Training Model Research Articles

River Stream Flow Prediction through Advanced Machine Learning Models for Enhanced Accuracy

The Narmada River basin, a significant water resource in central India, plays a crucial role in supporting agricultural, industrial, and domestic water supply. Effective management of this basin requires accurate streamflow forecasting, which has become increasingly important. This study delves into streamflow forecasting using historical data from five major river stations, covering the upper reaches in the East and middle sections. The dataset spans from 1978 to 2020 and undergoes rigorous screening and preparation, including normalization using StandardScaler. The research adopts a comprehensive approach, developing models for training on 70% of historical data, validation on the most current 15%, and testing against future targets with another 15% of the data. To achieve precise predictions, a suite of machine learning models is employed, including CatBoost, LGBM (Light Gradient Boosting Machine), Random Forest, and XGBoost. Performance evaluation of these models relies on key indices such as mean squared error (MSE), mean absolute error (MAE), root mean square error (RMSE), root mean square percent error (RMSPE), normalized root mean squared error (NRMSE), and R-squared (R2). Notably, among these models, Random Forest emerges as the most robust for streamflow prediction, showcasing its effectiveness in handling the complexities of hydrological forecasting. This research contributes significantly to the field of streamflow forecasting in the Narmada River basin by providing insights into the performance of various machine learning models. The findings not only enhance our understanding of watershed dynamics but also highlight the pivotal role that machine learning can play in improving hydrological forecasting for sustainable watershed management.

Uncovering botnets in IoT sensor networks: a hybrid self-organizing maps approach

The integration of the internet of things (IoT) has revolutionized diverse industries, introducing interconnected devices and IoT sensor networks for improved data acquisition. However, this connectivity exposes IoT ecosystems to emerging threats, with botnets posing significant risks to security. This research aims to develop an innovative solution for detecting botnets in IoT sensor networks. Leveraging insights from existing research, the study focuses on designing a hybrid self-organization map (SOM) Approach that integrates lightweight deep learning (DL) techniques. The objective is to enhance detection accuracy by exploring various DL architectures. Proposed methodology aims to balance computational efficiency for resource-constrained IoT devices while improving the discriminatory power of the detection system. The study advancing IoT cybersecurity and addresses critical challenges in botnet detection within IoT sensor networks. The testing of the artificial neural networks (ANN) classifier involves three models, each represented based on parameters related to the construction of the training models. The most effective ANN achieves 86%, works on anomaly intrusion detection systems (AIDS).

Analysis and Prediction of Multiple Disease using Machine Learning

Abstract: This paper uses the machine learning methods for predicting diabetes, heart disease, and Parkinson's disease based on user-provided input data. Its primary goal is to enhance early disease detection and prompt medical intervention using personalized predictions. By training models on relevant datasets, accurate predictions are achieved across various diseases, expanding our understanding and predictive capabilities in healthcare. The research's broad scope contributes significantly to addressing multiple medical conditions, fostering a more holistic approach to healthcare delivery. Through data-driven insights, this tool aids in reducing medical costs by enabling early intervention and proactive management of medical conditions. Overall, this study highlights the potential of model in healthcare for disease pre-diction and underscores its role in advancing personalized medicine and improving clinical decision-making.

Athlete Performance Prediction Using Random Forest

Abstract: Fitness and health applications are increasingly being integrated into athletes' training routines, providing new opportunities for personalized performance optimization. This research paper investigates the utilization of smartwatch metrics and machine learning algorithms for predicting athlete performance, focusing on metrics such as power. Through a comprehensive analysis of collected data and employing advanced machine learning techniques, the study aims to provide insights into the predictive capabilities of smartwatch data in the realm of sports science. The introduction of a user-friendly athlete interface further enhances the accessibility and usability of this innovative technology. Smartwatch metrics combined with machine learning algorithms offer a potent toolset for predicting athlete performance. Smartwatches collect vast amounts of data on an athlete's biometrics, such as heart rate, activity levels, and more. Machine learning algorithms analyse this data to uncover patterns, correlations, and trends that are often imperceptible to human observation. By training models on historical data from athletes and their performances, these algorithms can make predictions about future performance, injury risk, optimal training schedules, and even suggest personalized strategies for improvement. This fusion of technology enables coaches and athletes to make data-driven decisions, optimize training regimens, and enhance overall performance while minimizing the risk of injury.

