Multiclass classification of motor imagery tasks based on multi-branch convolutional neural network and temporal convolutional network model.

Motor imagery (MI) is a cognitive process wherein an individual mentally rehearses a specific movement without physically executing it. Recently, MI-based brain-computer interface (BCI) has attracted widespread attention. However, accurate decoding of MI and understanding of neural mechanisms still face huge challenges. These seriously hinder the clinical application and development of BCI systems based on MI. Thus, it is very necessary to develop new methods to decode MI tasks. In this work, we propose a multi-branch convolutional neural network (MBCNN) with a temporal convolutional network (TCN), an end-to-end deep learning framework to decode multi-class MI tasks. We first used MBCNN to capture the MI electroencephalography signals information on temporal and spectral domains through different convolutional kernels. Then, we introduce TCN to extract more discriminative features. The within-subject cross-session strategy is used to validate the classification performance on the dataset of BCI Competition IV-2a. The results showed that we achieved 75.08% average accuracy for 4-class MI task classification, outperforming several state-of-the-art approaches. The proposed MBCNN-TCN-Net framework successfully captures discriminative features and decodes MI tasks effectively, improving the performance of MI-BCIs. Our findings could provide significant potential for improving the clinical application and development of MI-based BCI systems.

Mechanism Study of Molecular Trap in All-Organic Polystyrene-Based Dielectric Composite.

It is a huge challenge to explore how charge traps affect the electric breakdown of polymer-based dielectric composites. In this paper, alkane and aromatic molecules with different substituents are investigated according to DFT theoretical method. The combination of strong electron-withdrawing groups and aromatic rings can establish high electron affinity molecules. 4'-Nitro-4-dimethylaminoazobenzene (NAABZ) with a vertical electron affinity of 1.39 eV and a dipole moment of 10.15 D is introduced into polystyrene (PSt) to analyze the influence of charge traps on electric properties. Marcus charge transfer theory is applied to calculate the charge transfer rate between PSt and NAABZ. The nature of charge traps is elaborated from a dynamic perspective. The enhanced breakdown mechanism of polymers-based composites stems from the constraint of carrier mobility caused by the change in transfer rate. But the electrophile nature of high electron affinity filler can decrease the potential barriers at the metal-polymer interface. Simultaneously, the relationship between the electron affinity of fillers and the breakdown strength of polymer-based composites is nonlinear because of the presence of the inversion region. Based on the deep understanding of the molecular trap, this work provides the theoretical calculation for the design and development of high-performance polymer dielectrics.

Dynamic modulation of multicolor upconversion luminescence of Er3+ via excitation pulse width.

Lanthanide-doped upconversion (UC) luminescent materials display multicolor emissions, making them ideal for a variety of applications, such as multi-channel biological imaging, fluorescence encryption, anti-counterfeiting, and 3D display. Manipulating the UC emissions of the luminescent materials with a fixed composition is crucial for their applications. Herein, we propose a facile strategy to achieve pulse-width-dependent multicolor UC emissions in NaYF4:Yb/Er/Tm nanocrystals. Upon excitation with a 980 nm continuous-wave laser diode, Er3+ ions in NaYF4:20%Yb,15%Er,1%Tm nanocrystals exhibited UC emissions with a red-to-green (R/G) ratio of 11.3. Nevertheless, by employing a 980 nm pulse laser with pulse widths from 0.1 to 10 ms, the UC R/G ratio can be easily adjusted from 0.9 to 11.3, resulting in continuous and remarkable color transformation from green, yellow, orange, to red. By virtue of the dynamic luminescence color variation of these NaYF4:20%Yb,15%Er,1%Tm nanocrystals, we demonstrated their potential applications in the areas of anti-counterfeiting and information encryption. These findings provide deep insights into the excited-state dynamics and energy transfer of Er3+ in NaYF4:Yb/Er/Tm nanocrystals upon 980 nm pulse excitation, which may pave the way for designing multicolor UC materials toward versatile applications.

