Search

Nano-TiCp/GH3536 composite powders were prepared by mechanical alloying （MA） and thermal plasma spheroidization （TPS） methods. The flowability， bulk density， particle size distribution， sphericity， and microstructure of the composite powders before and after MA-TPS treatment were studied by Hall flow meter，Scott volumeter， XRD， particle size analyzer， scanning electron microscope， and transmission electron microscope. The results show that the preformed powders obtained by MA-5 h treatment are irregular shape powders with non-uniform adhesion of nano-TiC particles on the surface of the GH3536 substrate. Compared with the original GH3536 powders， the number of equiaxed grains in the preformed powders increases and the crystal grains are refined， but the flowability and bulk density decrease.The nano-TiCp/GH3536 composite powders obtained by the second TPS treatment are spherical， and the original irregular nano-TiC particles are spheroidized and uniformly distributed inside the GH3536 substrate. After removing the submicron particles with an average particle size of 183.27 nm， the flowability， bulk density， and sphericity of TPS composite powders are further improved.The TEM images of submicron particles show that the nanoscale GH3536 microspheres are uniformly coated with a layer of TiC， the structure is similar to the core-shell， and spherical TiC particles are found in the powder. The average spherical degree， average particle size， flowability， and bulk density of the nano-TiCp/GH3536 composite powders prepared by the two-step method are 0.965，33.65 μm，17.1 s/50 g，and 4.184 g/cm3， respectively.

A method based on graph transformation and transfer learning is proposed to further enhance the accuracy of household-transformer relationships and phase identification in low-voltage distribution networks. Firstly， a graph transformation method based on Gramian angular field （GAF） is introduced to convert electricity consumption data into a two-dimensional representation， facilitating the identification of differences in one-dimensional time-series electricity consumption data. Next， to address challenges such as sparse user data in low-voltage distribution networks， limited data acquisition methods， and a scarcity of samples， a deep learning model suitable for household-transformer relationship and phase identification is constructed using transfer learning and leveraging pre-trained parameter weights. Experimental validation demonstrates that the proposed model outperforms mainstream methods in both household-transformer relationship and phase identification， exhibiting improved accuracy and stability.

Objectives. This study aimed to use laboratory and clinical data of systemic lupus erythematosus (SLE) patients to construct prediction models for fundus complications in SLE. Methods. Routine blood test data and clinical information of 277 SLE patients were collected retrospectively. Based on their fundus examination, they were divided into two groups, with or without fundus lesions, defined as retinopathy and choroidopathy in this study. The data of the two groups were compared, and the prediction model was established using binary logistic regression analysis. Results. There were 85 patients in the fundus lesions' group and 192 patients in the control group. Between the two groups, age, SLEDAI, serositis, hypertension, diabetes, anticardiolipin antibody (ACA), anti‐Sm antibody, C‐reactive protein (CRP), hemoglobin (Hb), platelet count (PLT), albumin (Alb), serum creatinine(Scr), urea, uric acid(UA), and immunoglobulin G(IgG) were significantly different (p < 0.05). Besides, age, SLEDAI, serositis, hypertension, diabetes, anti‐SSB, CRP, Hb, PLT, FIB, Alb, Scr, urea, UA, GLU, and IgG were significantly correlated with SLE‐related fundus lesions. PLT, fibrinogen (FIB), IgG, and urea were independent risk factors of SLE‐related fundus lesions. The area under the curve (AUC) was 0.830 (p < 0.001; 95% CI = 0.762–0.898), and the nomogram was established with great evaluation efficiency demonstrated by the calibration curve and the Hosmer–Lemeshow test. The result of k‐fold cross‐validation also showed high prediction accuracy. Conclusions. We have found the independent risk factors of SLE‐related fundus lesions and developed a model to improve the prediction of fundus lesions in SLE. [ABSTRACT FROM AUTHOR]

