Recently, deep learning-based underwater object detection technology has achieved remarkable success. However, the accuracy and completeness of dataset instance annotation are crucial for its success. The quality of underwater images is low, severe objects clustering, and occlusion, acquiring object's annotations demands substantial time and labor costs, while mis annotation and missed annotation can also degrade model performance and limit their application in practical scenarios. To address this issue, this paper presents a novel weakly supervised underwater object real-time detection method, which is divided into two subtasks: weakly supervised object localization and real-time object detection. In the weakly supervised object localization task, we design a novel category hierarchy structure network that integrates the high-resolution attention-class activation mapping algorithm to obtain high-quality object class activation maps, weaken background interference, and obtain more complete object regions. The parameterized spatial loss module is devised to enable the model to escape from local optimal solutions, thus accurately and efficiently obtaining object pseudo-detection annotation boxes. For the real-time object detection task, the single-stage detector YOLOv7 is selected as the basic detection model, and an object perception loss function is designed based on the class activation map to jointly supervise the training process. A method for filtering noisy pseudo-supervision information is proposed to enhance the pseudo-supervision information involved in training. Ablation experiments and multi-method comparison experiments were conducted on the URPC and RUOD datasets, and the results verify the effectiveness of the proposed strategy, and our model exhibits significant advantages in detection performance and detection efficiency compared to current mainstream and advanced models.

Electroencephalogram (EEG) signals are critical for human-computer interaction due to their objective, continuous, and tamper-resistant emotional data. However, cross-subject/session emotion recognition faces challenges from EEG's small amplitude, low signal-to-noise ratio, and non-stationarity. To address this, we propose the Multi-Branch EEG Emotion Recognition (MBEER) model, integrating spatiotemporal and frequency information. The MBEER model comprises four modules: Brain-Area Global Dynamic Graph Attention Module (BG-DGAT), Channel Frequency Feature Extraction Module (CFFE), Hierarchical Flash Cross-Attention Module (HFCAT), and Multi-Branch Classifier (MBC). BG-DGAT dynamically models functional connectivity between brain regions using Graph Attention Networks (GAT), capturing non-stationary EEG characteristics and suppressing noise with multi-scale temporal convolution and Savitzky-Golay filtering. The CFFE module employs Transformer and Bi-LSTM architectures to extract time-frequency features from eight critical brain regions, highlighting symmetric hemispheric differences. HFCAT fuses spatial-temporal features via multi-head cross-attention to model inter-regional dynamics. The MBC module separates domain-specific and invariant features, aligning distributions across domains using Maximum Mean Discrepancy (MMD) loss for robust generalization. Experimental results demonstrate that MBEER achieves cross-subject accuracies of 92.10 % on SEED, 83.57 % on SEED-IV and 59.59 % on DEAP-Arousal. In cross-session experiments, MBEER achieves accuracies of 92.70 % (SEED) and 65.38 % (SEED-IV). These results also highlight performance variability across datasets, with comparatively lower accuracy on DEAP reflecting the dataset's higher complexity and noise. The ablation experiment further shows that the multi-view feature fusion strategy contributes to a 7.06 % performance gain. Extensive evaluations underline the robustness and superior emotion recognition accuracy of the proposed method under substantial inter-individual variability.

Purpose: Sepsis-associated acute lung injury (SALI) represents a severe complication in sepsis patients, leading to poor clinical outcomes and increased mortality. This study aimed to develop and validate a reliable nomogram for early prediction of SALI in adult critically ill patients, addressing the critical need for timely risk stratification and intervention. Methods: A retrospective cohort study was conducted involving 345 intensive care unit sepsis patients from the First People's Hospital of Huzhou. Patients were randomly divided into training (n = 241) and validation (n = 104) cohorts. Multivariate logistic regression and LASSO regression were employed to identify independent risk factors and construct a predictive nomogram. Results: Four independent risk factors were identified: PaCO2, PaO2, serum uric acid, and Sequential Organ Failure Assessment score. The developed nomogram demonstrated excellent discriminative performance, with area under the receiver operating characteristic curve of 0.916 in the training cohort and 0.931 in the validation cohort. Calibration curves, decision curve, and SHapley Additive exPlanations analysis confirmed the model's robust predictive performance and clinical utility. Conclusions: The novel nomogram provides a practical, visualized risk assessment tool for early SALI recognition, potentially improving patient outcomes through enhanced understanding and timely interventions in critically ill patients.