Layer-by-Layer-Processed All-Polymer Solar Cells with Enhanced Performance Enabled by Regulating the Microstructure of Upper Layer.

The layer-by-layer (LBL) fabrication method allows for controlled microstructure morphology and vertical component distribution, and also offers a reproducible and efficient technique for fabricating large-scale organic solar cells (OSCs). In this study, the polymers D18 and PYIT-OD are employed to fabricate all-polymer solar cells (all-PSCs) using the LBL method. Morphological studies reveal that the use of additives optimizes the microstructure of the active layer, enhancing the cells' crystallinity and charge transport capability. The optimized device with 2% CN additive significantly reduces bimolecular recombination and trap-assisted recombination. All-PSCs fabricated by the LBL method based on D18/PYIT-OD deliver a power conversion efficiency (PCE) of 15.07%. Our study demonstrates the great potential of additive engineering via the LBL fabrication method in regulating the microstructure of active layers, suppressing charge recombination, and enhancing the photovoltaic performance of devices.

Geometric differences in the ribosome exit tunnel impact the escape of small nascent proteins.

The exit tunnel is the subcompartment of the ribosome that contains the nascent polypeptide chain and, as such, is involved in various vital functions, including regulation of translation and protein folding. As the geometry of the tunnel shows important differences across species, we focus on key geometrical features of eukaryote and prokaryote tunnels. We used a simple coarse-grained molecular dynamics model to study the role of the tunnel geometry in the post-translational escape of short proteins (short open reading frames [sORFs]) with lengths ranging from 6 to 56 amino acids. We found that the probability of escape for prokaryotes is one for all but the 12-mer chains. Moreover, proteins of this length have an extremely low escape probability in eukaryotes. A detailed examination of the associated single trajectories and energy profiles showed that these variations can be explained by the interplay between the protein configurational space and the confinement effects introduced by the constriction sites of the ribosome exit tunnel. For certain lengths, either one or both of the constriction sites can lead to the trapping of the protein in the "pocket" regions preceding these sites. As the distribution of existing sORFs indicates some bias in length that is consistent with our findings, we finally suggest that the constraints imposed by the tunnel geometry have impacted the evolution of sORFs.

Evolution-guided multiomics provide insights into the strengthening of bioactive flavone biosynthesis in medicinal pummelo.

Pummelo (Citrus maxima or Citrus grandis) is a basic species and an important type for breeding in Citrus. Pummelo is used not only for fresh consumption but also for medicinal purposes. However, the molecular basis of medicinal traits is unclear. Here, compared with wild citrus species/Citrus-related genera, the content of 43 bioactive metabolites and their derivatives increased in the pummelo. Furthermore, we assembled the genome sequence of a variety for medicinal purposes with a long history, Citrus maxima 'Huazhouyou-tomentosa' (HZY-T), at the chromosome level with a genome size of 349.07 Mb. Comparative genomics showed that the expanded gene family in the pummelo genome was enriched in flavonoids-, terpenoid-, and phenylpropanoid biosynthesis. Using the metabolome and transcriptome of six developmental stages of HZY-T and Citrus maxima 'Huazhouyou-smooth' (HZY-S) fruit peel, we generated the regulatory networks of bioactive metabolites and their derivatives. We identified a novel MYB transcription factor, CmtMYB108, as an important regulator of flavone pathways. Both mutations and expression of CmtMYB108, which targets the genes PAL (phenylalanine ammonia-lyase) and FNS (flavone synthase), displayed differential expression between Citrus-related genera, wild citrus species and pummelo species. This study provides insights into the evolution-associated changes in bioactive metabolism during the origin process of pummelo.

Catalpol improves impaired neurovascular unit in ischemic stroke rats via enhancing VEGF-PI3K/AKT and VEGF-MEK1/2/ERK1/2 signaling.

