Training Progress Research Articles

National Palliative Care Strategy in a Conflict Affected Country: A Jordanian Demonstration Project

ContextPalliative care (PC) integration is vital, as endorsed by the World Health Organization. Yet, Jordan, a Middle Eastern country with limited resources, faces ongoing challenges despite efforts to improve palliative and home care. Establishing a national PC strategic framework, with government and stakeholder consensus, is essential for ensuring universal access to high-quality palliative care. However, processes for achieving this, particularly in the Middle East, are underreported. ObjectivesThis study delineates the process of developing Jordan's National Palliative and Home Care Strategic Framework, reflecting on its five-year impact. Additionally, it identifies barriers to PC progress in Jordan and offers recommendations to stakeholders. MethodsWe conducted a rapid review and analyzed reports, minutes, meetings, and publications. The sequential framework development involved content and situational analysis, expert review, transparent expert consultation, multistage consensus procedures, and high-level advocacy meetings. ResultsThe National Palliative and Home Care Strategic Framework encompasses six domains: 1) policy, 2) finance, 3) service delivery, 4) opioid access, 5) capacity building, and 6) information, research, monitoring, and evaluation. Government endorsement in April 2018 ignited national engagement, driving policy, service delivery, workforce development, education, training, and research progress. Nonetheless, workforce shortages, limited opioid access, and inadequate funding persist as barriers. ConclusionJordan's collaborative development of the inaugural National Palliative and Home Care Strategic Framework, endorsed by the government and stakeholders, provides a comprehensive roadmap for PC advancement. While it promises improved services, effective solutions to workforce and opioid access issues are crucial for successful implementation.

Novel training concept utilizing a real world-like EP simulator

Abstract Background Training in the field of invasive cardiac electrophysiology (EP) is challenging and requires intellectual skills related to the interpretation of intracardiac electrograms. Classic approach to education of young electrophysiologists focused solely on theoretical knowledge and overseen procedures in patients as no real life like simulation of complete diagnostic and therapeutic EP studies was available. Purpose The purpose of this study was to assess a novel tool for EP training based on fully interactive, online simulator providing real clinical experience to the users. Methods EP Simulator replicates EP recording system routinely used in cardiac electrophysiology laboratories and therefore enable easier transition from training to clinical situation. Users are able to access a fully functional system with simulated surface electrocardiogram and intracardiac electrograms mimicking signals recorded by a catheter located at high right atrium, His bundle, coronary sinus, and right ventricular apex. Normal electrophysiology study and the following cardiac arrhythmias are available for simulation: sick sinus syndrome, AV conduction blocks, atrial ectopy, atrial flutter and fibrillation, atrio-ventricular and atrio-ventricular nodal reentrant tachycardia and ventricular tachycardias. Evaluation by 40 electrophysiologists (mean experience 8±6y) from 16 countries was collected via online questionnaire. Results Realism of ECG signals was described as excellent or very good by 90% of responders, realism of intracardial signals by 82.5%. Realism of signal interactions and user experience was judged as excellent or very good by 75% and 70%, respectively. 100% of users agree definitely or mostly that EP Simulator helps to translate theoretical into practical knowledge. 97.5% of responders would include it in EP training programs as it is extremely or very useful for training purposes in opinion of 87.5%. 72.5% of responders think that training on EP simulator can potentially reduce rate of complications. In 87.5% the overall experience of EP simulator was completely or mostly satisfying and it would be recommended by 100% of respondents. Conclusion EP simulator is a feasible tool for training of young electrophysiologist and it may be potentially included in cardiologist curriculum. We should particularly emphasize the positive respondents’ evaluation of EP simulator overall realism. Further systematic evaluation of the training progress is needed.