To solve the problems that the conventional object detector is hard to extract features and miss detec-tion of small objects when detecting underwater objects due to the noise of underwater environment and the scale change of objects, this paper designs a novel feature enhancement & progressive dynamic aggregation strategy, and proposes a new underwater object detector based on YOLOv5s. Firstly, a fea-ture enhancement gating module is designed to selectively suppress or enhance multi-level features and reduce the interference of underwater complex environment noise on feature fusion. Then, the adjacent feature fusion mechanism and dynamic fusion module are designed to dynamically learn fusion weights and perform multi-level feature fusion progressively, so as to suppress the conflict information in multi -scale feature fusion and prevent small objects from being submerged by the conflict information. At last, a spatial pyramid pool structure (FMSPP) based on the same size quickly mixed pool layer is proposed, which can make the network obtain stronger description ability of texture and contour features, reduce the parameters, and further improve the generalization ability and classification accuracy. The ablation experiments and multi-method comparison experiments on URPC and DUT-USEG data sets prove the effectiveness of the proposed strategy. Compared with the current mainstream detectors, our detector achieves obvious advantages in detection performance and efficiency.(c) 2023 Published by Elsevier Ltd.

Geometric Interpretation of Theorem 1. For two points xi and xj with their neighborhood matrices Xis and Xjs, if there exists a rotation (or reflection) matrix R satisfying with Theorem 1, then we can conclude the geometric structures consisting of two points and their neighbors are similar. For example, given the geometric design, the orange region (red point with its neighbors shown in an auxiliary Fig. 1 [25]) is similar to the blue region (blue points with their corresponding neighbors)* The advantage of finding similar points in the geometric space is that more information can be collected especially when point clouds are sparse. As for the comparison shown in Fig. 2, the traditional feature extraction with