Neurovascular unit (NVU) is organized multi-cellular and multi-component networks that are essential for brain health and brain homeostasis maintaining. Neurovascular unit dysfunction is the central pathogenesis process of ischemic stroke. Thus integrated protection of NVU holds great therapeutic potential for ischemic stroke. Catalpol, classified into the iridoid monosaccharide glycoside, is the main active ingredient of the radix from traditional Chinese medicine, Rehmannia glutinosa Libosch, that exhibits protective effects in several brain-related diseases. In the present study, we investigated whether catalpol exerted protective effects for NVU in ischemic stroke and the underlying mechanisms. MCAO rats were administered catalpol (2.5, 5.0, 10.0 mg·kg-1·d-1, i.v.) for 14 days. We showed that catalpol treatment dose-dependently reduced the infarction volume and significantly attenuated neurological deficits score in MCAO rats. Furthermore, catalpol treatment significantly ameliorated impaired NVU in ischemic region by protecting vessel-neuron-astrocyte structures and morphology, and promoting angiogenesis and neurogenesis to replenish lost vessels and neurons. Moreover, catalpol treatment significantly increased the expression of vascular endothelial growth factor (VEGF) through up-regulating PI3K/AKT signaling, followed by increasing FAK and Paxillin and activating PI3K/AKT and MEK1/2/ERK1/2 pathways. The protective mechanisms of catalpol were confirmed in an in vitro three-dimensional NVU model subjected to oxygen-glucose deprivation. In conclusion, catalpol protects NVU in ischemic region via activation of PI3K/AKT signaling and increased VEGF production; VEGF further enhances PI3K/AKT and MEK1/2/ERK1/2 signaling, which may trigger a partly feed-forward loop to protect NVU from ischemic stroke.

Wide-range optical fiber temperature sensor based on up-conversion luminescent nanocrystals.

In this study, an optical fiber temperature sensor based on up-conversion luminescence (UCL) is proposed. The core part is a new plan of the sensing unit, which is constructed with UCL materials of NaYF4: Yb3+, Er3+ nanocrystals by fiber fusion technology to achieve a wide range of temperature measurements. Experimental results show that the proposed optical fiber temperature sensor shows significant spectrum-temperature characteristics in 80-373 K temperature range. Its relative sensitivities at low temperature and normal temperature are 2.2×10-3/K with a determination coefficient of 0.957 and 9.1×10-3/K with a determination coefficient of 0.994, respectively. This type of sensor also has good mechanical strength and system stability and shows great potential for development, particularly in the aerospace field with large temperature differences.

PP121 suppresses RANKL-Induced osteoclast formation in vitro and LPS-Induced bone resorption in vivo.

Bone resorption, caused by osteoclasts (OCs), is important to bone homeostasis. The abnormalities of bone resorption may induce a series of diseases, including osteoarthritis, osteoporosis and aseptic peri-implant loosening. The latest research developed,a novel tyrosine and phosphoinositide kinase dual inhibitor, named PP121, inhibited Src in anaplastic thyroid carcinoma cell. However, the therapeutic function of PP121 on abnormal bone resorption is still uncertain. In the present study, we showed that PP121 could potently suppress osteoclast differentiation, osteoclast-specific gene expression and bone resorption via suppressing Src/MAPK (ERK and p38)/Akt-mediated NFATc1 induction in vitro. \It was found that PP121 could suppress the formation of osteoclasts from bone marrow macrophages (BMMs) without causing cytotoxicity, inhibit bone resorption and downregulate the mRNA level of osteoclast-specific markers, including calcitonin receptor (CTR), tartrate resistant acid phosphatase (TRAP), cathepsin K (CTSK), matrix metalloproteinase 3 (MMP3), Cellular oncogene fos (C-Fos) and nuclear factor of activated T-cells cytoplasmic 1 (NFATc1). Consistent with in vitro observation, we found that PP121 greatly ameliorated LPS-induced bone resorption. Our results provide promising evidence of the therapeutic potential of PP121 for osteolytic diseases related to excessive osteoclast-mediated bone resorption.