Domain Adaptation and Semantic Drawing Driven Sketch-to-Photo Retrieval using Collaborative Generative Representation Learning

Abstract: Sketch-based face recognition is an interesting task in vision and multimedia research yet it is quite challenging due to the great difference between face photos and sketches. In this paper we propose a novel approach for photo-sketch generation aiming to automatically transform face photos into detail-preserving personal sketches. Unlike the traditional models synthesizing sketches based on a dictionary of exemplars we develop a fully convolutional network to learn the end-to-end photosketch mapping. Our approach takes whole face photos as inputs and directly generates the corresponding sketch images with efficient inference and learning in which the architecture is stacked by only convolutional kernels of very small sizes. The exemplar-based method is most frequently used in face sketch synthesis because of its efficiency in representing the nonlinear mapping between face photos and sketches. However, the sketches synthesized by existing exemplar-based methods suffer from block artifacts and blur effects. In addition, most exemplar-based methods ignore the training sketches in the weight representation process. To improve synthesis performance, a novel joint training model is proposed in this paper, taking sketches into consideration. First, we construct the joint training photo and sketch by concatenating the original photo and its sketch with a high-pass filtered image of their corresponding sketch. Then, an offline random sampling strategy is adopted for each test photo patch to select the joint training photo and sketch patches in the neighboring region. Finally, a novel locality constraint is designed to calculate the reconstruction weight, allowing the synthesized sketches to have more detailed information

Developing professional capital through technology-enabled university-school-enterprise collaboration: an innovative model for C-STEAM preservice teacher education in the Greater Bay area

Purpose Preservice teacher (PST) professional development programs are crucial for cultivating high-quality STEAM teachers of the future, significantly impacting the quality of regional STEAM education. The Guangdong-Hong Kong-Macao Greater Bay Area, as a region of cross-border cooperation, integrates the resources and advantages of Guangdong, Hong Kong, and Macao, possessing rich cultural heritage and innovative capabilities. Transdisciplinary Education for Cultural Inheritance (C-STEAM) is an effective approach to promoting educational collaboration within the Greater Bay Area, facilitating the integration of both technological and humanities education. This study aims to develop a Technology-Enabled University-School-Enterprise (T-USE) collaborative education model and implement it in the Greater Bay Area, to explore its role as a support mechanism in professional development and its impact on C-STEAM PSTs' professional capital. Design/methodology/approach Adopting a qualitative methodology, the study interviewed PSTs who participated in a C-STEAM teacher education course under the T-USE model. Thematic coding is used to analyze their knowledge acquisition, interaction benefits with community members, and autonomous thinking and decision-making in theoretical learning and teaching practice. Findings The findings show that the T-USE model significantly enhanced the PSTs' human capital, including teaching beliefs, knowledge, and skills. In terms of social capital, PSTs benefited from collaboration with PST groups, university teaching teams, in-service teachers, and enterprises, though challenges such as varying levels of expertise among in-service teachers and occasional technical instability emerged. For decisional capital, the T-USE model provided opportunities for autonomous thinking and promoted teaching judgment skills through real teaching challenges and scenarios. Reflective practice activities also supported PSTs' professional growth. Originality/value This study reveals the effectiveness and internal mechanism of the T-USE model in C-STEAM PST training, offering significant theoretical and practical references for future PST education.

Application of 3D printing surgical training models in the preoperative assessment of robot-assisted partial nephrectomy