To the Editor: Despite major advances in medical care, the incidence and mortality of bloodstream infection (BSI) remain high, which is still a global public health challenge. BSI can be caused by various microorganisms, and the most common organisms are Escherichia coli, Klebsiella pneumoniae, and Staphylococcus aureus (S. aureus), according to the China Antimicrobial Surveillance Network (CHINET).[1]S. aureus is the third most common cause of BSI, which is associated with short-term mortality rates of 15–30%, long-term excess mortality, and increased healthcare costs. At present, there are many studies on Gram-negative bacteremia but relatively few on S. aureus bloodstream infection (SA-BSI), especially in China. With the occurrence of new treatments and clinical conditions, such as aging and extensive antibiotic resistance, it is necessary to reanalyze the clinical characteristics and prognosis of SA-BSI. This single-center retrospective cohort study was conducted in the Second Affiliated Hospital, Zhejiang University School of Medicine. The Ethics Committee of our hospital granted ethics approval (No. 2019-194) for the present study. The requirement for signed informed consent was exempted because of the retrospective nature of the study. Furthermore, a statement of permission from patients for submission was not needed, as no personal information was included. During the six-year study period (2013–2018), a total of 1174 blood culture specimens positive for S. aureus were initially included. Four patients aged <18 years, 54 patients with nonpathogenic bacteria, 45 patients with incomplete or missing data, and nine patients lost to follow-up were excluded. Finally, 349 patients were included. The patients were followed for at least 28 days after the onset of BSI. According to 28-day mortality, the patients were divided into nonsurvival (61 cases) and survival groups (288 cases) [Figure 1A].Figure 1: (A) Flowchart of study participant enrollment with blood culture specimens positive for Staphylococcus aureus. (B) Age and sex distributions of patients with SA-BSI. (C) Distribution proportions of MRSA and MSSA from 2013 to 2018. (D) Forest plot of the results of the univariate and multivariate logistic regression analyses for prognostic factors for death in patients with SA-BSI. APACHE: Acute Physiology and Chronic Health Evaluation; COPD: Chronic obstructive pulmonary disorder; CVC: Central venous catheter; ICU: Intensive care unit; MRSA: Methicillin-resistant Staphylococcus aureus; MSSA: Methicillin-sensitive Staphylococcus aureus; SA-BSI: Staphylococcus aureus bloodstream infections; SOFA: Sequential Organ Failure Assessment.Using a self-created Excel sheet, the clinical data of cases were collected through the hospital identification of blood culture-positive specimens for S. aureus provided by the microbiology laboratory. We recorded demographic and microbiological data, sensitivity to antibiotics, and clinical treatment and evaluation data (the Sequential Organ Failure Assessment [SOFA] score, the Acute Physiology and Chronic Health Evaluation [APACHE] II score in the first 24 h following the onset of BSI) by reviewing electronic medical records. To ensure data quality, a homogeneous data collection form was carefully prepared before data collection. Additionally, two days of training were provided to data collectors and supervisors. Furthermore, the supervisor and principal investigators supervised the data collectors throughout the entire data collection period. After data collection was completed, the supervisor organized and carefully cleaned the data and then entered the data into Statistical Package for Social Sciences (SPSS, 26.0, IBM Corp., Armonk, NY, USA) for statistical analysis. Variables with P <0.05 in the univariate analysis were entered into the multivariable model. Continuous variables were treated as dichotomous variables based on Youden index. Multivariate analysis was performed with logistic regression to identify independent prognostic factors for death in patients with SA-BSI. A two-tailed P <0.05 was considered statistically significant. The demographics showed that the median age was 59.0 (45.5–68.0) years, and 69.6% (243/349) were male [Supplementary Table 1, https://links.lww.com/CM9/B557]. The age distribution was left-skewed, with a peak incidence in the group of 60–69 years. In addition, the proportion of men was significantly higher than that of women in all age groups except for those >90 years old [Figure 1B]. Patients in the nonsurvivor group were significantly older than survivor group (median, 65.0 vs. 58.0 years, P = 0.005). In terms of comorbidities, significantly higher percentages of chronic obstructive pulmonary disease (COPD) and severe asthma were observed in the nonsurvivor group than in the survivor group (6.6% [4/61] vs. 1.0% [3/288], P = 0.005). In comparison with survivors, nonsurvivors had a more severe condition, evidenced by a higher APACHE II score (median, 24 vs. 11, P <0.001), a higher SOFA score (median, 12 vs. 3, P <0.001), and had a higher rate of intensive care unit (ICU) admission (85.2% [52/61] vs. 27.4% [79/288], P <0.001), indwelling central venous catheter placement (73.8% [45/61] vs. 42.4% [122/288], P <0.001), and invasive mechanical ventilation (82.0% [50/61] vs. 34.4% [99/288], P <0.001). Compared with those in survivors, the proportions of polymicrobial SA-BSI (26.2% [16/61] vs. 13.2% [38/288], P = 0.011), methicillin-resistant S. aureus (MRSA) (82.0% [50/61] vs. 57.3% [165/288], P <0.001), and septic shock (50.8% [31/61] vs. 1.4% [4/288], P <0.001) in nonsurvivors were higher. In comparison with survivors, nonsurvivors had a lower hematocrit (median [%], 26.8 vs. 29.1, P = 0.009), a lower platelet count (median [109/L], 85 vs. 187, P <0.001), a lower albumin level (mean [g/L], 27.73 vs. 29.95, P =0.005), and worse liver and kidney function. In addition, the absolute neutrophil count (ANC) and procalcitonin (PCT) level were significantly higher in nonsurvivors than in survivors [Supplementary Table 2, https://links.lww.com/CM9/B557]. In comparison with those in survivors, the ratios of resistance of S. aureus to ciprofloxacin, levofloxacin, and moxifloxacin were significantly higher in nonsurvivors [Supplementary Table 3, https://links.lww.com/CM9/B557]. Of note, MRSA occurred significantly more frequently in nonsurvivors than in survivors (82% [50/61] vs. 57.3% [165/288], P <0.001). The proportion of MRSA had a significant downwards trend from 70.8% (46/65) in 2014 to 43.6% (24/55) in 2018 [Figure 1C]. The main source of SA-BSI was pneumonia (26.6%, 93/349), followed by skin/soft tissue infection (24.4%, 85/349). Compared with survivors, nonsurvivors had higher rates of pneumonia (37.7% [23/61] vs. 24.3% [70/288], P =0.032) and intra-abdominal infection (27.9% [17/61] vs. 8.7% [25/288], P <0.001) [Supplementary Table 4, https://links.lww.com/CM9/B557]. In terms of infection control, there was no significant difference in the rates of drainage of the infection source and removal of contaminated sutures between the two groups. We also did not observe any differences in mortality between the survivors and nonsurvivors based on antibiotic exposure (P >0.05). For targeted treatment, 189 (54.2%) patients received glycopeptides (vancomycin or teicoplanin), 25 (7.2%) patients received piperacillin/tazobactam, 58 (16.6%) patients received linezolid, 26 (7.4%) patients received tigecycline, and 43 (12.3%) patients received fluoroquinolone (levofloxacin or moxifloxacin), but there was no statistical significance between the two groups. In addition, a total of 8.3% (29/349) of patients did not receive appropriate therapy within 24 h after the release of antibiotic susceptibility results, but there was no difference between the two groups (6.6% [4/61] vs. 8.7% [25/288], P =0.585) [Supplementary Table 4, https://links.lww.com/CM9/B557]. The multivariate logistic regression model showed that the independent prognostic factors for 28-day mortality were age >67 years (adjusted odds ratio [aOR], 4.46; 95% confidence interval [CI], 1.18–16.88), an APACHE II score >17 (aOR, 42.47; 95% CI, 8.11–222.48), a SOFA score >7 (aOR, 8.01; 95% CI, 2.06–31.12), septic shock (aOR, 9.86; 95% CI, 1.18–82.37), and albumin <30 g/L (aOR, 5.14; 95% CI, 1.34–19.71) [Figure 1D]. There is still some controversy about the relationship between polymicrobial BSI and mortality in ICU patients. A study by Park et al[2] showed that polymicrobial SA-BSI was an independent prognostic factor for bacteremia-related mortality, which is not consistent with our current study. In their study, patients with polymicrobial SA-BSI were significantly less likely to receive appropriate empirical antibiotics than those with monomicrobial SA-BSI; 25% were infected with S. aureus plus a Gram-positive pathogen; and evidence did not shown that coinfection with Gram-positive pathogens resulted in death faster than coinfection with Gram-negative organisms. In our study, appropriate antibiotic therapy did not differ between the two groups, and the proportion of Gram-positive coinfections (36.1% [22/61]) with S. aureus was significantly higher than that in the study by Park et al[2] (25%). Therefore, we speculate that fatal polymicrobial SA-BSI is associated with a more severe condition and is not a direct independent prognostic factor of death for patients with SA-BSI. The incidence of SA-BSI has generally been reported to be higher in males than in females, while some studies have reported increased mortality in females.[3] We found that most of the proportions of infections in males in the different age groups were generally higher than those in females [Figure 1B], and the mortality rates of SA-BSI in men and women were 19.3% [47/243] and 13.2% [14/106], respectively. We did not detect any sex difference in outcomes. We found that the age distribution was left-skewed, with a peak incidence in the 60–69 years group [Figure 1B], and age >67 years was an independent prognostic factor for death in patients with SA-BSI. This might be related to more comorbidities in older patients. As described in many studies,[3] age was the strongest independent predictor of mortality, and the mortality rate increased from 6% in young individuals (<15 years old) to 57% in adults >85 years of age according to Lamagni et al.[4] Therefore, clinicians should pay more attention to the age of patients with SA-BSI than to their sex. In our current study, the proportion of MRSA had a significant downwards trend from 70.8% (46/65) in 2014 to 43.6% (24/55) in 2018, which is in line with the MRSA trend reported by CHINET,[3] probably as a result of a greater understanding of SA-BSI management. Although MRSA was associated with mortality, it was not an independent prognostic factor for death in patients with SA-BSI. This might be related to a decrease in the pathogenicity of MRSA in recent years, which is indirectly refected by the decrease in the detection rate of MRSA in our hospital in recent years. In addition, we found that septic shock was an independent prognostic factor for death in patients with SA-BSI. In conclusion, SA-BSI was characterized by a peak incidence in 60–69 years old, and pneumonia and skin/soft tissue infection as the main infection sources. There were several independent prognostic factors for 28-day mortality of SA-BSI patients, including age >67 years, an APACHE II score >17, a SOFA score >7, septic shock, and albumin <30 g/L. By contrast, MRSA and polymicrobial BSI were associated with 28-day mortality, but not independent prognostic factors. To properly manage SA-BSI, clinicians should be aware of patients with one or more of the abovementioned independent prognostic factors. Funding This work was supported in part by grants from the National Natural Science Foundation of China (No. 81901941), Natural Science Foundation of Zhejiang Province (No. LY19H150007; No. LY20H150008), Medical and Health Research Program of Zhejiang Province (No. 2019RC038; No. 2018KY427; No. 2022KY1396, No. 2022KY1398). Conflicts of interest None.

