Disentangled self-attention neural network based on information sharing for click-through rate prediction.

With the exponential growth of network resources, recommendation systems have become successful at combating information overload. In intelligent recommendation systems, the prediction of click-through rates (CTR) plays a crucial role. Most CTR models employ a parallel network architecture to successfully capture explicit and implicit feature interactions. However, the existing models ignore two aspects. One limitation observed in most models is that they focus only on the interaction of paired term features, with no emphasis on modeling unary terms. The second issue is that most models input characteristics indiscriminately into parallel networks, resulting in network input oversharing. We propose a disentangled self-attention neural network based on information sharing (DSAN) for CTR prediction to simulate complex feature interactions. Firstly, an embedding layer transforms high-dimensional sparse features into low-dimensional dense matrices. Then, the disentangled multi-head self-attention learns the relationship between different features and is fed into a parallel network architecture. Finally, we set up a shared interaction layer to solve the problem of insufficient information sharing in parallel networks. Results from experiments conducted on two real-world datasets demonstrate that our proposed method surpasses existing methods in predictive accuracy.

Umbilical cord mesenchymal stem cells promote the repair of trochlear groove reconstruction in dogs.

Trochlear groove reconstruction (TGR) is a common treatment for patellar luxation (PL) in dogs. Nevertheless, the prognosis of TGR is poor due to the cartilage damage and secondary inflammation. To study the repair effect of canine umbilical cord mesenchymal stem cells (UC-MSCs) after TGR, 10 experimental dogs were given TGR surgery and then randomized into two groups: Treatment group (1 ml suspension allogeneic UC-MSCs (106 cells/kg) was injected into the cavum articulare on days 0, 7, and 14 after TGR); and the Model group (injected with 1 ml of physiological saline as negative control). The therapeutic effect of UC-MSCs was studied by blood routine examination, inflammatory factor index detection, double-blind knee score, histopathology, and computed tomography (CT) scans. The results showed that the total number of white blood cells and neutrophils in the model group were significantly higher than those in the treatment group on both 7 days and 21 days, postoperatively (P < 0.05); there were no significant changes in the levels of IL-6, MMP-13, and TGF-ß1 between the model group and the treatment group throughout the days of testing. The double-blind knee scores of the treatment group were significantly lower than the model group on 1st, 4th, and 5th days postoperatively (P < 0.05). The treatment group showed low-pain sensation, stable gait, and fast recovery of muscle strength in the knee score, and the wound healing of the treatment group returned to normal on the 5th day after surgery; CT scans and gross observation showed that the cartilage growth in the treatment group was faster than that in the model group. Histological observation of cases showed that fibro chondrocytes were predominantly found in the treatment group, and the distribution of chondrocytes was uneven, while the model group showed a large number of fibrous tissue hyperplasia, fissures, and unequal matrix staining. Intra-articular injection of UC-MSCs after TGR has the effect of relieving pain and promoting the repair of bone defects, making the operative limb recover function earlier, making up for the deficiency of TGR, and improving the effect of PL treatment. Future studies should furthermore explore the dose and frequency of therapy based on the multiple advantages of UC-MSCs and the mechanism of cartilage repair in dogs.

A Comparative Study of Biological Characteristics and Transcriptome Profiles of Mesenchymal Stem Cells from Different Canine Tissues.

Mesenchymal stem cells (MSCs) are the most promising seed cells for cell therapy. Comparing the biological and transcriptome gene characteristics of MSCs from different sources provides an important basis for the screening of clinically used cells. The main purpose of this experiment was to establish methods for the isolation and culture of MSCs from five different canine sources, including adipose tissue, bone marrow, umbilical cord, amniotic membrane, and placenta, and compare biological and transcriptome characteristics of MSCs, in order to provide a basis for the clinical application of canine MSCs. MSCs were isolated from Chinese pastoral dogs, and the following experiments were performed: (1) the third, sixth, and ninth generations of cells were counted, respectively, and a growth curve was plotted to calculate the MSC population doubling time; (2) the expression of CD34 and CD44 surface markers was studied by immunofluorescence; (3) the third generation of cells were used for osteogenetic and adipogenic differentiation experiments; and (4) MSC transcriptome profiles were performed using RNA sequencing. All of the five types of MSCs showed fibroblast-like adherent growth. The cell surface expressed CD44 instead of CD34; the third-generation MSCs had the highest proliferative activity. The average population doubling time of adipose mesenchymal stem cells (AD-MSCs), placenta mesenchymal stem cells (P-MSCs), bone marrow mesenchymal stem cells (BM-MSCs), umbilical cord mesenchymal stem cells (UC-MSCs), and amniotic mesenchymal stem cells (AM-MSCs) were 15.8 h, 21.2 h, 26.2 h, 35 h, and 41.9 h, respectively. All five types of MSCs could be induced to differentiate into adipocytes and osteoblasts in vitro, with lipid droplets appearing after 8 days and bone formation occurring 5 days after AD-MSC induction. However, the multilineage differentiation for the remaining of MSCs was longer compared to that of the AD-MSCs. The MSC transcriptome profiles showed that AD-MSC and BM-MSCs had the highest homology, while P-MSCs were significantly different compared to the other four types of MSCs. All the isolated MSCs had the main biological characteristics of MSCs. AD-MSCs had the shortest time for proliferation, adipogenesis, and osteogenic differentiation.

