Latest ArticlesTo propose a transfer learning-based method for breath sound feature recognition and autonomous determination of sputum suction timing.
An electronic stethoscope was used to collect breath sounds from the main airways of clinically ventilated patients before and after sputum suction, with pre-suction breath sounds labeled as requiring suction. The collected data underwent high-pass filtering and wavelet soft-threshold denoising, followed by the extraction of log-Mel spectrograms. A VGGish model pretrained on the Audio Set dataset was then employed to extract feature vectors from these spectrograms, which were subsequently classified using a support vector machine to determine whether suction was required.
The precision, recall and F1 score for recognition of breath sounds requiring sputum suction were 86.73%, 93.06% and 89.78%, respectively.
The proposed breath sound recognition method based on transfer learning effectively determines the timing of sputum suction and shows a significant clinical potential.
To investigate the flow field characteristics of peritoneal fluid flowing in the cavity between the liver and the inner wall of the diaphragm peritoneum in patients with ovarian cancer and the effects on deformation of the liver and diaphragm peritoneum.
A bidirectional fluid-structure interaction (FSI) analysis was conducted using COMSOL to investigate the interaction between the peritoneal fluid and the liver and diaphragm peritoneum under varying inlet velocities and viscosity functions.
The accuracy of the simulation was validated by comparing the simulation results with the contour lines of the CT scans, and the displacement error between the two was smaller than 5%. When the inlet velocity of the abdominal fluid increased from 0.1 m/s to 0.15 m/s, convex deformation of the diaphragm peritoneum increased by 193.3 μm, and concave deformation decreased by 304.1 μm. Meanwhile, the increase of the inlet velocity made the viscosity near the wall of deformed area increased, which improved the probability of the metastatic implantation of the cancer cells. The higher the viscosity in the main body region of the viscosity function, the larger convex deformation of the diaphragm peritoneum; the larger the linear fitting value in the tail, the smaller the concave deformation. The viscosity of the concave deformation area near the outlet of the right lobe was much larger than that of other areas, and cancer cells were more likely to metastasise in this area.
This study elucidates the relationship between peritoneal fluid flow and solid deformation, predicting the regions prone to cancer cell metastasis and implantation under various conditions. The findings provide a theoretical foundation for studying the motion of cancer cells within the flow field.
The wear debris generated during artificial joint prosthesis service can react with bone tissues to form osteolysis, seriously affecting the life-time of artificial joint prostheses. This paper reviews, summarizes, and analyzes domestic and international research literature on the extraction, characterization, and identification of wear debris from different artificial joint materials, aiming to provide references and feasible ideas for the future construction of a systematic and hierarchical research system for artificial joint wear debris. The main findings are as follows: strong alkali protein degradation test, strong acid protein degradation test, and protease protein degradation test are the commonly used method for extracting artificial joint wear debris, and researchers have clarified the protein degradation mechanisms of these three debris extraction methods. The characterization of wear debris in-vitro and in-vivo is mostly for hip and knee joints, with a small amount involving cervical spine and ankle joints. Studies have shown that the size, quantity, shape, and volume of wear particles are influenced by factors such as joint type, contact area, material selection, and implantation time. Both domestic and international studies have conducted characterization research on wear debris after in-vitro simulation testing, but there is still a lack of wear debris characterization analysis of clinical retrievals in China. Currently, most research is on the recognition of wear debris in the traditional mechanical field, but research on the intelligent recognition of artificial joint wear debris is relatively few, indicating that there is a certain lag in the application of computer technology in the field of artificial joint wear debris recognition.
A microfluidic extracorporeal membrane oxygenator is an advanced extracorporeal life support device designed using microfluidic technology, capable of providing oxygenation support to patients with severe respiratory failure and other pulmonary diseases via extracorporeal circulation. Compared to conventional extracorporeal membrane oxygenators, it features a more superior biomimetic design, demonstrating potentials for improved therapeutic outcomes and reduced complications. This review summarizes the research progress of microfluidic extracorporeal membrane oxygenators in terms of hemodynamics, membrane materials, biocompatibility, gas exchange efficiency, and structural design. It analyzes how factors such as blood channel design, material selection and surface modification techniques impact the performance of microfluidic extracorporeal membrane oxygenators, such as biomimetic flow paths minimizing shear stress and endothelial cell linings significantly reducing thrombosis. Finally, the limitations of microfluidic extracorporeal membrane oxygenators are discussed, along with prospects for future development. Innovations are still needed in enhancing biocompatibility, portability, manufacturability, and cost reduction for microfluidic extracorporeal membrane oxygenators.
Cardiovascular diseases are the leading cause of death worldwide, and hemodynamics plays a significant role in understanding the mechanisms of these diseases, predicting disease progression, and guiding treatment strategies. Traditional methods for obtaining personalized hemodynamic parameters in clinical settings have numerous limitations, while the rise of deep learning technology has brought new opportunities for their computation. This review focuses on the application of deep learning in obtaining hemodynamic parameters in clinical settings, covering its progress in computational fluid dynamics preprocessing, hemodynamic computation (data-driven and PINN method), and magnetic resonance anagiography. It analyzes the advantages and challenges of each method and discusses future development directions, aiming to provide a reference for research on obtaining hemodynamic parameters in clinical settings using artificial intelligence method.
Constructing functional microvascular networks in vitro represents a pivotal step in the creation of engineered tissues, organ-on-chip models, and organoids, holding profound implications for tissue engineering, regenerative medicine, drug screening, and disease modeling. As a cutting-edge bio-manufacturing approach, bioprinting enables the precise deposition of biomaterials, cells, and bioactive molecules to fabricate intricate microvascular networks that faithfully replicate the geometric architecture and functional properties of native microvasculature. This review summarizes the research progress in bioprinting microvascular networks, with a focus on bioprinting technologies, bioinks, and the biomechanical functional evaluation of microvascular networks.
No. 86 Xueyuan South Road, Haidian District, Beijing
010-62662699(Operation and Technical Support)
qkjq@cast.org.cn
Copyright © 2025 China Association for Science and Technology. All rights reserved. For all open access content, the relevant licensing terms apply.
Sponsored by the Office of the Leading Group for Cybersecurity and Informatization of CAST, and supported by Science and Technology Review Publishing House