Abstract Background Although Staphylococcus aureus bloodstream infections (SA-BSI) are a common and important infection, polymicrobial SA-BSI are infrequently reported. The aim of this study was to investigate the clinical characteristics and risk factors of polymicrobial SA-BSI in comparison with monomicrobial SA-BSI. Methods A single-center retrospective observational study was performed between Jan 1, 2013, and Dec 31, 2018 at a tertiary hospital. All patients with SA-BSI were enrolled, and their clinical data were gathered by reviewing electronic medical records. Results A total of 349 patients with SA-BSI were enrolled including 54 cases (15.5%) with polymicrobial SA-BSI. In multivariable analysis, burn injury (adjusted odds ratio [OR], 7.04; 95% confidence interval [CI], 1.71–28.94), need of blood transfusion (aOR, 2.72; 95% CI, 1.14–6.50), use of mechanical ventilation (aOR, 3.11; 95% CI, 1.16–8.30), the length of prior hospital stay (aOR, 1.02; 95% CI, 1.00–1.03), and pneumonia as primary site of infection (aOR, 4.22; 95% CI, 1.69–10.51) were independent factors of polymicrobial SA-BSI. In comparison with monomicrobial SA-BSI, patients with polymicrobial SA-BSI had longer length of ICU stay [median days, 23(6.25,49.25) vs. 0(0,12), p < 0.01] and hospital stay [median days, 50(21.75,85.75) vs. 28(15,49), p < 0.01], and showed a higher 28-day mortality (29.6% vs. 15.3%, p = 0.01). Conclusions Burn injury, blood transfusion, mechanical ventilation, the length of prior hospital stay, and pneumonia as a primary site of infection are independent risk factors for polymicrobial SA-BSI. In addition, patients with polymicrobial SA-BSI might have worse outcomes compared with monomicrobial SA-BSI.