Abstract 15657: Stroke Survivors’ and Caregivers’ Experience of Using Game-Based Virtual Reality Training Modules Designed for Promoting Social Participation

Introduction: Virtual reality (VR) technology may provide a promising alternative to promote stroke survivors’ social participation. Three innovative social participation-focused, game-based VR training modules for community-dwelling stroke survivors were developed. Aim: This study sought to understand the survivors’ and caregivers’ perceptions of the usefulness of VR-based interventions in supporting post-stroke recovery and their experience of using the specially developed VR modules. Methods: A qualitative study with semi-structured, in-depth, individual interviews was conducted with a convenience sample of 10 stroke survivor-caregiver dyads (n=20). Each survivor and caregiver attended a first interview about their perceptions and expectations of VR-based interventions. They were then invited to experience all games of the VR modules and were interviewed again about their experience and satisfaction with using the VR modules, and suggestions on areas for improvement. All interview data were transcribed verbatim and analyzed thematically. Results: Most of the survivors (mean age=60.80; SD=6.71) were female (70%), had an ischemic stroke (50%), and walked with a stick (70%). The mean years after stroke was 4.80 (SD=4.16). Most of the caregivers were female (80%) and had a mean age of 56.20 years (SD=7.73). All survivors had no prior VR experience. Three themes were identified: A modern strategy instilling enjoyment into the regular physical training; confidence boosting for reengaging into life situations and outdoor activities; and tailored modifications needed for reviewing training progress, supporting more social interactions and controller-free experience. They suggested enabling availability of the modules in survivors’ home or community settings for increasing the convenience to access the modules. Conclusions: The findings showed that the new VR training modules are acceptable to the survivors. Feedback elicited informed refinement of the modules and the usefulness of the modules in improving social participation of survivors will be evaluated in a large-scale randomized controlled trial.

Upper limb biomechanical differences in volleyball spikes among young female players

The performance assessment of relevant biomechanical factors is essential for appropriate age-related training progression in volleyball spike (VS). This study aimed to investigate the differences in upper limbs’ (a) range of motion (ROM), (b) angular velocities, and (c) muscular activity during VS among under 16- (U-16), under 18- (U-18), and under 20- (U-20) year-old female volleyball players. The kinematic variables of the spike performances were recorded using six optoelectronic cameras (Vicon Motion system, UK). Additionally, the wireless Trigno system (Delsys, USA) was used to record the electromyographic signals. The one-way ANOVA (1d SPM analysis) exhibited a significantly greater wrist internal rotation ( p < .001), wrist adduction velocity (p < .001), and palmaris longus activity ( p < .001) in the acceleration phase in U-20, compared to the U-16 group. Additionally, the U-20 and the U-18 players portrayed a significantly greater shoulder external rotation ( p < .001) and internal angular velocity ( p = .035) relative to the U-16 group. A significantly greater pectoralis major (p < .001), biceps brachii ( p < .001), and anterior deltoid ( p < .001) activity was observed in the U-20 group, compared to younger groups. This study suggests that older players employ more upper limb muscular activities during spike with the presence of blocks, which enables them to hit the ball with greater velocities and direct the ball in their desired direction through increased wrist ROMs.

GPARS: Graph predictive algorithm for efficient resource scheduling in heterogeneous GPU clusters

Efficient resource scheduling in heterogeneous graphics processing unit (GPU) clusters are critical for maximizing system performance and optimizing resource utilization. However, prior research in resource scheduling algorithms typically employed machine learning (ML) algorithms to estimate job durations or GPU utilization in the cluster based on training progress and task speed. Regrettably, these studies often overlooked the performance variations among different GPU types within these clusters, as well as the presence of spatiotemporal correlations among jobs. To address these limitations, this paper introduces the graph predictive algorithm for efficient resource scheduling (GPARS) designed specifically for heterogeneous clusters. GPARS leverages spatiotemporal correlations among jobs and utilizes graph attention networks (GANs) for precise job duration prediction. Building upon the prediction results, we develop a dynamic objective function to allocate suitable GPU types for newly submitted jobs. By conducting a comprehensive analysis of Alibaba’s heterogeneous GPU cluster, we delve into the impact of GPU capacity and type on job completion time (JCT) and resource utilization. Our evaluation, using real traces from Alibaba and Philly, substantiates the effectiveness of GPARS. It achieves a remarkable 10.29% reduction in waiting time and an average improvement of 7.47% in resource utilization compared to the original scheduling method. These findings underscore GPARS’s superior performance in enhancing resource scheduling within heterogeneous GPU clusters.

