Training Model Research Articles

Hub and Spokes-Implementation of Basic Palliative Care into University Hospital Practice-A Pilot Study.

Background: The aim is to implement knowledge of basic palliative care in selected departments by the Hub and Spokes model. Methods: Implementation of basic palliative care was designed as a stepwise training model by skills lectures over a time period of 2 years. In each of the six selected oncological and nononcological departments, one physician and two nurses were trained in semi-annual half-day meetings as expert representatives within their departments. Results: Semi-structured interviews were conducted to assess implementation outcomes with 12 nurses and 6 physicians. Overall acceptability was high for all departments and professions. Feasibility was given in all departments. Adoption and penetration of a trained expert representative differed between medical and nursing professions. Implementation was more appropriate in the stationary sector. Implementation costs were low. Conclusion: Expansion of the system into a second follow-up period, including more departments, is planned to ensure sustainability.

FedUB: Federated Learning Algorithm Based on Update Bias

Federated learning, as a distributed machine learning framework, aims to protect data privacy while addressing the issue of data silos by collaboratively training models across multiple clients. However, a significant challenge to federated learning arises from the non-independent and identically distributed (non-iid) nature of data across different clients. non-iid data can lead to inconsistencies between the minimal loss experienced by individual clients and the global loss observed after the central server aggregates the local models, affecting the model’s convergence speed and generalization capability. To address this challenge, we propose a novel federated learning algorithm based on update bias (FedUB). Unlike traditional federated learning approaches such as FedAvg and FedProx, which independently update model parameters on each client before direct aggregation to form a global model, the FedUB algorithm incorporates an update bias in the loss function of local models—specifically, the difference between each round’s local model updates and the global model updates. This design aims to reduce discrepancies between local and global updates, thus aligning the parameters of locally updated models more closely with those of the globally aggregated model, thereby mitigating the fundamental conflict between local and global optima. Additionally, during the aggregation phase at the server side, we introduce a metric called the bias metric, which assesses the similarity between each client’s local model and the global model. This metric adaptively sets the weight of each client during aggregation after each training round to achieve a better global model. Extensive experiments conducted on multiple datasets have confirmed the effectiveness of the FedUB algorithm. The results indicate that FedUB generally outperforms methods such as FedDC, FedDyn, and Scaffold, especially in scenarios involving partial client participation and non-iid data distributions. It demonstrates superior performance and faster convergence in tasks such as image classification.

Response prediction and probabilistic analysis of the vehicle-ballasted track system considering track irregularity based on long-short term memory neural network

It is necessary to efficiently and accurately predict the dynamic behavior of the train-track system subjected to massive random track geometric excitations for system evaluation and maintenance. This paper proposed an engineering practical approach for the train-track system behavior prediction by integrating the deep learning surrogate model-based long-short term memory (LSTM) and physical model of the train-ballasted track interaction (TBTI) system, and then the generalized probability density evolution method (GPDEM) is introduced to achieve the probabilistic assessment of the TBTI system under limited track irregularities, where the track irregularity probability model (TIPM) coupled with generalized-F discrepancy points sets is employed to compress the track irregularity sets to reduce the number of system responses (as training target) but without loss of the responses completeness, therefore, it can promote the efficiency and accuracy of training model. The prediction performance of the surrogate model is illustrated by comparing the time-frequency domain information, mean and standard deviation of the responses obtained by the physical model and surrogate model. The results indicated the proposed method can accurately predict the system response, especially the low and medium frequency response below 50 Hz, and improve the efficiency of stochastic analysis for the TBTI system by approximately 20 times. Furthermore, the applicability of the proposed method is further examined by investigating the influence of probability levels of track irregularity and train speeds on train performance. Its extended model is also implemented to predict the response of systems induced by any measured track irregularity in real-time, which promotes the generalization capability of the proposed surrogate model.

The influence of gender and sport on popliteal angle and dorsiflexion in junior high school students.

The aim of the study was to assess factors affecting the popliteal angle and foot dorsiflexion, in particular gender. The subjects were 142 students from the 2nd and 3rd year of Poznań junior high schools. The participants included 57 girls and 87 boys. Three raters examined each subject: a specialist in orthopaedics, a resident doctor and a physical therapy student. Foot dorsal flexion was tested in a supine position with lower limbs extended. Next, dorsal flexion was evaluated with the knee and hip in 90 degrees of flexion. Finally, a passive knee extension (PKE) test was carried out. The significance of the PKE test is that the lower the angle the more flexible the hamstrings. This is because the PKE measurement is the distance to the right angle, that is a full knee extension with the hip flexed. The non-parametric test (Mann-Whitney) and the Student's t-test showed differences between the female and male gender in the measurements of the popliteal angle (p < .05000). The correlation was negative, which means that the hamstrings are more flexible in girls. No differences were found between gender and passive foot dorsiflexion and dorsiflexion with a flexed hip and knee. No differences were found between the group with the extended PE curriculum and the group with the standard number of PE classes in the range of motion of foot dorsiflexion and the value of the popliteal angle. Girls between 13 and 15years old have a significantly larger hamstring flexibility, which is confirmed by the tests of the popliteal angle. No differences were found in dorsiflexion between girls and boys who have not been trained using a training model.