Purpose To construct a clinical noncontrastive computed tomography (NCCT) deep learning joint model for predicting early hematoma expansion (HE) after cerebral hemorrhage (sICH) and evaluate its predictive performance. Methods All 254 patients with primary cerebral hemorrhage from January 2017 to December 2022 in the General Hospital of the Western Theater Command were included. According to the criteria of hematoma enlargement exceeding 33% or the volume exceeding 6 ml, the patients were divided into the HE group and the hematoma non-enlargement (NHE) group. Multiple models and the 10-fold cross-validation method were used to screen the most valuable features and model the probability of predicting HE. The area under the curve (AUC) was used to analyze the prediction efficiency of each model for HE. Results They were randomly divided into a training set of 204 cases in an 8:2 ratio and 50 cases of the test set. The clinical imaging deep feature joint model (22 features) predicted the area under the curve of HE as follows: clinical Navie Bayes model AUC 0.779, traditional radiology logistic regression (LR) model AUC 0.818, deep learning LR model AUC 0.873, and clinical NCCT deep learning multilayer perceptron model AUC 0.921. Conclusion The combined clinical imaging deep learning model has a high predictive effect for early HE in sICH patients, which is helpful for clinical individualized assessment of the risk of early HE in sICH patients. Key message What is already known on this topic: Prior to this study, existing research has highlighted various factors associated with hematoma expansion (HE) in intracerebral hemorrhage. These factors include age, Glasgow Coma Scale (GCS) score, hematoma characteristics, and time-related parameters. However, the predictive accuracy of individual factors has been limited, necessitating the development of more comprehensive prediction models. Deep learning features methods have shown promise in improving risk prediction but have not been applied to HE prediction. What this study adds: This study introduces the clinical noncontrastive computed tomography (NCCT) deep learning combined model, a novel approach that effectively predicts HE and enhances predictive accuracy. By incorporating a combination of clinical, imaging, and deep learning features, this model offers improved predictive capabilities compared to existing methods. Independent risk factors identified in this study for early hematoma enlargement include age, age group, GCS score, hematoma shape and location, midline shift, blend sign, pre-H-slicer, and time from onset to examination (H). How this study might affect research, practice or policy: The development of the clinical NCCT deep learning combined model presents a significant advancement in predicting HE, potentially aiding in the identification of high-risk patients for targeted interventions and enrollment in clinical trials. By demonstrating the value of combining clinical, imaging, and deep learning features in predicting hematoma enlargement, this study paves the way for future research to explore innovative approaches in risk assessment and management strategies for patients with intracerebral hemorrhage. [ABSTRACT FROM AUTHOR]

High-quality, chromosome-scale genomes are essential for genomic analyses. Analyses, including 3D genomics, epigenetics, and comparative genomics rely on a high-quality genome assembly, which is often accomplished with the assistance of Hi-C data. Curation of genomes reveal that current Hi-C-assisted scaffolding algorithms either generate ordering and orientation errors or fail to assemble high-quality chromosome-level scaffolds. Here, we offer the software Puzzle Hi-C, which uses Hi-C reads to accurately assign contigs or scaffolds to chromosomes. Puzzle Hi-C uses the triangle region instead of the square region to count interactions in a Hi-C heatmap. This strategy dramatically diminishes scaffolding interference caused by long-range interactions. This software also introduces a dynamic, triangle window strategy during assembly. Initially small, the window expands with interactions to produce more effective clustering. Puzzle Hi-C outperforms available scaffolding tools. [ABSTRACT FROM AUTHOR]

As one of the most important food crops in the world, rice yield is directly related to national food security. Lodging is one of the most important factors restricting rice production, and the cultivation of rice varieties with lodging resistance is of great significance in rice breeding. The lodging resistance of rice is directly related to the mechanical strength of the stalks. In this paper, we reviewed the cell wall structure, its components, and its genetic regulatory mechanism, which improved the regulatory network of rice stalk mechanical strength. Meanwhile, we analyzed the new progress in genetic breeding and put forward some scientific problems that need to be solved in this field in order to provide theoretical support for the improvement and application of rice breeding. [ABSTRACT FROM AUTHOR]