Phoneme Training for Adult Cochlear Implant Users: A Review of the Literature and Study Protocol.

This study describes a protocol for a novel individualized phoneme training program for adult cochlear implant (CI) users, based on individual phoneme confusion errors. The protocol is underpinned by a literature review on phoneme training and a focus group with adult CI users. After a literature search, five studies were included for review and evaluation of quality and level of evidence. A focus group with experienced adult CI users (n = 7) was then conducted to gain insights into their experiences of auditory training post-implantation and recommendations for future training programs. The knowledge gained from the literature review and focus group was used as the foundation for a novel, individualized phoneme training program for adult CI users, for which the protocol is described in this study. A review of the literature shows that phoneme training in adult CI users has variable outcomes for on-task and off-task measures. Overall, the concept of individualized training relates to adaptive difficulty within training tasks and not to tailoring training content to participants' individual needs, as indicated by clinical outcomes. The focus group revealed that participants want to be able to track their training progress, have training content tailored to their individual needs, and expressed a preference for shorter training sessions. Using learnings from a literature review and focus group, this study describes a protocol for a novel, individualized phoneme training program for adult CI users. Study findings from this phoneme training program will be disseminated when available. https://doi.org/10.23641/asha.24392863.

Barriers to resilience among healthcare staff in Ireland in the post-austerity period

Abstract Background The health workforce is the motor of any health system. Although the focus of health systems has been on increasing the availability of health workers, less attention has been devoted to exploring the factors associated with the satisfaction/dissatisfaction of the existing workforce or how to improve it. Objective: To assess job satisfaction among Irish healthcare workers and to model the association between psychological, sociodemographic, and work-related factors and the satisfaction of the health staff. Methodology A secondary data analysis and subgroup analysis were carried out utilising the data collected in the waves (2016, 2018) of the Irish Staff Survey “Your Opinion Counts”. Descriptive statistics were used to describe the baseline characteristics of the participants. Post-stratification weights were generated to decrease the bias associated with the representativeness of the sample. Ordinal logistic regression was utilised for determining associations between job satisfaction and work-related, sociodemographic, and psychological state variables. Results Just over half of the health staff were satisfied with their organisation (52.1%). At country and organisational levels, the communication within the organisation and the possibilities of training and career progression showed the greatest percentage of dissatisfaction within the participants. At the individual level, being a woman, experiencing bullying or discrimination, working in the National Ambulance Service or being a doctor/dentist were the variables highly correlated with dissatisfaction of health staff. Conclusions The results highlighted the dissatisfaction experienced by many healthcare workers. Therefore, understanding and addressing their concerns is crucial for policy makers developing national strategies and knowledge-based interventions.

PLACE Dropout: A Progressive Layer-wise and Channel-wise Dropout for Domain Generalization

Domain generalization (DG) aims to learn a generic model from multiple observed source domains that generalizes well to arbitrary unseen target domains without further training. The major challenge in DG is that the model inevitably faces a severe overfitting issue due to the domain gap between source and target domains. To mitigate this problem, some dropout-based methods have been proposed to resist overfitting by discarding part of the representation of the intermediate layers. However, we observe that most of these methods only conduct the dropout operation in some specific layers, leading to an insufficient regularization effect on the model. We argue that applying dropout at multiple layers can produce stronger regularization effects, which could alleviate the overfitting problem on source domains more adequately than previous layer-specific dropout methods. In this article, we develop a novel layer-wise and channel-wise dropout for DG, which randomly selects one layer and then randomly selects its channels to conduct dropout. Particularly, the proposed method can generate a variety of data variants to better deal with the overfitting issue. We also provide theoretical analysis for our dropout method and prove that it can effectively reduce the generalization error bound. Besides, we leverage the progressive scheme to increase the dropout ratio with the training progress, which can gradually boost the difficulty of training the model to enhance its robustness. Extensive experiments on three standard benchmark datasets have demonstrated that our method outperforms several state-of-the-art DG methods. Our code is available at https://github.com/lingeringlight/PLACEdropout .