Enhancing visual comfort prediction in space layouts using deep learning models

ABSTRACT In designing performance-based facilities, such as educational buildings, assessing visual comfort is crucial. The computational cost and resource-intensive modeling necessitate an efficient analytical tool for diverse layout assessment. In the current study, predictive models were developed using deep learning and machine learning to predict visual comfort in typical elementary schools in Tehran. Firstly, layouts were modeled, considering influential parameters. Subsequently, simulation-based datasets were analyzed and labeled. The VGG16 and VGG19 Architectures from convolutional neural networks (CNNs), along with the pix2pix model from generative adversarial networks for forecasting respectively, numerical and pictorial indices regarding each visual comfort metric. The pix2pix model performs approximately SSIM of 0.9 for sDA, ASE, and UDIexceed3000 and 0.86 for DAp. The extracted features by CNNs were harnessed in training models. Eventually, models with the Bayesian Ridge algorithm had a promising performance which exhibited acceptable R2 values for metrics in the range of 0.90–0.96. Toward a depth analysis, parameter sensitivity using the Shapley additive explanations method was evaluated with XGBoost models. Additionally, the Spearman Correlation underscores the substantial impact of factors like WWR and aspect ratio on metrics.

Multitask Learning Strategy with Pseudo-Labeling: Face Recognition, Facial Landmark Detection, and Head Pose Estimation

Most facial analysis methods perform well in standardized testing but not in real-world testing. The main reason is that training models cannot easily learn various human features and background noise, especially for facial landmark detection and head pose estimation tasks with limited and noisy training datasets. To alleviate the gap between standardized and real-world testing, we propose a pseudo-labeling technique using a face recognition dataset consisting of various people and background noise. The use of our pseudo-labeled training dataset can help to overcome the lack of diversity among the people in the dataset. Our integrated framework is constructed using complementary multitask learning methods to extract robust features for each task. Furthermore, introducing pseudo-labeling and multitask learning improves the face recognition performance by enabling the learning of pose-invariant features. Our method achieves state-of-the-art (SOTA) or near-SOTA performance on the AFLW2000-3D and BIWI datasets for facial landmark detection and head pose estimation, with competitive face verification performance on the IJB-C test dataset for face recognition. We demonstrate this through a novel testing methodology that categorizes cases as soft, medium, and hard based on the pose values of IJB-C. The proposed method achieves stable performance even when the dataset lacks diverse face identifications.

Training models of highly skilled cyclists in the pre-competitive mesocycle using a pedal power meter

Abstract. The need for systematic improvement of results and demonstration of high sports results, which are sometimes at the limit of human capabilities, substantiates the problem of finding ways to improve the level of special physical fitness of highly qualified cyclists. Studies of the experience of implementing pedaling power control tools in training have shown their effectiveness in various leading cycling countries of the world, which can contribute to improving the level of special physical fitness of athletes in track and highway cycling. The effectiveness of the use of power measuring devices has shown its positive effect in preparing cyclists for highway competitions, which makes it possible to assume the great need to use these devices in the training of track cyclists. The purpose of this work is to develop and test models of pre-competition mesocycles based on pedaling power data. Work results. Experimental development and verification of the effectiveness of training programs with means of controlling pedaling power confirmed their informativeness in the process of conducting mesocycles and the need to use them in the preparation and development of training plans for track cyclists for the main competitions of the year. Conclusions. Based on the data of two macrocycles of 2021 and 2022 training of cyclists, the training and results of highly qualified track cyclists were compared. The intensity of each training session in two mesocycles was established. Structures of microcycles based on power and heart rate data are compared. The training of track cyclists, which includes not only training for power development, but also volume training, has proven to be effective.

Preceptorship in robotic colorectal surgery: experience from the Australian private sector

This article describes a post-fellowship preceptorship training program to train sub-specialty colorectal surgeons in gaining proficiency in robotic colorectal surgery using a dual-surgeon model in the Australian private sector. The Australian colorectal surgeon faces challenges in gaining robotic colorectal surgery proficiency with limited exposure and experience in the public setting where the majority of general and colorectal surgery training is currently conducted. This training model uses graded exposure with a range of simulation training, wet lab training, and clinical operative cases to progress through both competency and proficiency in robotic colorectal surgery which is mutually beneficial to surgeons and patients alike. Ongoing audit of practice has shown no adverse impacts.