Evaluation of the Curative Effect of Umbilical Cord Mesenchymal Stem Cell Therapy for Knee Arthritis in Dogs Using Imaging Technology.

OBJECTIVE: The aim of this study was to assess the efficacy of canine umbilical cord mesenchymal stem cells (UC-MSCs) on the treatment of knee osteoarthritis in dogs. METHODS: Eight dogs were evenly assigned to two groups. The canine model of knee osteoarthritis was established by surgical manipulation of knee articular cartilage on these eight dogs. UC-MSCs were isolated from umbilical cord Wharton's jelly by 0.1% type collagenase I and identified by immunofluorescence staining and adipogenic and osteogenic differentiation in vitro. A suspension of allogeneic UC-MSCs (1 × 106) and an equal amount of physiological saline was injected into the cavitas articularis in the treated and untreated control groups, respectively, on days 1 and 3 posttreatment. The structure of the canine knee joint was observed by magnetic resonance imaging (MRI), B-mode ultrasonography, and X-ray imaging at the 3rd, 7th, 14th, and 28th days after treatment. Concurrently, the levels of IL-6, IL-7, and TNF-α in the blood of the examined dogs were measured. Moreover, the recovery of cartilage and patella surface in the treated group and untreated group was compared using a scanning electron microscope (SEM) after a 35-day treatment. RESULTS: Results revealed that the isolated cells were UC-MSCs, because they were positive for CD44 and negative for CD34 surface markers, and the cells were differentiated into adipocytes and osteoblasts. Imaging technology showed that as treatment time increased, the high signal in the MRI T2-weighted images decreased, the echo-free space in B ultrasonography images disappeared basically, and the continuous linear hypoechoic region at the trochlear sulcus thickened. On X-ray images, the serrate defect at the ventral cortex of the patella improved, and the low-density gap of the ventral patella and trochlear crest gradually increased in the treated group. On the contrary, the high signal in the MRI T2-weighted images and the echo-free space in B ultrasonography images still increased after a 14-day treatment in the untreated control group, and the linear hypoechoic region was discontinuous. On the X-ray images, there was no improvement in the serrate defect of the ventral cortex of the patella. Results for inflammatory factors showed that the blood levels of IL-6, IL-7, and TNF-α of the untreated control group were significantly higher than those of the treated group (P < 0.05) 7-14 days posttreatment. The result of SEM showed that the cartilage neogenesis in the treated group had visible neonatal tissue and more irregular arrangement of new tissue fibers than that of the untreated control group. Furthermore, more vacuoles but without collagen fibers were observed in the cartilage of the untreated control group, and the thickness of the neogenetic cartilage in the treated group (65.13 ± 5.29, 65.30 ± 5.83) and the untreated control group (34.27 ± 5.42) showed a significant difference (P < 0.01). CONCLUSION: Significantly higher improvement in cartilage neogenesis and recovery was observed in the treated group compared to the untreated control group. The joint fluid and the inflammatory response in the treated group decreased. Moreover, improved recovery in the neogenetic cartilage, damaged skin fascia, and muscle tissue around the joints was more significant in the treated group than in the untreated control group. In conclusion, canine UC-MSCs promote the repair of cartilage and patella injury in osteoarthritis, improve the healing of the surrounding tissues, and reduce the inflammatory response.

Trilobatin attenuates the LPS-mediated inflammatory response by suppressing the NF-κB signaling pathway.

We investigated the anti-inflammatory effect of trilobatin, the flavonoid isolated from the leaves of Lithocarpus polystachyus Rehd, as well as the underlying molecular mechanisms. Treatment with trilobatin (0.005-5 µM) dose-dependently inhibited the lipopolysaccharide (LPS)-induced mRNA expression and secretion of pro-inflammatory cytokines, including tumor necrosis factor α (TNFα), interleukin-1ß (IL-1ß) and interleukin-6 (IL-6), in RAW 264.7 macrophages. However, no further inhibition was detected when the concentration of trilobatin was increased to 50 µM. Western blot analysis confirmed that the mechanism of the anti-inflammatory effect was correlated with the inhibition of LPS-induced inhibitor of nuclear factor-kappa B α (IκBα) degradation and nuclear factor-kappa B (NF-κB) p65 phosphorylation. In addition, trilobatin also showed a significant inhibition of LPS-induced TNFα and IL-6 at both the mRNA and protein levels in a mouse model. Our results suggest that trilobatin potentially inhibits the LPS-induced inflammatory response by suppressing the NF-κB signaling pathway.

MgFe-layered double hydroxide modified electrodes for direct electron transfer of heme proteins.

In this study, the Fe-based layered double hydroxides (Mg(3)Fe LDH) were used to immobilize heme proteins including hemoglobin (Hb), myoglobin (Mb) and horseradish peroxidase (HRP) for fabrication of heme/Mg(3)Fe LDH film on glassy carbon electrode (Mg(3)Fe-heme/GCE). The possible role of iron in framework of LDH to promote direct electron transfer (DET) of heme proteins was investigated using an LDH containing non-iron as a reference. Hb was selected as a model protein for studying the electrocatalytic activity of immobilized heme in LDH film. The Mg(3)Fe-Hb/GCE displayed an enhanced electrocatalytic reduction towards H(2)O(2). The biosensor showed a very low detection limit (0.036 µM) and apparent Michaelis-Menten constant (7.98 µM). This work outlines that Fe-based LDH modified electrode provides a promising platform for immobilization of heme proteins and development of sensitive biosensors.