Threat Detection Model for WLAN of Simulated Data Using Deep Convolutional Neural Network

Security identification solutions against WLAN network attacks according to straightforward digital detectors, such as SSID, IP addresses, and MAC addresses, are not efficient in identifying such hacking or router impersonation. These detectors can be simply mocked. Therefore, a further protected key uses new information by combining these simple digital identifiers with an RF signature of the radio link. In this work, a design of a convolutional neural network (CNN) based on fingerprinting radio frequency (RF) is developed with computer-generated data. The developed CNN is trained with beacon frames of a wireless local area network (WLAN) that is simulated as a result of identified and unidentified router nodes of fingerprinting RF. The proposed CNN is able to detect router impersonators by comparing the data pair of the MAC address and RF signature of the received signal from the known and unknown routers. ADAM optimizer, which is the extended version of stochastic gradient descent, has been used with a developed deep learning convolutional neural network containing three fully connected and two convolutional layers. According to the training progress graphic, the network converges to around 100% accuracy within the first epoch, which indicates that the developed architecture was efficient in detecting router impersonations.

Attitude towards motivation and formation of healthy fundamentals of life in modern youth

The article reveals the essence and content of the concept of "motivation for a healthy lifestyle", outlines the need and substantiates the main factors, factors and approaches to the formation of motivation and key competence for leading a healthy lifestyle. The mechanisms of influence on the state of health and the absence or presence of healthy motivation of the population as a whole and each person separately are revealed. Physical education and sports help to strengthen health, improve functional and motor capabilities. They are an integral part of a healthy lifestyle and lead to harmonious physical and spiritual development of the individual. Systematic physical education and sports preserve youth, health, and longevity, which is accompanied by creative work growth. A conscious attitude to physical culture and sports is of great importance. The principle of constant motivation, an active attitude towards arbitrary self-improvement and self-development in sports should be laid at the heart of any physical education program. Motivation to do sports is the most important aspect, in many ways it is not inferior in importance to the training process. It reflects interest in classes, an active and conscious attitude to sports and is one of the factors of successful mastering of the training program and progress. “Motivation is the urge to action; a psycho-physiological process that controls human behavior, capable of determining its direction, organization, activity and stability; a person’s ability to satisfy his own needs”.

Sparse Pressure-Based Machine Learning Approach for Aerodynamic Loads Estimation During Gust Encounters

Estimation of aerodynamic loads is a significant challenge in complex gusty environments due to the associated complexities of flow separation and strong nonlinearities. In this study, we explore the practical feasibility of multilayer perceptron (MLP) for estimating aerodynamic loads in gusts, when confounded by noisy and spatially distributed sparse surface pressure measurements. As a demonstration, a nonslender delta wing experiencing various gusts with different initial and final conditions is considered. Time-resolved lift and drag, and spatially distributed sparse surface pressure measurements are collected in a towing-tank facility. The nonlinear MLP model is used to estimate gust scenarios that are unseen in training progress. A filtering process allows us to examine the fluctuation of the dynamic response from the pressure measurements on the MLP. Estimation results show that the MLP model is able to capture the relationship between surface pressure and aerodynamic loads with a minimum quantity of learning samples, delivering accurate estimations, despite the slightly large errors for the cases at the boundary of the datasets. The results also indicate that the dynamic response of the pressure measurements has an influence on the learning of MLP. We further utilize gradient maps to perform a sensitivity analysis, so as to evaluate the contribution of the pressure data to the estimation of gust loads. This study reveals the significant contribution of the sensors located near the leading edge and at the nose of the delta wing. Our findings suggest the potential for an efficient sensor deployment strategy in data-driven aerodynamic load estimation.

Generating a Prioritized List of Operative Procedures for Simulation-based Assessment of General Surgery Trainees Through Consensus.