Effects of wheelchair Tai Chi ball exercise on physical and mental health and functional abilities among elderly with physical disability.

The purpose of this study was to examine the effect of a 12-week Wheelchair Tai Chi Ball (WTCB) intervention, a combination of mind-body exercise with strength training, on physical and mental health and functional abilities among elderly with disability. Twenty-six elderly persons participated in the study, nine WTCB group participants and ten control group participants completed the study. The WTCB group practised WTCB12 twice/week for one hour each time. The control group did their daily routine without WTCB intervention. The outcomes measures were: Pain Self-Efficacy Questionnaire (PSEQ), SF-36v2 for physical and mental health, heart rate, blood pressure, range of motion and muscle strength of the dominant arm at the shoulder, elbow and wrist joints. The Mixed Model ANOVA was employed to examine the differences between and within the two groups using pre-test and post-test scores. The results demonstrated the WTCB group had significant improvements on PSEQ, general physical health and had positive effects on maintaining muscle strength at the shoulder, elbow and wrist joints as compared to the control group. The WTCB12 exercise had positive effects on self-efficacy for pain management, general physical health, and maintain upper extremity muscle strength and is a feasible exercise for elderly with disability.

Chitooligosaccharide inhibits RANKL-induced osteoclastogenesis and ligation-induced periodontitis by suppressing MAPK/ c-fos/NFATC1 signaling.

Considering the high rate of osteoclast-related diseases worldwide, research targeting osteoclast formation/function is crucial. In vitro, we demonstrated that chitooligosaccharide (CS) dramatically inhibited osteoclastogenesis as well as osteoclast function dose-dependently. CS suppressed osteoclast-specific genes expression during osteoclastogenesis. Furthermore, we found that CS attenuated receptor activator of nuclear factor kappa B ligand (RANKL)-mediated mitogen-activated protein kinase (MAPK) pathway involving p38, erk1/2, and jnk, leading to the reduced expression of c-fos and nuclear factor of activated T cells c1 (NFATc1) during osteoclast differentiation. In vivo, we found CS protected rats from periodontitis-induced alveolar bone loss by micro-computerized tomography and histological analysis. Overall, CS inhibited RANKL-induced osteoclastogenesis and ligature-induced rat periodontitis model, probably by suppressing the MAPK/c-fos/NFATc1 signaling pathway. Therefore, CS may be a safe and promising treatment for osteoclast-related diseases.

Stabilizing γ-CsPbI3 Perovskite via Phenylethylammonium for Efficient Solar Cells with Open-Circuit Voltage over 1.3 V.

Cesium lead iodide (CsPbI3 ) perovskite has gained great attention due to its potential thermal stability and appropriate bandgap (≈1.73 eV) for tandem cells. However, the moisture-induced thermodynamically unstable phase and large open-circuit voltage (VOC ) deficit and also the low efficiency seriously limit its further development. Herein, long chain phenylethylammonium (PEA) is utilized into CsPbI3 perovskite to stabilize the orthorhombic black perovskite phase (γ-CsPbI3 ) under ambient condition. Furthermore, the moderate lead acetate (Pb(OAc)2 ) is controlled to combine with phenylethylammonium iodide to form the 2D perovskite, which can dramatically suppress the charge recombination in CsPbI3 . Unprecedentedly, the resulted CsPbI3 solar cells achieve a 17% power conversion efficiency with a record VOC of 1.33 V, the VOC deficit is only 0.38 V, which is close to those in organic-inorganic perovskite solar cells (PSCs). Meanwhile, the PEA modified device maintains 94% of its initial efficiency after exceeding 2000 h of storage in the low-humidity controlled environment without encapsulation.

Study on the characteristic of overburden rock structure and support system of the retraction channel in layered fully mechanized caving face.