In the present paper, we construct a new type of two-hidden-layer feedforward neural network operators with ReLU activation function. We estimate the rate of approximation by the new operators by using the modulus of continuity of the target function. Furthermore, we analyze features such as parameter sharing and local connectivity in this kind of network structure.

In some applications, there are signals with a piecewise structure to be recovered. In this paper, we propose a piecewise sparse approximation model and a piecewise proximal gradient method (JPGA) which aim to approximate piecewise signals. We also make an analysis of the JPGA based on differential equations, which provides another perspective on the convergence rate of the JPGA. In addition, we show that the problem of sparse representation of the fitting surface to the given scattered data can be considered as a piecewise sparse approximation. Numerical experimental results show that the JPGA can not only effectively fit the surface, but also protect the piecewise sparsity of the representation coefficient.

Background:Although the clinical features of Acinetobacter baumannii bloodstream infection are well described, the specific clinical characteristics of polymicrobial Acinetobacter baumannii bloodstream infection have been rarely reported. The objective of this study was to examine the risk factors for and clinical outcomes of polymicrobial Acinetobacter baumannii bloodstream infection.Methods:A retrospective observational study was performed from January 2013 to December 2018 in a tertiary hospital. All patients with Acinetobacter baumannii bloodstream infection were enrolled, and the data were collected from the electronic medical records.Results:A total of 594 patients were included, 21% (126/594) of whom had polymicrobial infection. The most common copathogen was Klebsiella pneumoniae (20.81%), followed by Pseudomonas aeruginosa (16.78%) and Enterococcus faecium (12.08%). Compared with monomicrobial Acinetobacter baumannii bloodstream infection, polymicrobial Acinetobacter baumannii bloodstream infection mostly originated from the skin and soft tissue (28.6% vs. 10.5%, p < 0.001). Multivariate analysis revealed that burn injury was independently associated with polymicrobial Acinetobacter baumannii bloodstream infection (adjusted odds ratio, 3.569; 95% confidence interval, 1.954-6.516). Patients with polymicrobial Acinetobacter baumannii bloodstream infection were more likely to have a longer hospital length of stay [40 (21, 68) vs. 27 (16, 45), p < 0.001] and more hospitalization days after bloodstream infection than those with monomicrobial Acinetobacter baumannii bloodstream infection [22 (8, 50) vs. 13 (4, 28), p < 0.001]. However, no significant difference in mortality was observed between the two groups.Conclusions:Approximately one-fifth of Acinetobacter baumannii bloodstream infections were polymicrobial in this cohort. The main sources were skin and soft tissue infections, and burn injury was the only independent risk factor. Although mortality did not differ between the groups, considering the limitations of the study, further studies are required to assess the impact of polymicrobial (vs. monomicrobial) Acinetobacter baumannii bloodstream infection on outcomes.