BackgroundTo explore the application effect of 3D printing surgical training models in the preoperative assessment of robot-assisted partial nephrectomy.MethodsEighty patients who underwent robot-assisted partial nephrectomy surgery between January 2022 and December 2023 were selected and divided into two groups according to the chronological order. The control group (n = 40) received preoperative assessment with verbal and video education from January 2022 to December 2022, while the observation group (n = 40) received preoperative assessment with 3D printing surgical training models combined with verbal and video education from January 2023 to December 2023. The preoperative anxiety, information demand score, and surgical awareness were compared between the two groups. The physiological stress indicators, including interleukin-6 (IL-6), angiotensin II (AT II), adrenocorticotropic hormone (ACTH), cortisol (Cor), mean arterial pressure (MAP), and heart rate (HR), were also measured at different time points before and after surgery.They were 6:00 am on the day before surgery (T0), 6:00 am on the day of the operation (T1), 6:00 am on the first day after the operation (T2), and 6:00 am on the third day after the operation (T3).The preparation rate before surgery was compared between the two groups.ResultsThe anxiety and surgical information demand scores were lower in the observation group than in the control group before anesthesia induction, and the difference was statistically significant (P < 0.001). Both groups had lower scores before anesthesia induction than before preoperative assessment, and the difference was statistically significant (P < 0.05). The physiological stress indicators at T1 time points were lower in the observation group than in the control group, and the difference was statistically significant (P < 0.05). The overall means of the physiological stress indicators differed significantly between the two groups (P < 0.001). Compared with the T0 time point, the T1, T2, and T3 time points in both groups were significantly lower, and the difference was statistically significant (P < 0.05). The surgical awareness and preparation rate before surgery were higher in the observation group than in the control group, and the difference was statistically significant (P < 0.05).ConclusionThe preoperative assessment mode using 3D printing surgical training models combined with verbal and video education can effectively reduce the psychological and physiological stress responses of surgical patients, improve their surgical awareness, and enhance the preparation rate before surgery.

Comprehensive study on the development of an automatic helmet violator detection system (AHVDS) using advanced machine learning techniques

The number of motorcycle accidents has increased rapidly due to different causes. Helmet is a one of the major safety equipment that helps to reduce the fatal accidents even though many of the motorcyclists are not used that. Currently many researchers have focused on traffic violator detection but they are yet to identify the violated person details as it requires high resolution cameras and expensive sensors for number plate extraction. This research article is focused on comprehensive study on the development of an automated helmet violation detection with number plate extraction in real time. This study divides the dataset into rider with helmet, rider without helmet, passenger with helmet and passenger without helmet classes, and four training model was created to detect helmet in rider and passengers. The proposed model was constructed in three stages to attain high accuracy with a vast amount of training dataset. In the first stage, the helmet was detected in the input image using the deep learning object-detection model .In the second stage, if the helmet identification confirmed in the input image then the entire motorcyclist image is segmented based on object segmentation model. In the third stage, number plate training model is deployed into the same object detection and object segmentation model to identify and segment the license plates in the motorcyclist image. This study primarily incorporates the Mask Region based Convolution Neural Networks (Mask R-CNN) for object detection and segmentation Further, it incorporates the three different deep neural networks like Dense block Network (DenseNet-121), Residual Network (ResNet-101) and Inception Residual Network architecture V2layers (Inception Res Net V2) to create accurate helmet violator detections with more extensive information. The experimental results indicates that this comprehensive helmet violator detection study improves the helmet detection accuracy by 6.2 %.In addition, the above three comprehensive proposed models are evaluated under Wilcoxon rank test for better sensitive information. Further, the less object detection loss (2 %) and significant computational time variation (69 %) indicates the performance of the comprehensive helmet violator detection system was good compared to existing study.

A Visual Survey of Tunnel Boring Machine (TBM) Performance in Tunneling Excavation: Mainstream Direction, Brief Review and Future Prospects

This study employs scientometric analysis to investigate the current trajectory of research on tunnel boring machine (TBM) performance and collaborative efforts. Utilizing software tools like Pajek 5.16 and VOSviewer 1.6.18, it scrutinizes literature from 2000 to 2021 sourced from the Web of Science (WOS). The findings illuminate TBM research as an interdisciplinary and intersectoral field attracting increasing national and institutional attention. Notable contributions from China, Iran, the United States, Turkey, and Australia underscore the global significance of TBM research. The recent upsurge in annual publications, primarily driven by Chinese research initiatives, reflects a renewed vigor in TBM exploration. Additionally, the paper presents a succinct evaluation of TBM advantages and drawbacks compared to conventional drill and blast methods, discussing key considerations in excavation methodology selection. Moreover, the study comprehensively reviews TBM performance prediction models, categorizing them into theoretical, empirical, and artificial intelligence-driven approaches. Finally, rooted in metaverse theory, the discourse delves into the immersive learning model and the architecture of a TBM metaverse. In the future, the immersive training and learning model diagram can be employed in scenarios such as employee training and the promotion of safety knowledge. Additionally, the TBM metaverse architecture can simulate, monitor, diagnose, predict, and control the organization, management, and service processes and behaviors of TBMs. This will enhance efficient collaboration across various aspects of the project production cycle. This forward-looking perspective anticipates future trends in TBM technology, emphasizing societal impact and enhancement of economic benefits.

