Explore the vast range of titles available.

The task of UAV-based maritime rescue object detection faces two significant challenges: accuracy and real-time performance. The YOLO series models, known for their streamlined and fast performance, offer promising solutions for this task. However, existing YOLO-based UAV maritime rescue object detection methods tend to prioritize high accuracy, often at the expense of real-time performance and ease of implementation and expansion. This study proposes a modular plug-and-play optimization approach based on the YOLOv8 framework, aiming to enhance real-time performance while maintaining high accuracy for UAV maritime rescue object detection. The proposed optimization modules are flexible, easy to implement, and extendable. In experiments on the large-scale publicly available SeaDronesSee dataset, our method achieved a 13.53% improvement in accuracy over YOLOv8x while reducing computational cost by 85.63%. Additionally, it surpassed the detection speed of the SeaDronesSee official code’s two-stage detector by over 20 times, while maintaining comparable accuracy. Furthermore, our analysis of the experimental results highlights differences in detection difficulty among various objects and potential biases within the dataset.

Accurately and rapidly detecting objects and their locations in drone-captured images from maritime search and rescue scenarios provides valuable information for rescue operations. The YOLO series, known for its balance between lightweight architecture and high accuracy, has become a popular method among researchers in this field. Recent advancements in the newly released YOLO11 model have demonstrated significant progress in general object detection tasks across everyday scenarios. However, its application to the specific task of drone-based maritime search and rescue still leaves substantial room for improvement. To address this gap, we propose targeted optimizations to enhance YOLO11’s performance in this domain. These include integrating a Space-to-Depth module into the Backbone, incorporating a content-aware upsampling algorithm in the Neck, and adding an extra detection head to better exploit shallow image features. These modifications significantly improve the model’s ability to detect small, overlapping, and rarely occurring objects, which are common challenges in maritime search and rescue tasks. Experimental evaluations conducted on the large-scale SeaDronesSee dataset demonstrate that the proposed optimized YOLO11 outperforms YOLOv8, YOLO11, and MambaYOLO across all scales. Moreover, under lightweight configurations, the model achieves substantial performance gains over YoloOW, a method renowned for its accuracy but depends on heavyweight configurations. In the lightweight complexity range, the proposed model achieves a relative accuracy improvement of 20.85% to 43.70% compared to these state-of-the-art methods. The code supporting this research is available at https://github.com/bgno1/sds_yolo11 .

Subtle changes in white matter (WM) microstructure have been associated with normal aging and neurodegeneration. To study these associations in more detail, it is highly important that the WM tracts can be accurately and reproducibly characterized from brain diffusion MRI. In addition, to enable analysis of WM tracts in large datasets and in clinical practice it is essential to have methodology that is fast and easy to apply. This work therefore presents a new approach for WM tract segmentation: Neuro4Neuro, that is capable of direct extraction of WM tracts from diffusion tensor images using convolutional neural network (CNN). This 3D end-to-end method is trained to segment 25 WM tracts in aging individuals from a large population-based study (N ​= ​9752, 1.5T MRI). The proposed method showed good segmentation performance and high reproducibility, i.e., a high spatial agreement (Cohen’s kappa, κ=0.72−0.83) and a low scan-rescan error in tract-specific diffusion measures (e.g., fractional anisotropy: ε=1%−5%). The reproducibility of the proposed method was higher than that of a tractography-based segmentation algorithm, while being orders of magnitude faster (0.5s to segment one tract). In addition, we showed that the method successfully generalizes to diffusion scans from an external dementia dataset (N ​= ​58, 3T MRI). In two proof-of-principle experiments, we associated WM microstructure obtained using the proposed method with age in a normal elderly population, and with disease subtypes in a dementia cohort. In concordance with the literature, results showed a widespread reduction of microstructural organization with aging and substantial group-wise microstructure differences between dementia subtypes. In conclusion, we presented a highly reproducible and fast method for WM tract segmentation that has the potential of being used in large-scale studies and clinical practice. •We present a direct CNN-based method for white matter tract segmentation from DTI.•The method accurately segmented 25 tracts from a large population dataset.•It showed higher reproducibility than compared tractography-based methodology.•The method generalized well to clinical-quality, dementia, and cross-scanner data.•It enables fast, easy, and reliable analysis of brain microstructure in aging and disease.