The surface defects of a hot-rolled strip will adversely affect the appearance and quality of industrial products. Therefore, the timely identification of hot-rolled strip surface defects is of great significance. In order to improve the efficiency and accuracy of surface defect detection, a lightweight network based on coordinate attention and self-interaction (CASI-Net), which integrates channel domain, spatial information, and a self-interaction module, is proposed to automatically identify six kinds of hot-rolled steel strip surface defects. In this paper, we use coordinate attention to embed location information into channel attention, which enables the CASI-Net to locate the region of defects more accurately, thus contributing to better recognition and classification. In addition, features are converted into aggregation features from the horizontal and vertical direction attention. Furthermore, a self-interaction module is proposed to interactively fuse the extracted feature information to improve the classification accuracy. The experimental results show that CASI-Net can achieve accurate defect classification with reduced parameters and computation.

Purpose: Candida albicans (C. albicans) candidemia has been well reported in previous studies, while research on non-albicans Candida (NAC) bloodstream infections remains poorly explored. Therefore, the present study aimed to investigate the clinical characteristics and outcomes of patients with NAC candidemia. Patients and Methods: We recruited inpatients with candidemia from January 2013 to June 2020 in a tertiary hospital for this retrospective observational study. Results: A total of 301 patients with candidemia were recruited in the current study, including 161 (53.5%) patients with NAC candidemia. The main pathogens in NAC candidemia were Candida tropicalis (C. tropicalis) (23.9%), Candida parapsilosis (15.6%) and Candida glabrata (10.3%). Patients with NAC candidemia had more medical admissions (P=0.034), a higher percentage of hematological malignancies (P=0.007), a higher frequency of antifungal exposure (P=0.012), and more indwelling peripherally inserted central catheters (P=0.002) than those with C. albicans candidemia. In a multivariable analysis, prior anti fungal exposure was independently related to NAC candidemia (adjusted odds ratio [aOR], 0.312; 95% confidence interval [CI], 0.113-0.859). Additionally, NAC was obviously resistant to azoles, especially C. tropicalis had a high cross-resistance to azoles. However, no significant differences were noted in the mortality rates at 14 days, 28 days and 60 days between these two groups. Conclusion: NAC is dominant in candidemia, and prior antifungal exposure is an independent risk factor. Of note, although the outcomes of NAC and C. albicans candidemia are similar, drug resistance to specific azoles as well as cross-resistance frequently occurs in patients with NAC candidemia, and this drug resistance deserves attention in clinical practice and further in-depth investigation.

Clustering and cell type classification are a vital step of analyzing scRNA-seq data to reveal the complexity of the tissue (e.g. the number of cell types and the transcription characteristics of the respective cell type). Recently, deep learning-based single-cell clustering algorithms become popular since they integrate the dimensionality reduction with clustering. But these methods still have unstable clustering effects for the scRNA-seq datasets with high dropouts or noise. In this study, a novel single-cell RNA-seq deep embedding clustering via convolutional autoencoder embedding and soft K-means (scCAEs) is proposed by simultaneously learning the feature representation and clustering. It integrates the deep learning with convolutional autoencoder to characterize scRNA-seq data and proposes a regularized soft K-means algorithm to cluster cell populations in a learned latent space. Next, a novel constraint is introduced to the clustering objective function to iteratively optimize the clustering results, and more importantly, it is theoretically proved that this objective function optimization ensures the convergence. Moreover, it adds the reconstruction loss to the objective function combining the dimensionality reduction with clustering to find a more suitable embedding space for clustering. The proposed method is validated on a variety of datasets, in which the number of clusters in the mentioned datasets ranges from 4 to 46, and the number of cells ranges from 90 to 30 302. The experimental results show that scCAEs is superior to other state-of-the-art methods on the mentioned datasets, and it also keeps the satisfying compatibility and robustness. In addition, for single-cell datasets with the batch effects, scCAEs can ensure the cell separation while removing batch effects.