Talent Training Optimization for Surveying and Mapping Majors in Higher Vocational Colleges Under the Background of Production-Education Integration

In recent years, the surveying and mapping geographic information industry has experienced rapid expansion. The application scenarios for “pan-surveying and mapping” have diversified, and advancements in aerial surveying equipment technology have raised the requirements for data application and expression skills among staff. This has created an urgent need for a team of highly skilled and practically capable surveying and mapping professionals. Higher vocational colleges, as the primary institutions for training surveying and mapping professionals, must align with industry development trends and proactively collaborate with enterprises. Building a training model that integrates production and education is essential to enhancing the quality of surveying and mapping professional training. This paper explores how to optimize the training mode for surveying and mapping professionals in higher vocational colleges within the context of this production and education integration.

The Influence of Models of Different Structures of Strength Training Classes in Horting on the Indicators of Bioimpedansometry

Topicality. In the modern system of training in mixed martial arts, in particular in the sport of horticulture, the search for optimal models of training programs aimed at improving strength capabilities and body composition of athletes is actively investigated. The Purpose of the Research. To study the peculiarities of the influence of different models of strength training in horting on the indicators of the body composition of athletes.. Methods. 60 athletes who have been involved in horting for 3–4 years took part in the study. 3 research groups were formed. The representatives of the examined groups used experimental training models during 3 months of strength training. Using the non-invasive method of bioimpedance measurement during all stages of the study, the nature of changes in parameters of body composition indicators was assessed. The Results. It was established that only representatives of the 2nd group showed a positive increase in parameters of active cellular body mass (ACM, %) by 3,7 % (p&lt;0,05) during the 3 months of the study compared to the initial data. A significant decrease in the parameters of this body composition indicator was found in the athletes of the other two groups. The greatest increase in BJM among the participants was found in athletes of group 3. At the same time, in representatives of group 1, the level of the studied indicator is almost half as low as compared to the results that were found in athletes of group 3 during the three months of the experiment. The greatest dynamics towards a decrease in the BM indicator (%) by 2,8 % (p&lt;0,05), over the entire period of the study, is observed in athletes of group 1. At the same time, in the representatives of the other two groups, during the entire period of the experiment, we observe identical dynamics of the investigated body composition indicator, the level of which differs by only 0,7 % from the results demonstrated by the athletes of the 1st group. Findings. The use in the process of strength training in horting of variations of exercise complexes effective in strength fitness on simulators under the conditions of the principle of “premature fatigue” of agonist muscles due to exercises of an isolating nature, allows to maximally influence the parameters of body composition compared to the results that were found in other conditions of training activities.

MODELS OF PROFESSIONAL DEVELOPMENT FOR BIOLOGY TEACHERS IN THE CONTEXT OF SOCIAL CHANGES

This article aims to explore the impact of social changes on the professional development of biology teachers. The purpose of the study is the analysis of modern challenges in the field of education and the development of a model of professional growth of a biology teacher. The following methods are employed in the study: analysis and synthesis of literature on the research topic to determine the current state of the investigated problem, systematic analysis, and synthesis of obtained information to identify the main models of professional development for biology teachers. The regulatory framework ensuring conditions for the professional development of teachers has been analyzed. Modern economic and technological changes, socio-cultural and political transformations, dynamics of interethnic and cultural relations in Ukraine, high population mobility, and migration processes affect the education system and the professional development of teachers. In the scientific discourse, among effective models of professional development for teachers, scholars distinguish: concepts of teaching and development of teachers' professional abilities based on knowledge (knowledge for practice, knowledge in practice, personal knowledge); models of professional development for educators (training model, award-bearing model, deficit model, cascading model, standards-based model, coaching/mentoring model, community of practice model, action research model, transformational model); competency, cognitive, human-centered models; models of applied science, craftsmanship, reflective, apprentice-expert model, rationalist model, case studies model, cascading. The author proposes an original model of professional development, which involves six stages: analysis of contemporary challenges, professional training based on teachers' needs, creation of an individual development plan, integration of innovative technologies into professional activities, self-evaluation, and the establishment and support of professional communities. The main research results indicate the need to reconsider traditional approaches to teaching biology and emphasize the use of modern technologies in the professional development of teachers for successful implementation of changes in the educational environment.