To develop appropriate content for high-stakes simulation-based assessments of operative competence in general surgery training through consensus. Valid methods of summative operative competence assessment are required by competency-based training programs in surgery. An online Delphi consensus study was conducted. Procedures were derived from the competency expectations outlined by the Joint Committee on Surgical Training Curriculum 2021, and subsequent brainstorming. Procedures were rated according to their perceived importance, perceived procedural risk, how frequently they are performed, and simualtion feasibility by a purposive sample of 30 surgical trainers and a 5-person steering group. A modified Copenhagen Academy for Medical Education and Simulation Needs Assessment Formula was applied to the generated data to produce ranked procedural lists, which were returned to participants for re-prioritization. Prioritized lists were generated for simulation-based operative competence assessments at 2 key stages of training; the end of 'phase 2' prior to the development of a sub-specialty interest, and the end of 'phase 3', that is, end-of-training certification. A total of 21 and 16 procedures were deemed suitable for assessments at each of these stages, respectively. This study describes a national needs assessment approach to content generation for simulation-based assessments of operative competence in general surgery using Delphi consensus methodology. The prioritized procedural lists generated by this study can be used to further develop operative skill assessments for use in high-stakes scenarios, such as trainee progression, entrustment, and end-of-training certification, before subsequent validity testing.

Comparison of Pre- and Postoperative Motor-proprioceptive Abilities in Patients with Gonarthrosis.

Both surgeons and patients want to achieve a high level of satisfaction and the best possible functional results within a short time after knee TEP surgery. By using a tool that digitally records various measurement parameters of balance and motor function preoperatively and postoperatively on a mobile basis and with little time expenditure, progressive results can be compared. Individual factors can thus be determined and these can influence the progress in regeneration and training progress perioperatively.In a prospective study, 100 patients before and 66 patients after installation of a cement-retained knee TEP were evaluated for the following parameters: balance, maximum strength, and power. All measurements were performed with the KMP measurement platform from MotoSana. The second measurements were performed in each case after a standardised follow-up treatment.It was shown that there are significant relationships between personal factors such as age, height, body weight and with baseline values and performance measures: maximum strength and power. Furthermore, it was shown that postoperative improvement could be achieved for the most part around balance support. All patients who previously had to hold on with one hand or both hands no longer needed support after surgery to maintain the single-leg stance for the specified time of 15s. For a more detailed analysis of the balance parameters, the samples were adjusted and only the patients who did not hold on for support pre- and postoperatively were counted. In patients with low and medium initial stance, the sway area increased at the second measurement session, and in patients with large sway areas, it decreased, and the stance became more stable. In the area of maximum strength and power, patients with high baseline values still had higher values after AHB compared with the other patients, but lower values compared with their own baseline values.Patients who already had very good motor skills before surgery were able to achieve a greater increase in motor skills compared to the weaker group. However, all patients failed to reach their preoperative baseline values after completion of the AHB. Deficits in balance were still detectable in all groups. By using the presented force plate, measurement-based coordinated rehabilitation procedures are possible during and after completion of the AHB. Rehabilitation with individualised improvement of balance and motor function could be expected to prevent dissatisfaction after knee arthroplasty, e.g. due to muscular imbalance in femoropatellar pain syndromes.

Identity-Aware Contrastive Knowledge Distillation for Facial Attribute Recognition

Facial attribute recognition (FAR) is an important and yet challenging multi-label learning task in computer vision. Existing FAR methods have achieved promising performance with the development of deep learning. However, they usually suffer from prohibitive computational and memory costs. In this paper, we propose an identity-aware contrastive knowledge distillation method, termed ICKD, to compress the FAR model. A nonlinear weight-sharing mapping (NWSM) mechanism is firstly designed to avoid the difficulty of directly matching features of the teacher and student networks due to the lower representation ability of the student network. Furthermore, an identity-aware contrastive distillation (ICD) loss is employed to guide the student network to effectively learn the mutual relations between samples with multiple attributes. In addition, an adjustable ladder distillation (ALD) loss is developed to automatically adjust the importance of different distillation points with the progress of training. Extensive experiments demonstrate that our method can significantly improve the performance of student networks and outperforms the existing FAR methods on the public challenging datasets.

Defect driven physics-informed neural network framework for fatigue life prediction of additively manufactured materials.