According to proteomics technology, as impacted by the complexity of sampling in the experimental process, several problems remain with the reproducibility of mass spectrometry experiments, and the peptide identification and quantitative results continue to be random. Predicting the detectability exhibited by peptides can optimize the mentioned results to be more accurate, so such a prediction is of high research significance. This study builds a novel method to predict the detectability of peptides by complying with the capsule network (CapsNet) and the convolutional block attention module (CBAM). First, the residue conical coordinate (RCC), the amino acid composition (AAC), the dipeptide composition (DPC), and the sequence embedding code (SEC) are extracted as the peptide chain features. Subsequently, these features are divided into the biological feature and sequence feature, and separately inputted into the neural network of CapsNet. Moreover, the attention module CBAM is added to the network to assign weights to channels and spaces, as an attempt to enhance the feature learning and improve the network training effect. To verify the effectiveness of the proposed method, it is compared with some other popular methods. As revealed from the experimentally achieved results, the proposed method outperforms those methods in most performance assessments.

Recently, an increasing number of studies have demonstrated that miRNAs are involved in human diseases, indicating that miRNAs might be a potential pathogenic factor for various diseases. Therefore, figuring out the relationship between miRNAs and diseases plays a critical role in not only the development of new drugs, but also the formulation of individualized diagnosis and treatment. As the prediction of miRNA-disease association via biological experiments is expensive and time-consuming, computational methods have a positive effect on revealing the association. In this study, a novel prediction model integrating GCN, CNN and Squeeze-and-Excitation Networks (GCSENet) was constructed for the identification of miRNA-disease association. The model first captured features by GCN based on a heterogeneous graph including diseases, genes and miRNAs. Then, considering the different effects of genes on each type of miRNA and disease, as well as the different effects of the miRNA-gene and disease-gene relationships on miRNA-disease association, a feature weight was set and a combination of miRNA-gene and disease-gene associations was added as feature input for the convolution operation in CNN. Furthermore, the squeeze and excitation blocks of SENet were applied to determine the importance of each feature channel and enhance useful features by means of the attention mechanism, thus achieving a satisfactory prediction of miRNA-disease association. The proposed method was compared against other state-of-the-art methods. It achieved an AUROC score of 95.02% and an AUPR score of 95.55% in a 10-fold cross-validation, which led to the finding that the proposed method is superior to these popular methods on most of the performance evaluation indexes.

Abstract Background Candida albicans (C. albicans) candidemia were well reported in previous studies, while researches on Candida non-albicans (C. non-albicans) candidemia remain poorly explored. Therefore the present study was aimed to investigate the clinical characteristics, and outcomes of C. non-albicans candidemia. Methods We recruited inpatients with candidemia from January 2013 to June 2020 in a tertiary hospital for this retrospective observational study. Results Total 301 patients with candidemia were recruited in current study, including 161 (53.5%) patients with C. non-albicans candidemia. The main pathogens in C. non-albicans candidemia were Candida tropicalis (23.9%), Candida parapsilosis (15.6%) and Candida glabrata (10.3%). Patients with C. non-albicans candidemia had more medical admissions (P = 0.034), more percentage of hematological malignancies (P = 0.007), more frequency of antifungal exposure (P = 0.012), and more indwelling peripherally inserted central catheter (P = 0.002) in comparison with C. albicans candidemia. In multivariable analysis, prior antifungal exposure was independently related to C. non-albicans candidemia (adjusted odds ratio [aOR], 0.312; 95% confidence interval [CI], 0.113–0.859). Additionally, C. non-albicans was obviously resistant to azoles, especially for C. tropicalis with a high cross-resistance to azoles. However, no significant differences were noted about the mortalities of 14 days, 28 days and 60 days between these two groups. Conclusions C. non-albicans are dominant in candidemia, and prior antifungal exposure is an independent risk factor. Of note, although outcomes between C. non-albicans and C. albicans candidemia are similar, the drug-resistance to specific azoles as well as cross-resistance frequently occurs in patients with C. non-albicans candidemia, which deserves attentions in clinical practice and further in-depth investigation.

