Grab Bar Use Influences Fall Hazard During Bathtub Exit

Objective This study evaluated the hazard (risk of unrecovered balance loss and hazardous fall) and strategies associated with grab bar use, compared to no grab bar use, during unexpected balance loss initiated whilst exiting a bathtub. Background While independent bathing is critical for maintaining self-sufficiency, injurious falls during bathing transfer tasks are common. Grab bars are recommended to support bathing tasks, but no evidence exists regarding fall prevention efficacy. Method Sixty-three adults completed a hazardous bathtub transfer task, experiencing an unpredictable external balance perturbation while stepping from a slippery bathtub to a dry surface. Thirty-two were provided a grab bar, while 31 had no grab bar available. Slips and grab bar use were recorded via four video cameras. Slip occurrence and strategy were identified by two independent video coders. Results Participants who had a grab bar were 75.8% more likely to recover their balance during the task than those who did not have a grab bar. Successful grab bar grasp was associated with balance recovery in all cases. Attempts to stabilize using other environmental elements, or using internal strategies only, were less successful balance recovery strategies. Grab bar presence appeared to cue use of the environment for stability. Proactive grasp and other strategies modified grasping success. Conclusion Grab bars appear to provide effective support for recovery from unexpected balance loss. Grab bar presence may instigate development of fall prevention strategies prior to loss of balance. Application Bathroom designs with grab bars may reduce frequency of fall-related injuries during bathing transfer tasks.

Effect of Parameters on Lumbar Compressive Force during Patient Transfer

Patient transfer (PT) tasks are a significant cause of low back pain (LBP) in caregivers. Adopting proper motion strategies is an effective and inexpensive approach to reduce the risk of LBP. However, since the standardization of PT tasks is not specified in ISO 11228, there is an increasing need to develop a quantitative assessment method for the lumbar safety of caregivers. Therefore, we aim to determine the effect of representative factors, extracted from caregivers’ movements and of external force, on peak compressive force (CF) in patient transfer tasks using the lumbar compressive force as a criterion. The CF at the lumbar region is estimated using a biomechanical simulator, and regression analysis is performed between the estimated CF and representative factors. The results imply that peak CF occurs in the incipience of transfer and occurs after the occurrence of the peak trunk angle. The results also indicate that the peak CF can be reduced by preventing the representative factors from simultaneously reaching the maximum values. In this study, we provide a method of reducing peak CF by estimating the timing and magnitude of the peak to help caregivers assess the severity of LBP risk in actual PT, which is expected to contribute to the standardization of PT tasks.

QOLP-11. WORKING MEMORY TRAINING FOR ADULTS WITH GLIOMA

BACKGROUND CogMed Working Memory Training (CWMT) is a computer-based program shown to improve working memory (WM) among those with cognitive impairment. No study to date has investigated its feasibility, acceptability, and satisfaction in adult patients with glioma, despite the well-documented incidence of WM impairment in this population. METHODS Twenty patients with glioma and objective and/or perceived WM deficits enrolled in the study: 52% high-grade, 57% female, Mage=47 (range=21-72 years). Adverse events were monitored to determine safety. Feasibility and acceptability were assessed based on established metrics. Satisfaction was explored by exit-interviews. Neurocognitive tests and measures of psychological distress were administered pre-/post-CWMT to assess preliminary efficacy. RESULTS Of 20 enrolled patients, 16 completed the study protocol (80% retention rate). Reasons for withdrawal included time burden (n=2); tumor-related fatigue (n=1) or lost to follow-up (n=1). No adverse events were determined to be study-related. Adherence was 69%. The perceived degree of benefit was only moderate. Pre- to post-CWMT assessments showed medium to large effects on near-transfer tasks (h p 2 =.35, p=.01 and h p 2 =.25, p=.04) and far-transfer tasks (h p 2 =.20, p=.07 and h p 2 =.16, p=.12) but small to no effects on perceived WM (h p 2 =.01, p=.79) and psychological distress (h p 2 =.01-.06, p=.35-.79). CONCLUSION CWMT was found to be safe and acceptable in adult patients with glioma. Enrollment, retention rates, and treatment adherence were all adequate, yet only moderate perceived benefit was reported. Given that objective measures of WM improved but psychological distress did not, it may be worth considering a less burdensome CWMT protocol, perhaps investigating a less time intensive intervention with respect to both frequency and length of training sessions.

Design and Compliant Control of a Piggyback Transfer Robot

Patient transfer, such as lifting and moving a bedridden patient from a bed to a wheelchair or a pedestal pan, is one of the most physically challenging tasks in nursing care. Although many transfer devices have been developed, they are rarely used because of the large time consumption in performing transfer tasks and the lack of safety and comfortableness. We developed a piggyback transfer robot that can conduct patient transfer by imitating the motion when a person holds another person on his/her back. The robot consisted of a chest holder that moves like a human back. In this paper, we present an active stiffness control approach for the motion control of the chest holder, combined with a passive cushion, for lifting a care-receiver comfortably. A human-robot dynamic model was built and a subjective evaluation was conducted to optimize the parameters of both the active stiffness control and the passive cushion of the chest holder. The test results of 10 subjects demonstrated that the robot could transfer a subject safely and the combination of active stiffness and passive stiffness were essential to a comfortable transfer. The objective evaluation demonstrated that an active stiffness of k= 4 kPa/mm along with a passive stiffness lower than the stiffness of human chest was helpful for a comfort feeling.