Aiming at the technical problems of frequent dynamic pressure disturbance of regenerated roof and roadway stability control in the process of forming large-section support return channel under the delamination mining, taking 1200-2 fully mechanized caving face of Baiyinhua No. 4 Mine as the engineering background, numerical simulation and theoretical analysis were adopted. The overburden structure of the lower stratified caving face and the loading characteristics of the roof after the expansion are analyzed, and the breaking forms and the limiting conditions of the roof are given.The combined support scheme of anchor rod + metal mesh + steel ladder belt + I-beam insertion + suspension beam + anchor cable + single pillar is studied to form a set of safe and efficient construction support technology system. Practice shows that this scheme can effectively reduce the roof subsidence, narrow the scope of plastic failure zone, and ensure the stability of roadway surrounding rock during the support withdrawal period. Successfully complete the efficient withdrawal of hydraulic support. The research results can provide a good reference for the roof management and the smooth output of coal resources under similar conditions.

Large-Scale Cortical Network Analysis and Classification of MI-BCI Tasks Based on Bayesian Nonnegative Matrix Factorization.

Motor imagery (MI) is a high-level cognitive process that has been widely applied to clinical rehabilitation and brain-computer interfaces (BCIs). However, the decoding of MI tasks still faces challenges, and the neural mechanisms underlying its application are unclear, which seriously hinders the development of MI-based clinical applications and BCIs. Here, we combined EEG source reconstruction and Bayesian nonnegative matrix factorization (NMF) methods to construct large-scale cortical networks of left-hand and right-hand MI tasks. Compared to right-hand MI, the results showed that the significantly increased functional network connectivities (FNCs) mainly located among the visual network (VN), sensorimotor network (SMN), right temporal network, right central executive network, and right parietal network in the left-hand MI at the ß (13-30Hz) and all (8-30Hz) frequency bands. For the network properties analysis, we found that the clustering coefficient, global efficiency, and local efficiency were significantly increased and characteristic path length was significantly decreased in left-hand MI compared to right-hand MI at the ß and all frequency bands. These network pattern differences indicated that the left-hand MI may need more modulation of multiple large-scale networks (i.e., VN and SMN) mainly located in the right hemisphere. Finally, based on the spatial pattern network of FNC and network properties, we propose a classification model. The proposed model achieves a top classification accuracy of 78.2% in cross-subject two-class MI-BCI tasks. Overall, our findings provide new insights into the neural mechanisms of MI and a potential network biomarker to identify MI-BCI tasks.

Multifunctional Prosthesis Surface: Modification of Titanium with Cinnamaldehyde-Loaded Hierarchical Titanium Dioxide Nanotubes.

Orthopedic prostheses are the ultimate therapeutic solution for various end-stage orthopedic conditions. However, aseptic loosening and pyogenic infections remain as primary complications associated with these devices. In this study, a hierarchical titanium dioxide (TiO2) nanotube drug delivery system loaded with cinnamaldehyde for the surface modification of titanium implants, is constructed. These specially designed dual-layer TiO2 nanotubes enhance material reactivity and provide an extensive drug-loading platform within a short time. The introduction of cinnamaldehyde enhances the bone integration performance of the scaffold (simultaneously promoting bone formation and inhibiting bone resorption), anti-inflammatory capacity, and antibacterial properties. In vitro experiments have demonstrated that this system promoted osteogenesis by upregulating both Wnt/ß-catenin and MAPK signaling pathways. Furthermore, it inhibits osteoclast formation, suppresses macrophage-mediated inflammatory responses, and impedes the proliferation of Staphylococcus aureus and Escherichia coli. In vivo experiments shows that this material enhances bone integration in a rat model of femoral defects. In addition, it effectively enhances the antibacterial and anti-inflammatory properties in a subcutaneous implant in a rat model. This study provides a straightforward and highly effective surface modification strategy for orthopedic Ti implants.

Learning Gait Representation From Massive Unlabelled Walking Videos: A Benchmark.