Pulmo-protection of long-term swimming exercise via improving insulin sensitivity in monocrotaline-induced pulmonary hypertensive rats

Exercise has been recognized as an effective intervention in the treatment of pulmonary arterial hypertension (PAH), supported by numerous studies. However, the precise effects of exercise on pulmonary function remain to be fully elucidated. In this study, using a rat model of swimming exercise training and monocrotaline-induced PAH, we aimed to explore its impact on pulmonary morphology and function. Our investigations revealed that MCT-treated rats exhibited augmented mean pulmonary arterial pressure (MPAP) and pulmonary vascular remodeling, which can be attenuated by 4 weeks of swimming exercise training (60 min/day, 5 days/week). Notably, MCT-treated rats showed impaired pulmonary function, as manifested by decreased tidal volume and dynamic compliance, which were reversed by exercise training. Assessment of pulmonary substrate in PAH rats indicated a prominent pro-inflammatory substrate, evidenced by macrophage accumulation through quantitative immunohistological analysis of macrophage-like cell expression (CD68), and extracellular matrix remodeling, evaluated by Masson staining. Importantly, both the pro-inflammatory substrate and extracellular matrix remodeling were ameliorated by swimming exercise training. Additionally, serum biochemical analysis demonstrated elevated levels of low-density lipoprotein cholesterol and Apolipoprotein B following MCT treatment, which were reduced with exercise intervention. Moreover, exercise enhanced systemic insulin sensitivity in both MCT-treated and untreated rats. Notably, MCT and exercise treatment both decreased fasting blood glucose (FBG) levels in rats, whereas exercise training reinstated FBG levels to normal in MCT-treated rats. In summary, our study suggests that swimming exercise confers a pulmonary protective effect in MCT-induced PAH rats, highlighting the potential importance of exercise-based rehabilitation in the management of PAH.

Hot rolled steel surface defect detection and classification using an automatic ensemble approach

This study introduces an ensemble-based Deep Neural Network (DNN) model for detecting defects on steel surfaces. The method suggested in this study classifies steel surface conditions into six possible fault categories, namely, crazing, inclusion, rolled in, pitted surface, scratches, and patches. The images undergo preprocessing and extraction of features in spatial and frequency domains using image segmentation techniques such as grey level difference method (GLDM), fast Fourier Transform (FFT), grey level co-occurrence matrix (GLCM), texture analysis and discrete wavelet transform (DWT). The ensembling of image features into a fused feature pool is carried out after the preprocessing of input images that are provided as input to a light-weight neural network model for training and testing. The performance of the model is comprehensively evaluated via an ablation study both before and after ensembling. In addition, the model capability is effectively analyzed using receiver operating characteristics (ROC) curve, confusion matrix from which classification accuracy of the model could be obtained and other parameters including precision and f1-score. It was observed that the proposed deep learning network presents phenomenally high accuracy of 99.72% for detection and classification of steel surface faults. This result was found to be superior when compared with the performance of the same neural network over each feature type individually. This study also compares the classification results of the model built based on the ensembled feature set with the results of various other classification approaches available in literature. The ensemble-based model could potentially be integrated into existing inspection systems for real-time, efficient and robust condition monitoring of steel surfaces.

Reform and Innovation of the Training Method of Outstanding Software Engineers under the "Six Excellence and One Top-Notch" Plan

In the face of scientific and technological revolution and industrial transformation in the new era, China's engineering education is facing new requirements and challenges. In order to support the new development strategy of engineering education in China, the "Excellence Plan 2.0" provides an overall idea for actively exploring the engineering education system with Chinese characteristics, building a new talent training model, implementing diversified teaching methods, and building a high-level engineering education teaching team. In order to promote China's engineering education standards to the world and promote the improvement of China's engineering education status in the world, it is urgent to cultivate interdisciplinary talents with engineering practice ability and innovation ability. In view of the current situation and the main problems of the engineering talent training system in China's colleges and universities, this paper makes a preliminary analysis and reflection on the reform and innovation of outstanding engineer training from the aspects of practical teaching, reform and innovation of talent training mode, talent team construction, and school-enterprise cooperation.