By applying fundamental mathematical knowledge, this paper proves that the function [Formula: see text] is an integer no less than [Formula: see text] has the property that the difference between the function value of middle point of arbitrarily two adjacent equidistant distribution nodes on [Formula: see text] and the mean of function values of these two nodes is a constant depending only on the number of nodes if and only if [Formula: see text] By them, we establish an important result about deep neural networks that the function [Formula: see text] can be interpolated by a deep Rectified Linear Unit (ReLU) network with depth [Formula: see text] on the equidistant distribution nodes in interval [Formula: see text] and the error of approximation is [Formula: see text] Then based on the main result that has just been proven and the Chebyshev orthogonal polynomials, we construct a deep network and give the error estimate of approximation to polynomials and continuous functions, respectively. In addition, this paper constructs one deep network with local sparse connections, shared weights and activation function [Formula: see text] and discusses its density and complexity.

BACKGROUND:With single-cell RNA sequencing (scRNA-seq) methods, gene expression patterns at the single-cell resolution can be revealed. But as impacted by current technical defects, dropout events in scRNA-seq lead to missing data and noise in the gene-cell expression matrix and adversely affect downstream analyses. Accordingly, the true gene expression level should be recovered before the downstream analysis is carried out.RESULTS:In this paper, a novel low-rank tensor completion-based method, termed as scLRTC, is proposed to impute the dropout entries of a given scRNA-seq expression. It initially exploits the similarity of single cells to build a third-order low-rank tensor and employs the tensor decomposition to denoise the data. Subsequently, it reconstructs the cell expression by adopting the low-rank tensor completion algorithm, which can restore the gene-to-gene and cell-to-cell correlations. ScLRTC is compared with other state-of-the-art methods on simulated datasets and real scRNA-seq datasets with different data sizes. Specific to simulated datasets, scLRTC outperforms other methods in imputing the dropouts closest to the original expression values, which is assessed by both the sum of squared error (SSE) and Pearson correlation coefficient (PCC). In terms of real datasets, scLRTC achieves the most accurate cell classification results in spite of the choice of different clustering methods (e.g., SC3 or t-SNE followed by K-means), which is evaluated by using adjusted rand index (ARI) and normalized mutual information (NMI). Lastly, scLRTC is demonstrated to be also effective in cell visualization and in inferring cell lineage trajectories.CONCLUSIONS:a novel low-rank tensor completion-based method scLRTC gave imputation results better than the state-of-the-art tools. Source code of scLRTC can be accessed at https://github.com/jianghuaijie/scLRTC .

Local self-similarity of 3D model is a fundamental problem in the shape analysis. The construction of a local shape descriptor is very important to the final result of self-similarity analysis. To solve this problem, a self-similarity analysis method based on the tensor fusion feature descriptor is proposed. Firstly, the shape diameter function (SDF) of a point cloud model is approximately calculated by using relevant facets and antipodal points. Then, spectral clustering is used to segment the model into sub-blocks, and the three-dimensional feature tensor is constructed from the SDF, shape index (SI) and Gauss curvature (GS) matrix of KNN neighborhood points. Finally, the shape descriptor is obtained by constructing the mapping with the tensor norm, and then the similarity measure is defined and the self-similarity between the sub-blocks of the model is analyzed. Several state-of-the-art methods (including partial matching and saliency detection) are tested. In terms of not only the visual effect, but also the similarity measure and the relative errors, the results show that this method can effectively describe the shape and improves the recognition accuracy of similar sub-blocks of a point cloud model.

The 3D structure of a protein is closely related to its function, and the similarity analysis between their structures can help reveal the function of proteins. However, there exist two problems arising from the analysis of 3D structures of proteins. The proteins with a similar sequence may have different structures, while the proteins with a similar structure may have different sequences. In the analysis of similarity in 3D structures of proteins, it remains difficult for the traditional methods using the spatial feature distribution and geometry or topology features of proteins to solve these problems. In this paper, a Tile-CNN network is proposed to analyze the similarity of proteins in 3D structure. In order to capture the overall and the local features as exhibited by the 3D structures of proteins, it projects 3D protein models into 2D protein images from different views and then cuts these 2D projected images using the tile strategy. After the training of proteins with these images in the Tile-CNN, the test protein model can be expressed by an analysis matrix, and then the similarity between 3D structures of proteins is computed using the root mean square distance (RMSD) for the benchmark matrix and the analysis matrix. As revealed by the experimental results, the proposed algorithm is more robust in analyzing the similarity of 3D structures of proteins and produces a satisfactory performance in solving the two aforementioned problems.