Gait depicts individuals' unique and distinguishing walking patterns and has become one of the most promising biometric features for human identification. As a fine-grained recognition task, gait recognition is easily affected by many factors and usually requires a large amount of completely annotated data that is costly and insatiable. This paper proposes a large-scale self-supervised benchmark for gait recognition with contrastive learning, aiming to learn the general gait representation from massive unlabelled walking videos for practical applications via offering informative walking priors and diverse real-world variations. Specifically, we collect a large-scale unlabelled gait dataset GaitLU-1M consisting of 1.02M walking sequences and propose a conceptually simple yet empirically powerful baseline model GaitSSB. Experimentally, we evaluate the pre-trained model on four widely-used gait benchmarks, CASIA-B, OU-MVLP, GREW and Gait3D with or without transfer learning. The unsupervised results are comparable to or even better than the early model-based and GEI-based methods. After transfer learning, GaitSSB outperforms existing methods by a large margin in most cases, and also showcases the superior generalization capacity. Further experiments indicate that the pre-training can save about 50% and 80% annotation costs of GREW and Gait3D. Theoretically, we discuss the critical issues for gait-specific contrastive framework and present some insights for further study. As far as we know, GaitLU-1M is the first large-scale unlabelled gait dataset, and GaitSSB is the first method that achieves remarkable unsupervised results on the aforementioned benchmarks.

FGF12 regulates cell cycle gene expression and promotes follicular granulosa cell proliferation through ERK phosphorylation in geese.

The granulosa cells play an important role in the fate of follicular development or atresia in poultry. Fibroblast growth factor 12 (FGF12) is downregulated in atretic follicles and may be involved in regulating granulosa cell survival in previous studies, but its molecular mechanism remains unclear. In this study, FGF12 overexpression and knockdown models of goose granulosa cells were constructed to investigate its function. The downstream expression of the cell cycle pathway was analyzed by qPCR. Granulosa cell proliferative activity and apoptosis were detected by CCK8 and TUNEL. Protein phosphorylation levels of ERK and AKT were measured using Western blotting to analyze the key pathway of FGF12 regulation of granulosa cell proliferation. ERK protein phosphorylation inhibitor was added for further verification. After overexpression of FGF12, cell proliferation activity was increased, the expressions of cell cycle pathway genes CCND1, CCNA2, MAD2, and CHK1 were upregulated, the apoptosis of granulosa cell was decreased, and Caspase 3 gene and protein expression were downregulated. After the knockdown of FGF12, cell proliferation activity decreased, the expression of downstream genes in the cell cycle pathway was downregulated, the apoptosis of granulosa cells was increased, and the Bcl-2 gene and protein were downregulated. Overexpression of FGF12 promoted the synthesis of P4 and upregulates the expression of the STAR gene. Overexpression of FGF12 promoted ERK protein phosphorylation but did not affect AKT phosphorylation. The addition of ERK phosphorylation inhibitors resulted in the elimination of the increase in cell proliferative activity caused by FGF12 overexpression. In conclusion, FGF12 could promote proliferation and inhibit apoptosis of goose granulosa cells by increasing ERK phosphorylation.

BMP4/SMAD8 signaling pathway regulated granular cell proliferation to promote follicle development in Wanxi white goose.

Granular cells proliferation in goose regulated by bone morphogenetic proteins (BMPs) signaling pathway is still unknown. In this experiment, BMPs and their receptor, and receptor activated mothers against decapentaplegic homologs (SMADs) were quantitatively expressed in granular cell layer of pre-hierarchycal and hierarchycal follicles in Wanxi White goose. The screened BMP was then used for construction of overexpressed and knockdown vectors and transfected into granular cells of goose to assess the cell proliferation and apoptosis. Granular cells with BMP-overexpressed were then used for ChIP-Seq analysis to elucidate the molecular mechanism of BMP affecting granular cell proliferation. The results showed that the mRNA expression of BMP4 was significantly expressed in pre-hierarchical follicles, and also highly expressed in hierarchical follicles than other BMPs, while the Ⅰ and Ⅱ type of BMP receptors were expressed in basic level. The mRNA expression of SMAD8 was significantly elevated in pre-hierarchical follicles. Overexpression of BMP4 could promote the proliferation of granular cells and inhibited the expression of BMP4 caused a higher cell apoptosis. ChIP-Seq identified multiple regulatory targets of SMAD4, which were mostly related to cell cycle and lipid metabolism according to the GO and KEGG pathway enrichment. From the five most significant binding motif and quantitative expression verification, the activin membrane binding inhibitor (BAMBI) was down regulated in BMP4 overexpressed granular cells. In conclusion, the BMP4 was highly expressed in granular cells and phosphorylates SMAD8, the activated SMAD8 combined with SMAD4 transfers into nucleus to regulate the expression of BAMBI to promote lipid synthesis.