Diagnostic errors represent a critical issue in clinical diagnosis and treatment. In China, the rate of misdiagnosis in clinical diagnostics is approximately 27.8%. By comparison, in the United States, which boasts the most developed medical resources globally, the average rate of misdiagnosis is estimated to be 11.1%. It is estimated that annually, approximately 795,000 Americans die or suffer permanent disabilities due to diagnostic errors, a significant portion of which can be attributed to physicians’ failure to make accurate clinical diagnoses based on patients’ clinical presentations. Differential diagnosis, as an indispensable step in the clinical diagnostic process, plays a crucial role. Accurately excluding differential diagnoses that are similar to the patient’s clinical manifestations is key to ensuring correct diagnosis and treatment. Most current research focuses on assigning accurate diagnoses for specific diseases, but studies providing reasonable differential diagnostic assistance to physicians are scarce. This study introduces a novel solution specifically designed for this scenario, employing machine learning techniques distinct from conventional approaches. We develop a differential diagnosis recommendation computation method for clinical evidence-based medicine, based on interpretable representations and a visualized computational workflow. This method allows for the utilization of historical data in modeling and recommends differential diagnoses to be considered alongside the primary diagnosis for clinicians. This is achieved by inputting the patient’s clinical manifestations and presenting the analysis results through an intuitive visualization. It can assist less experienced doctors and those in areas with limited medical resources during the clinical diagnostic process. Researchers discuss the effective experimental results obtained from a subset of general medical records collected at Shengjing Hospital under the premise of ensuring data quality, security, and privacy. This discussion highlights the importance of addressing these issues for successful implementation of data-driven differential diagnosis recommendations in clinical practice. This study is of significant value to researchers and practitioners seeking to improve the efficiency and accuracy of differential diagnoses in clinical diagnostics using data analysis.

The hydraulic pipeline is subject to the aero-engine base excitation and the pump fluid pulsation which can always damage the pipeline through overload to fatigue. So, the health monitoring technique of hydraulic pipeline is essential for the maintenance of the aero-engine. In this article, the Kalman filter combined with fiber Bragg grating method is proposed to detect the location faults of the hydraulic pipeline system. In this method, the description of state equations for the hydraulic pipeline vibration signals are presented based on autoregressive model. Then, a practical strategy detection method is proposed to investigate the vibration signals of hydraulic pipeline based on the Kalman filter technique. Finally, the clamp loosening and the collision faults of the hydraulic pipeline are conducted as an example to validate the proposed approach. The obtained results show that the present technique is convenient and efficient to detect the location faults of hydraulic pipeline where the fiber Bragg grating sensors are fixed, which can serve as an effective guidance for the health monitoring of hydraulic pipeline in aero-engine.

Existing buildings energy-saving renovation is an important means to cope with global warming and an essential component of achieving China’s energy conservation and pollution emissions reduction strategy goals. The development of the energy-saving renovation market is closely related to its influencing factors, which determine the reasons and conditions for the development of the system; therefore, it is necessary to study the influencing factors of energy-saving transformations. System dynamics was applied to explore the feedback relationship between the service subsystem, the demand market subsystem, and the market regulation subsystem. Analysis was performed for the intrinsic influencing factors of the development of the existing building energy-saving renovation market and the interaction law of feedback relationship. This paper discusses the basic characteristics of government incentives, Energy Service Company (ESCO) technology innovation, ESCO’s revenue, and owner’s awareness to promote the development of the existing building energy-saving renovation market. Base on those, it puts forward suggestions for promoting the market development of existing buildings energy-saving reconstruction. The findings provided a theoretical basis and guiding role for the Chinese government to formulate support policies for existing building energy-saving renovation. At the same time, it also provides reference for other countries to develop existing buildings energy-saving renovation market.

Writing various medical records takes significant daily workload for physicians. Generative AI technique has the advantage in tasks of data-to-text generation and text summarization, and brings opportunities to reduce workload for physicians to work on medical records. However, current general Large Language Models (LLMs) cannot satisfy the strict requirements to correctness of generative texts in specific tasks of medical record generation. In addition, due to the constraints to protect patient privacy, physicians cannot upload patient data to public cloud services for LLM cloud service. We develop optimized LLMs for medical record generation, which can be deployed in hospitals and integrated with the Electronic Medical Record (EMR) applications for physicians to reduce workload of writing medical records. We propose an approach for constructing data augmented LLM on medical record generation. As for each specific task, we extract annotated data with high quality from the EMR application in a hospital. Based on such data and customized instruct, we construct certain optimized models for specific tasks, including medical Data-to-Text generation (from structural medical data to history of present illness) and medical text summarization (from a series of progress notes to discharge summary). Furthermore, we propose Faithfulness score, a evaluation metrics, based on semantic similarity between the generative texts by LLMs and reference texts by physicians. Extensive experiments are conducted with high-quality task-specific medical data, and tested with our optimized models and two other models, including a general state-of-the-art (SOTA) model and a medical model, thereby evaluating the correctness of the generated medical records by Faithfulness score, separately on the two specific tasks.The experimental results demonstrate that our optimized models has improved the Faithfulness score of the generated medical records, respectively by 19.72% and 19.33% rather than the existing SOTA models, on medical Data-to-Text generation and medical text summarization.Our work has been validated and applied in the hospital we cooperate with, and save approximately 0.5-1 hour of working time per day for a physician, so that he or she can spend more time in taking care of his or her patients.This method can be generalized to any hospital, using its native medical data, to achieve a specialized model available for medical record generation tasks.The code is available at https://github.com/LotusPhilip/data-augmented-model