New Method of Paired Comparison for Improved Observer Shortage Using Deep Learning Models

The aim of this study was to validate the potential of substituting an observer in a paired comparison with a deep-learning observer. Phantom images were obtained using computed tomography. Imaging conditions included a standard setting of 120 kVp and 200 mA, with tube current variations ranging from 160 mA, 120 mA, 80 mA, 40 mA, and 20 mA, resulting in six different imaging conditions. Fourteen radiologic technologists with >10 years of experience conducted pairwise comparisons using Ura's method. After training, VGG16 and VGG19 models were combined to form deep learning models, which were then evaluated for accuracy, recall, precision, specificity, and F1value. The validation results were used as the standard, and the results of the average degree of preference and significance tests between images were compared to the standard if the results of deep learning were incorporated. The average accuracy of the deep learning model was 82%, with a maximum difference of 0.13 from the standard regarding the average degree of preference, a minimum difference of 0, and an average difference of 0.05. Significant differences were observed in the test results when replacing human observers with AI counterparts for image pairs with tube currents of 160 mA vs. 120 mA and 200 mA vs. 160 mA. In paired comparisons with a limited phantom (7-point noise evaluation), the potential use of deep learning was suggested as one of the observers.

Transfer learning method for specific emitter identification based on pseudo‐labelling and meta‐learning

AbstractSpecific emitter identification (SEI) represents a prominent research direction within the electronic countermeasures domain aimed at discerning carrier identity attributes by analysing subtle radar characteristics. At present, most established SEI techniques assume that both the source and target domain (TD) data adhere to the same distribution. However, this assumption is invalidated by semantic drift which frequently occurs between TD and source domain (SD) samples owing to variations in the collection environment, equipment, and other factors. Considering the aforementioned challenges, this article introduces a transfer learning approach for SEI to leverage pseudo‐label integration within the framework of meta‐learning. This approach employs the bispectral perimeter integral for extracting emitter signal features to construct a feature extractor and basic learner based on CNN13. To label and filter the TD samples, the proposed approach utilises the multiple pseudo‐label serial filtering mechanism, which comprises positive and negative pseudo‐labelling strategies, label uncertainty prediction methods, and hard sample filtering strategies. Ultimately, to address algorithmic real‐time requirements, the labelled TD samples are integrated into the feature extractor and learner of the SD through meta‐learning to facilitate the transfer of TD features to the SD training model. Experimental validation conducted on a real radar dataset demonstrated that the proposed algorithm significantly enhances identification accuracy, exhibiting an improvement from approximately 50% to approximately 90%. Furthermore, the algorithm exhibits a short runtime and robust adaptability, effectively catering to the demands of practical scenarios.

A comprehensive literature mining and machine learning to decipher and predict effects of concrete materials on concrete corrosion of sewer pipe

ABSTRACT The concrete composition of different pipelines varies greatly, and a comprehensive analysis of the impact of concrete design on pipeline corrosion is required. Developed an integrated model based on Scikit-Optimize, XGBoost and SHAP to analyze concrete’s carbonation and sulfuric acid resistance from five aspects: binder type, supplementary cementitious material (SCM’s), aggregate, admixture and w/b. Choose Scikit-Optimize for parameter optimization. As an explanation model, the SHAP model can perform analysis well and explain the analysis basis of the XGBoost training model. The established model can accurately predict the corrosion effect (R2 >0.987) without overfitting. Binder should be maintained at 380 to 400 (kg/m3). Water/binder and coarse aggregate/binder are the two main factors that influence the binder, and they should be maintained at a ratio of around 0.4 to 0.6 and 2.5, respectively. Fly ash (FA) should be SCM’s used as at a range of 300 to 350 (kg/m3) to improve corrosion resistance.

Teaching Medical Procedural Skills for Performance

Procedures are a core element of medical professional practice. Today’s training approach was formulated in the mid-twentieth century based on a computer analogue of the brain. Despite minor modifications, the system has remained relatively unchanged for the past 70 years. It delivers competence. However, competence is not reliable performance. The inability to adapt to the variety of patients and variations in the performance environments, such as the operating room, results in patient morbidity and mortality. There is a need for changes in the development and training of medical procedural skills based on current theories of skill acquisition, movement theory, and motor control. Achieving optimal performance necessitates the ability to adapt through training in diverse patient and performance environments rather than merely imitating prescribed movements. We propose a novel model of training, the Constraints-Led Approach, which allows for robust training by altering the factors affecting skill acquisition and lifelong learning.

