphene biomolecules can be affected by functionalizing the surface with different types of functional groups, which is in accordance with the theory of material design.

From moderately severe to severe hypertriglyceridemia induced acute pancreatitis: circulating miRNAs play role as potential biomarkers.

The incidence of hypertriglyceridemia induced acute pancreatitis (HTAP) continues to rise in China. It has systemic complications and high mortality, making the early assessment of the severity of this disease even more important. Circulating microRNAs (miRNAs) could be novel, non-invasive biomarkers for disease progression judgment. This study aimed to identify the potential role of serum miRNAs as novel biomarkers of HTAP progression. HTAP patients were divided into two groups: moderately severe (HTMSAP) and severe (HTSAP), healthy people were used as control group. The serum miRNA expression profiles of these three groups were determined by microarray and verified by qRT-PCR. The functions and pathways of the targeted genes of deregulated miRNAs were predicted, using bioinformatics analysis; miRNA-mRNA network was generated. Moreover, the correlation between miR-181a-5p and pancreatitis metabolism related substances were studied and the serum concentration of inflammatory cytokines and miRNAs at different time points during the MSAP and SAP were investigated, respectively. Finally, the receiver operating characteristic (ROC) curve of miRNAs was studied. Significant changes in the serum concentration of the following miRNAs of HTAP patients (P<0.05) were discovered: miR24-3p, 361-5p, 1246, and 222-3p (constantly upregulated), and 181a-5p (constantly downregulated) (P<0.05). Bioinformatics analysis predicted that 13 GOs and 36 pathways regulated by overlap miRNAs were involved in glucose, fat, calcium (Ca++), and insulin metabolism (P<0.001). miRNA-mRNA network revealed that the overlap miRNAs targeted genes participating in pancreas metabolism and miR-181a-5p, the only downregulated miRNA, had good negative correlation with triglyceride (TG), total cholesterol (TC), and fast blood glucose (FBG), but a positive correlation with Ca++. When compared with inflammatory cytokines, the changes of all five overlap miRNAs were more stable. It was found that when used for evaluating the progression of HTAP, miRNAs showed good AUC. These data suggested that serum miRNAs have the potential to be excellent HTAP biomarkers.

Dynamic functional and ultrastructural changes of gastric parietal cells induced by water immersion-restraint stress in rats.

AIM: To investigate the dynamic functional and ultrastructural changes of gastric parietal cells induced by water immersion-restraint stress (WRS) in rats. METHODS: WRS model of Sprague-Dawley (SD) rats was established. Fifty-six male SD rats were randomly divided into control group, stress group and post-stress group. The stress group was divided into 1, 2 and 4 h stress subgroups. The post-stress group was divided into 24, 48 and 72 h subgroups. The pH value of gastric juice, ulcer index (UI) of gastric mucosa and H(+), K(+)-ATPase activity of gastric parietal cells were measured. Ultrastructural change of parietal cells was observed under transmission electron microscope (TEM). RESULTS: The pH value of gastric juice decreased time-dependently in stress group and increased in post-stress group. The H(+), K(+)-ATPase activity of gastric parietal cells and the UI of gastric mucosa increased time-dependently in stress group and decreased in post-stress group. Compared to control group, the pH value decreased remarkably (P = 0.0001), the UI and H(+), K(+)-ATPase activity increased significantly (P = 0.0001, P = 0.0174) in 4 h stress subgroup. UI was positively related with stress time (r = 0.9876, P < 0.01) but negatively with pH value (r = -0.8724, P < 0.05). The parietal cells became active in stress group, especially in 4 h stress subgroup, in which plenty of intracellular canalicular and mitochondria were observed under TEM. In post-stress group, the parietal cells recovered to resting state. CONCLUSION: The acid secretion of parietal cells is consistent with their ultrastructural changes during the development and healing of stress ulcer induced by WRS and the degree of gastric mucosal lesions, suggesting gastric acid play an important role in the development of stress ulcer and is closely related with the recovery of gastric mucosal lesions induced by WRS.

[Protective effect of nitric oxide on gastric mucosa and its relationship to the acid secretion of gastric parietal cells under stress in rats].

AIM: To demonstrate the protective effect of nitric oxide (NO) on gastric mucosa and its relationship to the acid secretion of parietal cells under stress in rats. METHODS: Water immersion-restraint stress (WRS) model in SD rats was performed. The gastric mucosal ulcer index (UI), NO contents in gastric mucosa and H+, K(+) -ATPase activity of parietal cells were measured. The effects of N(G)-nitro-L-arginine methyl ester(L-NAME) and L-arginine (L-Arg) on the H+, K(+)-ATPase activity of parietal cells and stress-induced gastric mucosal lesion were observed. RESULTS: L-NAME pretreatment decreased NO contents in gastric mucosa, activated H+, K(+) -ATPase activity of parietal cells and aggravated gastric mucosal lesion, whereas L-Arg pretreatment increased NO contents, inhibited H+, K(+) -ATPase activity and significantly ameliorated stress-induced gastric mucosal lesion. CONCLUSION: Endogenous nitric oxide plays an important role in protecting gastric mucosa from stress-induced lesion by inhibiting H+, K(+) -ATPase activity of parietal cells.