Social Networks Effects on Spouse and Adult-Child Dementia Caregivers' Experiences: A Cross-Sectional Study.

OBJECTIVES: A large body of literature addresses experiences of spouse and adult-children caregiver of individuals with dementia (IWDs) but has not examined the role and strength of social networks in associations between spouses and adult-children caregivers' experience. Based on the stress process model, we aimed to explore the strength levels of social networks and their association with spouses/adult-children caregivers for IWDs. DESIGN: A cross-sectional study. SETTING AND PARTICIPANTS: A questionnaire-based survey was conducted with a total of 146 family caregivers of IWDs (78 adult-child, and 68 spouses) in China. METHODS: Data collection comprised 4 sections: (1) care-related stressors (dementia stage, neuropsychiatric symptoms); (2) caregiver context; (3) social network, using the Lubben Social Network Scale; and (4) caregiving experience, using the short-form Zarit Burden Interview and 9-item Positive Aspects of Caregiving Scale. Linear regression, mediation model analysis, and interactive analysis were performed to explore the mechanisms of associations between variables. RESULTS: Spouses had weaker social network strength (ß = -0.294, P = .001) and reported greater positive aspects of caregiving (ß = 0.234, P = .003) than adult-children caregivers; no significant difference was found between them for caregiver burden. Mediation analysis suggests that associations between caregiver type and caregiver burden are indirect-only mediation effects of social networks (ß = 0.140, 95% CI = 0.066-0.228). The social network strength suppressed the association between caregiver type and positive aspects of caregiving. The caregiver type/social network interaction statistically significantly (P = .025) affected the "positive aspects": a stronger social network was associated with more positive aspects of caregiving among the spouse subgroup (ß = 0.341, P = .003). CONCLUSIONS AND IMPLICATIONS: Social networks mediate responses to caregiving experiences among different care provider types and are vital intervention targets, especially for spousal caregivers. Our results can serve as references for identifying caregivers for clinical intervention.

A Model Combining Multi Branch Spectral-Temporal CNN, Efficient Channel Attention, and LightGBM For MI-BCI Classification.

Accurately decoding motor imagery (MI) brain-computer interface (BCI) tasks has remained a challenge for both neuroscience research and clinical diagnosis. Unfortunately, less subject information and low signal-to-noise ratio of MI electroencephalography (EEG) signals make it difficult to decode the movement intentions of users. In this study, we proposed an end-to-end deep learning model, a multi-branch spectral-temporal convolutional neural network with channel attention and LightGBM model (MBSTCNN-ECA-LightGBM), to decode MI-EEG tasks. We first constructed a multi branch CNN module to learn spectral-temporal domain features. Subsequently, we added an efficient channel attention mechanism module to obtain more discriminative features. Finally, LightGBM was applied to decode the MI multi-classification tasks. The within-subject cross-session training strategy was used to validate classification results. The experimental results showed that the model achieved an average accuracy of 86% on the two-class MI-BCI data and an average accuracy of 74% on the four-class MI-BCI data, which outperformed current state-of-the-art methods. The proposed MBSTCNN-ECA-LightGBM can efficiently decode the spectral and temporal domain information of EEG, improving the performance of MI-based BCIs.