The issue of corporate mandatory standards in production improvement programmes

Purpose: This paper aims to explore and describe how companies manage the level of standardisation of improvement practices in a multisite context. It seeks to explain the managerial strategies applied to change the standardisation level in manufacturing companies with multiple production sites worldwide.Design/methodology/approach: This paper collects data through interviews, observations and company documents from a large multinational producing company and, specifically, from of the largest production sites in the company. The research design resembles a grounded theory approach by being reflexive and open to emerging themes. The standardisation strategy was analysed at a company that strived to increase the standardisation of problem-solving practices within about 20 production sites as part of their corporate lean programme.Findings: Several managerial tools were applied at the corporate level to increase the standardisation level of problem-solving practices, such as developing standards and a company-specific toolbox aligned with an in-house maturity model. In addition, deploying change leaders and global implementation targets enabled audits and progress. However, consequences at the production-site level became minor adaptations of standards, the design of training models as a "roll-out", and a resource-demanding implementation process.Originality/Value: This paper empirically demonstrates strategic tools that corporate management teams apply to influence the company's standardisation level of practices. The study describes the purpose and consequences of the design of the toolbox, maturity model, training model, and implementation targets, which aims to simplify the complex task of managing standardisation in a corporate group. By applying a knowledge-based view, four processes (i.e. adaptation, integration, upskilling, and learning) were identified to improve the management strategies in multisite contexts.

Exploration of Curriculum Ideology and Politics Integrating Medical Humanities Cultivation Model to Enhance Medical Core Values

Purpose: To explore the application effect of the cultivation mode of Curriculum Ideology and Political Integration of Medical Humanities to enhance the core values of medicine in the clinical internship stage of medical students. Methods: Students in the clinical internship stage of the First Affiliated Hospital of Baotou Medical College were selected, and a total of 156 students in one random class in each grade were taken as the observation group and received the integration cultivation mode; 148 students were taken as the control group and employed the traditional mode. The teaching effect of interns in the two groups was analyzed. Results: The teaching performance of students in both groups after clinical internship teaching was improved compared with that before admission; the teaching performance, teaching effect, teaching evaluation, and teaching satisfaction of the observation group were higher than that of the control group. Conclusion: The integration of the medical humanities training model with curriculum ideology and politics in the clinical internship of medical students is conducive to the improvement of teaching performance, teaching evaluation, teaching satisfaction of teachers and students, and the development and improvement of core values of medical students, which is of good value for teaching application.

Serum bile acid and unsaturated fatty acid profiles of non-alcoholic fatty liver disease in type 2 diabetic patients.

The understanding of bile acid (BA) and unsaturated fatty acid (UFA) profiles, as well as their dysregulation, remains elusive in individuals with type 2 diabetes mellitus (T2DM) coexisting with non-alcoholic fatty liver disease (NAFLD). Investigating these metabolites could offer valuable insights into the pathophy-siology of NAFLD in T2DM. To identify potential metabolite biomarkers capable of distinguishing between NAFLD and T2DM. A training model was developed involving 399 participants, comprising 113 healthy controls (HCs), 134 individuals with T2DM without NAFLD, and 152 individuals with T2DM and NAFLD. External validation encompassed 172 participants. NAFLD patients were divided based on liver fibrosis scores. The analytical approach employed univariate testing, orthogonal partial least squares-discriminant analysis, logistic regression, receiver operating characteristic curve analysis, and decision curve analysis to pinpoint and assess the diagnostic value of serum biomarkers. Compared to HCs, both T2DM and NAFLD groups exhibited diminished levels of specific BAs. In UFAs, particular acids exhibited a positive correlation with NAFLD risk in T2DM, while the ω-6:ω-3 UFA ratio demonstrated a negative correlation. Levels of α-linolenic acid and γ-linolenic acid were linked to significant liver fibrosis in NAFLD. The validation cohort substantiated the predictive efficacy of these biomarkers for assessing NAFLD risk in T2DM patients. This study underscores the connection between altered BA and UFA profiles and the presence of NAFLD in individuals with T2DM, proposing their potential as biomarkers in the pathogenesis of NAFLD.

Improving Professional Training of Surgeons Based on an Integrated Model and Program-Target Training

Improving professional surgical training consists of comparing the results of professional training of surgeons based on the traditional training model and the integrated training model. Also from a comparison of the results of postgraduate education of surgeons at the advanced training level based on the traditional training program and program-targeted training.