Hyperspectral Imaging Combined With Deep Transfer Learning for Rice Disease Detection

Various rice diseases threaten the growth of rice. It is of great importance to achieve the rapid and accurate detection of rice diseases for precise disease prevention and control. Hyperspectral imaging (HSI) was performed to detect rice leaf diseases in four different varieties of rice. Considering that it costs much time and energy to develop a classifier for each variety of rice, deep transfer learning was firstly introduced to rice disease detection across different rice varieties. Three deep transfer learning methods were adapted for 12 transfer tasks, namely, fine-tuning, deep CORrelation ALignment (CORAL), and deep domain confusion (DDC). A self-designed convolutional neural network (CNN) was set as the basic network of the deep transfer learning methods. Fine-tuning achieved the best transferable performance with an accuracy of over 88% for the test set of the target domain in the majority of transfer tasks. Deep CORAL obtained an accuracy of over 80% in four of all the transfer tasks, which was superior to that of DDC. A multi-task transfer strategy has been explored with good results, indicating the potential of both pair-wise, and multi-task transfers. A saliency map was used for the visualization of the key wavelength range captured by CNN with and without transfer learning. The results indicated that the wavelength range with and without transfer learning was overlapped to some extent. Overall, the results suggested that deep transfer learning methods could perform rice disease detection across different rice varieties. Hyperspectral imaging, in combination with the deep transfer learning method, is a promising possibility for the efficient and cost-saving field detection of rice diseases among different rice varieties.

Commercial Brain Training: Efficacy, Transfer Effects, and the Influence of Personality Traits: A Study Conducted on Healthy Young Adults

In the present study, we investigated the effects of a four-week working memory (WM) and attention training program using commercial brain training (Synaptikon GmbH, Berlin). Sixty young healthy adults were assigned to the experimental and active control training programs. The training was conducted in a naturalistic home-based setting, while the pre- and post-examinations were conducted in a controlled laboratory setting. Transfer effects to an untrained WM task and to an untrained episodic memory task were examined. Furthermore, possible influences of personality, i.e., the five-factor model (FFM) traits and need for cognition (NFC), on training outcomes were examined. Additionally, the direct relationship between improvement in single trained tasks and improvement in the transfer tasks was investigated. Our results showed that both training groups significantly increased performance in the WM task, but only the WM training group increased their performance in the episodic memory transfer task. One of the training tasks, a visuospatial WM task, was particularly associated with improvement in the episodic memory task. Neuroticism and conscientiousness showed differential effects on the improvement in training and transfer tasks. It needs to be further examined whether these effects represent training effects or, for example, retest/practice or motivation effects.

GTAE: Graph Transformer–Based Auto-Encoders for Linguistic-Constrained Text Style Transfer

Non-parallel text style transfer has attracted increasing research interests in recent years. Despite successes in transferring the style based on the encoder-decoder framework, current approaches still lack the ability to preserve the content and even logic of original sentences, mainly due to the large unconstrained model space or too simplified assumptions on latent embedding space. Since language itself is an intelligent product of humans with certain grammars and has a limited rule-based model space by its nature, relieving this problem requires reconciling the model capacity of deep neural networks with the intrinsic model constraints from human linguistic rules. To this end, we propose a method called Graph Transformer–based Auto-Encoder, which models a sentence as a linguistic graph and performs feature extraction and style transfer at the graph level, to maximally retain the content and the linguistic structure of original sentences. Quantitative experiment results on three non-parallel text style transfer tasks show that our model outperforms state-of-the-art methods in content preservation, while achieving comparable performance on transfer accuracy and sentence naturalness.

Can inhibition or shifting training enhance cognitive abilities in emerging adults?

Executive functions enable and support most of our daily cognitive functioning. Within the number of executive functions proposed, updating, inhibition and shifting are most often considered as the three core executive functions. Cognitive training paradigms provide a platform for a possible enhancement of these functions. Since updating training has been studied to a greater extent, we wanted to investigate the effectiveness of inhibition and shifting training in this study. Emerging adults (psychology students) were randomly assigned either to the inhibition training (based on the Simon task; n = 36) or to the shifting training (based on the task switching paradigm; n = 35). Both groups underwent twelve 20-minute sessions distributed over four weeks. Measurements before and after the training included criterion tasks (i.e. the training tasks), near-transfer tasks (i.e. tasks that address the trained functions but use different types of stimuli or rules to respond), and far-transfer tasks (i.e., tasks that address untrained cognitive functions). The control participants (n = 36) were tested with a combination of these tasks. Both training groups improved their criteria task performance over time, while convincing training-related gains were not found in either near- or far-transfer tasks. This study raises some conceptual questions for the training of executive functions with respect to a sample of emerging adults with above-average cognitive abilities, motivational elements of training, and the role of executive functions in more complex everyday cognitive activities.

TransBERT

Recent advances, such as GPT, BERT, and RoBERTa, have shown success in incorporating a pre-trained transformer language model and fine-tuning operations to improve downstream NLP systems. However, this framework still has some fundamental problems in effectively incorporating supervised knowledge from other related tasks. In this study, we investigate a transferable BERT (TransBERT) training framework, which can transfer not only general language knowledge from large-scale unlabeled data but also specific kinds of knowledge from various semantically related supervised tasks, for a target task. Particularly, we propose utilizing three kinds of transfer tasks, including natural language inference, sentiment classification, and next action prediction, to further train BERT based on a pre-trained model. This enables the model to get a better initialization for the target task. We take story-ending prediction as the target task to conduct experiments. The final results of 96.0% and 95.0% accuracy on two versions of Story Cloze Test datasets dramatically outperform previous state-of-the-art baseline methods. Several comparative experiments give some helpful suggestions on how to select transfer tasks to improve BERT. Furthermore, experiments on six English and three Chinese datasets show that TransBERT generalizes well to other tasks, languages, and pre-trained models.