Additive manufacturing (AM) has attracted many attentions because of its design freedom and rapid manufacturing; however, it is still limited in actual application due to the existing defects. In particular, various defect features have been proved to affect the fatigue performance of components and lead to fatigue scatter. In order to properly assess the influences of these defect features, a defect driven physics-informed neural network (PiNN) is developed. By embedding the critical defects information into loss functions, the defect driven PiNN is enhanced to capture physical information during training progress. The results of fatigue life prediction for different AM materials show that the proposed PiNN effectively improves the generalization ability under small samples condition. Compared with the fracture mechanics-based PiNN, the proposed PiNN provides physically consistent and higher accuracy without depending on the choice of fracture mechanics-based model. Moreover, this work provides a scalable framework being able to integrate more prior knowledge into the proposed PiNN. This article is part of the theme issue 'Physics-informed machine learning and its structural integrity applications (Part 1)'.

Reflections on theses of experimental reports for the master of translation and interpreting in China (2012–2022)

PurposeThe number of universities offering Master of Translation and Interpreting (MTI) in China has increased to 316 in 15 years. This paper aims to take a closer look at the production patterns of experimental report theses in terms of total number, distribution across universities, supervision model and research content, reflects on the problems and suggests improvements, provides a reference for MTI education in China and beyond.Design/methodology/approachThis paper presents a bibliometric analysis of the published final theses of experimental reports for the MTI in China between the years 2012 and 2022, the period during which this type of thesis was produced, to identify the production patterns of these theses.FindingsThe number of experimental reports published by the nine leading universities accounts for 80% of the total 296 papers. The uneven development is also reflected in the supervision model and research content. Most universities can structure the main content of theses according to the suggestions of the MTI Guidance Training Outline. However, there are still deficiencies in the areas of experimental design, experimental validity, research questions and academic standards.Originality/valueThis paper reflects on the problems of MTI theses of experimental reports. Also, suggestions are made for linking top-level design and university education, for joint progress in faculty development and talent training and for integrating industrial aspects and international visions, to provide a reference for MTI education in China and beyond.

Exploring Laparoscopic Surgery Training Opportunities in the College of Surgeons of East, Central, and Southern Africa region

The resource-limited environment in Sub-Saharan countries, with a lack of expert trainers, impedes the progress of laparoscopic training. This study aimed to identify the opportunities and limitations of laparoscopic surgery training in the College of Surgeons of East, Central, and Southern Africa (COSECSA) countries. A multicountry online survey was conducted from January 2021 to October 2021 in COSECSA-accredited training hospitals within 16 countries. Available resources and challenges faced in order to set up well-structured laparoscopic training programs were explored. Ninety-four surgeons answered the questionnaire. The average resources reported per hospital were 3 trained laparoscopic surgeons, 2 laparoscopic towers, and 2 sets of laparoscopic instruments. The training of the majority of these surgeons has been in local institutions (53%), a further 37% within African countries and only 10% outside Africa. Approximately 45% of them declared that laparoscopic modules were planned within the University Curricula, while only 18% of surgeons recognized that laparoscopic modules are only planned within the COSECSA program. About 57% of participants reported that at the end of residency training, graduating surgeons were not able to perform basic laparoscopic procedures. The quoted barriers included: limited laparoscopic equipment, absence of simulation lab, lack of qualified trainers, lack of training programs and time for teaching by skilled doctors, and lack of institutional support. The well-structured set up of laparoscopic training programs in the COSECSA region is hindered due to the lack of qualified personnel and insufficient resources for the acquisition of equipment and simulation laboratories. Ongoing efforts to set up laparoscopic programs through the development of adaptive curricula, innovative strategies for reduction of equipment cost and adequate training of surgeons are crucial for patient safety and the development of laparoscopy.