Semi-supervised skin cancer diagnosis based on self-feedback threshold focal learning

Worldwide, skin cancer prevalence necessitates accurate diagnosis to alleviate public health burdens. Although the application of artificial intelligence in image analysis and pattern recognition has improved the accuracy and efficiency of early skin cancer diagnosis, existing supervised learning methods are limited due to their reliance on a large amount of labeled data. To overcome the limitations of data labeling and enhance the performance of diagnostic models, this study proposes a semi-supervised skin cancer diagnostic model based on Self-feedback Threshold Focal Learning (STFL), capable of utilizing partial labeled and a large scale of unlabeled medical images for training models in unseen scenarios. The proposed model dynamically adjusts the selection threshold of unlabeled samples during training, effectively filtering reliable unlabeled samples and using focal learning to mitigate the impact of class imbalance in further training. The study is experimentally validated on the HAM10000 dataset, which includes images of various types of skin lesions, with experiments conducted across different scales of labeled samples. With just 500 annotated samples, the model demonstrates robust performance (0.77 accuracy, 0.6408 Kappa, 0.77 recall, 0.7426 precision, and 0.7462 F1-score), showcasing its efficiency with limited labeled data. Further, comprehensive testing validates the semi-supervised model’s significant advancements in diagnostic accuracy and efficiency, underscoring the value of integrating unlabeled data. This model offers a new perspective on medical image processing and contributes robust scientific support for the early diagnosis and treatment of skin cancer.

ChatGPT-4 as a journalist: Whose perspectives is it reproducing?

The rapid emergence of generative AI models in the media sector demands a critical examination of the narratives these models produce, particularly in relation to sensitive topics, such as politics, racism, immigration, public health, gender and violence, among others. The ease with which generative AI can produce narratives on sensitive topics raises concerns about potential harms, such as amplifying biases or spreading misinformation. Our study juxtaposes the content generated by a state-of-the-art generative AI, specifically ChatGPT-4, with actual articles from leading UK media outlets on the topic of immigration. Our specific case study focusses on the representation of Eastern European Roma migrants in the context of the 2016 UK Referendum on EU membership. Through a comparative critical discourse analysis, we uncover patterns of representation, inherent biases and potential discrepancies in representation between AI-generated narratives and mainstream media discourse with different political views. Preliminary findings suggest that ChatGPT-4 exhibits a remarkable degree of objectivity in its reporting and demonstrates heightened racial awareness in the content it produces. Moreover, it appears to consistently prioritise factual accuracy over sensationalism. All these features set it apart from right-wing media articles in our sample. This is further evidenced by the fact that, in most instances, ChatGPT-4 refrains from generating text or does so only after considerable adjustments when prompted with headlines that the model deems inflammatory. While these features can be attributed to the model’s diverse training data and model architecture, the findings invite further examination to determine the full scope of ChatGPT-4’s capabilities and its potential shortcomings in representing the full spectrum of social and political perspectives prevalent in society.

Small-sample damage detection of bleacher structure based on GAN and MSS-CNN models

The damage scales and forms of bleacher structure are diverse, and the training by using neural network models may be inadequate when the data sample is limited, resulting in challenges such as overfitting or the inability to generalize new damage scenarios. In order to address the issue of damage detection in bleacher structures with small samples, this paper proposes a multi-scale stride convolutional neural network (MSS-CNN) model. It is trained as a generator and discriminator within a generative adversarial network (GAN) framework. By utilizing GAN to generate data and integrating real data with generated data, the mixed data is input into the MSS-CNN model for training, ultimately yielding damage detection results. In order to validate the effectiveness of this approach, a series of experimental studies are conducted by using a bleacher simulator at Qatar University as the research subject. Furthermore, the model is compared with ResNet, multi-layer perceptron, and support vector machine under identical experimental conditions by comparing real and mixed data. The experimental results consistently demonstrate the superior performance of the MSS-CNN model across multiple experiments. This paper presents a fresh research approach and perspective for addressing the challenge of small-sample damage detection in bleacher structures.

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.