Development of a colonic endoscopic submucosal dissection training model that simulates respiratory movements.

Watch a video of this article.

EfficientTrain++: Generalized Curriculum Learning for Efficient Visual Backbone Training.

The superior performance of modern computer vision backbones (e.g., vision Transformers learned on ImageNet-1K/22K) usually comes with a costly training procedure. This study contributes to this issue by generalizing the idea of curriculum learning beyond its original formulation, i.e., training models using easier-to-harder data. Specifically, we reformulate the training curriculum as a soft-selection function, which uncovers progressively more difficult patterns within each example during training, instead of performing easier-to-harder sample selection. Our work is inspired by an intriguing observation on the learning dynamics of visual backbones: during the earlier stages of training, the model predominantly learns to recognize some 'easier-to-learn' discriminative patterns in the data. These patterns, when observed through frequency and spatial domains, incorporate lower-frequency components, and the natural image contents without distortion or data augmentation. Motivated by these findings, we propose a curriculum where the model always leverages all the training data at every learning stage, yet the exposure to the 'easier-to-learn' patterns of each example is initiated first, with harder patterns gradually introduced as training progresses. To implement this idea in a computationally efficient way, we introduce a cropping operation in the Fourier spectrum of the inputs, enabling the model to learn from only the lower-frequency components. Then we show that exposing the contents of natural images can be readily achieved by modulating the intensity of data augmentation. Finally, we integrate these two aspects and design curriculum learning schedules by proposing tailored searching algorithms. Moreover, we present useful techniques for deploying our approach efficiently in challenging practical scenarios, such as large-scale parallel training, and limited input/output or data pre-processing speed. The resulting method, EfficientTrain++, is simple, general, yet surprisingly effective. As an off-the-shelf approach, it reduces the training time of various popular models (e.g., ResNet, ConvNeXt, DeiT, PVT, Swin, CSWin, and CAFormer) by [Formula: see text] on ImageNet-1K/22K without sacrificing accuracy. It also demonstrates efficacy in self-supervised learning (e.g., MAE). Code is available at: https://github.com/LeapLabTHU/EfficientTrain.

Swimming exercise support respiratory functions of children with mild intellectual disabilities: controlled study

Children with mild intellectual disabilities have lower lung function than typically developing peers. The aim of this study is to analyse the effects of an 18-week medium-intensity swimming exercise on lung parameters of children with mild intellectual disabilities. In the study, participants were divided into experimental (n:8) and control group (n:8). An 18-week basic swimming exercise (60 min/3 days/18 weeks) was performed in the experimental group, while no practice was performed in the control group. Forced Vital Capacity (FVC), Tiffeneau Index (FEV1/FVC ratio), Vital Capacity (VC) and Maximal Voluntary Ventilation (MVV) parameters of the participants were measured. Pre-test and post-test comparisons of experimental and control groups were made with the Mann Whitney U-Test. In addition, analysis of lung values of experimental and control groups before and after exercise was performed with the Wilcoxon Signed Ranks Test. We also calculated the effect size (ES) results of the interactions between the variables, using partial eta squared and Cohen’s d. Statistically there was significant difference between pre-test and post-test scores of VC parameter in experimental group. Examining the effect sizes, experimental had a higher impact level in FEV1/FVC, VC, MVV (.571medium/0.647medium/.490small). The results indicate that the Swimming is more effective in FEV1/FVC, MVV and VC. In order to understand the effectiveness of swimming exercise on lung function, studies which include time and training models (duration, severity, intensity) should be conducted in the future.

Federated Learning With Only Positive Labels by Exploring Label Correlations.

Federated learning (FL) aims to collaboratively learn a model by using the data from multiple users under privacy constraints. In this article, we study the multilabel classification (MLC) problem under the FL setting, where trivial solution and extremely poor performance may be obtained, especially when only positive data with respect to a single class label is provided for each client. This issue can be addressed by adding a specially designed regularizer on the server side. Although effective sometimes, the label correlations are simply ignored and thus suboptimal performance may be obtained. Besides, it is expensive and unsafe to exchange user's private embeddings between server and clients frequently, especially when training model in the contrastive way. To remedy these drawbacks, we propose a novel and generic method termed federated averaging (FedAvg) by exploring label correlations (FedALCs). Specifically, FedALC estimates the label correlations in the class embedding learning for different label pairs and utilizes it to improve the model training. To further improve the safety and also reduce the communication overhead, we propose a variant to learn fixed class embedding for each client, so that the server and clients only need to exchange class embeddings once. Extensive experiments on multiple popular datasets demonstrate that our FedALC can significantly outperform the existing counterparts.