Writing various medical records takes significant daily workload for physicians. Generative AI technique has the advantage in tasks of data-to-text generation and text summarization, and brings opportunities to reduce workload for physicians to work on medical records. However, current general Large Language Models (LLMs) cannot satisfy the strict requirements to correctness of generative texts in specific tasks of medical record generation. In addition, due to the constraints to protect patient privacy, physicians cannot upload patient data to public cloud services for LLM cloud service. We develop optimized LLMs for medical record generation, which can be deployed in hospitals and integrated with the Electronic Medical Record (EMR) applications for physicians to reduce workload of writing medical records. We propose an approach for constructing data augmented LLM on medical record generation. As for each specific task, we extract annotated data with high quality from the EMR application in a hospital. Based on such data and customized instruct, we construct certain optimized models for specific tasks, including medical Data-to-Text generation (from structural medical data to history of present illness) and medical text summarization (from a series of progress notes to discharge summary). Furthermore, we propose Faithfulness score, a evaluation metrics, based on semantic similarity between the generative texts by LLMs and reference texts by physicians. Extensive experiments are conducted with high-quality task-specific medical data, and tested with our optimized models and two other models, including a general state-of-the-art (SOTA) model and a medical model, thereby evaluating the correctness of the generated medical records by Faithfulness score, separately on the two specific tasks.The experimental results demonstrate that our optimized models has improved the Faithfulness score of the generated medical records, respectively by 19.72% and 19.33% rather than the existing SOTA models, on medical Data-to-Text generation and medical text summarization .Our work has been validated and applied in the hospital we cooperate with, and save approximately 0.5-1 hour of working time per day for a physician, so that he or she can spend more time in taking care of his or her patients.This method can be generalized to any hospital, using its native medical data, to achieve a specialized model available for medical record generation tasks.The code is available at https://github.com/LotusPhilip/data-augmented-model

Cen X-3 is an archetypical X-ray pulsar with strong flux variations and alternating torque reversals, both of which are similar to those of recently discovered pulsating ultra-luminous X-ray sources. We study a low state of Cen X-3 occurred in 2023 lasting for \\(\\sim100\\) days with Chandra and Insight-HXMT observations, supplemented with MAXI and Fermi/GBM data. The Chandra spectrum during the eclipse of Cen X-3 in the low state is very similar to that in the high state, especially, the Fe lines. The HXMT spectrum in the low state shows an enhanced Fe line, so do the MAXI data. The spin-up/spin-down trends of Cen X-3 are not affected by the low states. All these results indicate that the intrinsic emission in the low states is high, and the low states are just apparently low and are dominated by reprocessed emission. We found that the spin-up to spin-down reversals of Cen X-3 take longer time than the spin-down to spin-up reversals, which provides a definite observation test of any possible torque-reversal models. We discuss insights of these results for understanding the pulsating ultra-luminous X-ray sources.

Recently a retrograde neutron star is proposed for the classical wind-fed X-ray pulsar, GX 301-2, to explain the orbital spin-up to spin-down reversal near periastron, based on the stream model invoked to explain the pre-periastron flare of GX 301-2 previously. We study in detail three rare spin-up events detected by Fermi/GBM and find that the spin derivatives are correlated with the Swift/BAT fluxes, following a relation of \\( F^0.750.05\\). All the spin-up events of GX 301-2 started about 10 days after the periastron, which is the time needed for tidally stripped gas to reach the neutron star. The slow rotation of the optical companion implies that the accreted matter is likely to have angular momentum in the direction of the orbital motion, as in a Roche-Lobe-like overflow. As a result, the spin-up events of GX 301-2 would favor accretion of a prograde disk to a prograde neutron star. We also find that the flare of intrinsic X-ray emission of GX 301-2 happened 0.4 days before periastron, while the flare of low energy emission (2-10 keV) happened about 1.4 days before periastron. The preceding low energy flare can be explained by stronger absorption of the intrinsic X-ray emission closer to the periastron. This finding weakened the need of the stream model. The pulse fraction of GX 301-2 near periastron is reduced heavily, which is likely caused by Compton scattering process. Compton reflection from the optical companion might be responsible for the observed orbital spin reversal of GX 301-2.