Evaluating Text-to-Image GANs Performance: A Comparative Analysis of Evaluation Metrics

Generative Adversarial Networks (GANs) have emerged as powerful techniques for generating high-quality images in various domains but assessing how realistic the generated images are is a challenging task. To address this issue, researchers have proposed a variety of evaluation metrics for GANs, each with its own strengths and limitations. This paper presents a comprehensive analysis of popular GAN evaluation metrics, including FID, Mode Score, Inception Score, MMD, PSNR, and SSIM. The strengths, weaknesses, and calculation processes of these metrics are discussed, focusing on assessing image fidelity and diversity. Two approaches, pixel distance, and feature distance, are employed to measure image similarity, while the importance of evaluating individual objects using input captions is emphasized. Experimental results on a basic GAN trained on the MNIST dataset demonstrate improvement in various metrics across different epochs. The FID score decreases from 497.54594 at Epoch 0 to 136.91156 at Epoch 100, indicating improved differentiation between real and generated images. In addition, the Inception Score increases from 1.1533 to 1.6408, reflecting enhanced image quality and diversity. These findings highlight the effectiveness of the GAN model in generating more realistic and diverse images with training progression. However, when it comes to evaluating GANs on complex datasets, challenges arise, highlighting the need to combine evaluation metrics with visual inspection and subjective measures of image quality. By adopting a comprehensive evaluation approach, researchers can gain a deeper understanding of GAN performance and guide the development of advanced models.

Exploring Zero-Shot Semantic Segmentation with No Supervision Leakage

Zero-shot semantic segmentation (ZS3), the process of classifying unseen classes without explicit training samples, poses a significant challenge. Despite notable progress made by pre-trained vision-language models, they have a problem of “supervision leakage” in the unseen classes due to their large-scale pre-trained data. For example, CLIP is trained on 400M image–text pairs that contain large label space categories. So, it is not convincing for real “zero-shot” learning in machine learning. This paper introduces SwinZS3, an innovative framework that explores the “no-supervision-leakage” zero-shot semantic segmentation with an image encoder that is not pre-trained on the seen classes. SwinZS3 integrates the strengths of both visual and semantic embeddings within a unified joint embedding space. This approach unifies a transformer-based image encoder with a language encoder. A distinguishing feature of SwinZS3 is the implementation of four specialized loss functions in the training progress: cross-entropy loss, semantic-consistency loss, regression loss, and pixel-text score loss. These functions guide the optimization process based on dense semantic prototypes derived from the language encoder, making the encoder adept at recognizing unseen classes during inference without retraining. We evaluated SwinZS3 with standard ZS3 benchmarks, including PASCAL VOC and PASCAL Context. The outcomes affirm the effectiveness of our method, marking a new milestone in “no-supervison-leakage” ZS3 task performance.

Progressing On-Court Rehabilitation After Injury: The Control-Chaos Continuum Adapted to Basketball.

BACKGROUND: Sport-specific training is an integral component of returning to sport following injury. Frameworks designed to guide sport-specific rehabilitation need to integrate and adapt to the specific context of elite sport. The control-chaos continuum (CCC) is a flexible framework originally designed for on-pitch rehabilitation in elite football (soccer). The concepts underpinning the CCC transfer to other elite sport rehabilitation environments. CLINICAL QUESTION: How can practitioners and clinicians transfer the CCC to elite basketball, to support planning and return to sport? On-court rehabilitation is a critical sport-specific rehabilitation component of return to sport, yet there are no frameworks to guide practitioners when planning and delivering on-court rehabilitation. KEY RESULTS: Based on our experience working in the National Basketball Association, we report how the CCC framework can apply to elite basketball. We focus on the design and delivery of progressive training in the presence of injury in this basketball-specific edition of the CCC. Given the challenges when quantifying "load" in basketball, we encourage practitioners and clinicians to consider the qualitative aspects of performance such as skill, sport-specific movement, contact, and decision making. CLINICAL APPLICATION: The 5-phase framework describes training progression from high control, a return to on-court running, to high chaos, a return to "live" unrestricted basketball. The model can be adapted to both short- and long-term injuries based on injury and progression criteria. Strength and power "diagnostics" can be strategically implemented to enhance decision making throughout the return to sport continuum. J Orthop Sports Phys Ther 2023;53(9):1-12. Epub: 9 August 2023. doi:10.2519/jospt.2023.11981.