Explore the vast range of titles available.

Background To examine the effectiveness and safety of yoga of women with sleep problems by performing a systematic review and meta-analysis. Methods Medline/PubMed, ClinicalKey, ScienceDirect, Embase, PsycINFO, and the Cochrane Library were searched throughout the month of June, 2019. Randomized controlled trials comparing yoga groups with control groups in women with sleep problems were included. Two reviewers independently evaluated risk of bias by using the risk of bias tool suggested by the Cochrane Collaboration for programming and conducting systematic reviews and meta-analyses. The main outcome measure was sleep quality or the severity of insomnia, which was measured using subjective instruments, such as the Pittsburgh Sleep Quality Index (PSQI), Insomnia Severity Index (ISI), or objective instruments such as polysomnography, actigraphy, and safety of the intervention. For each outcome, a standardized mean difference (SMD) and confidence intervals (CIs) of 95% were determined. Results Nineteen studies in this systematic review included 1832 participants. The meta-analysis of the combined data conducted according to Comprehensive Meta-Analysis showed a significant improvement in sleep (SMD = − 0.327, 95% CI = − 0.506 to − 0.148, P  < 0.001). Meta-analyses revealed positive effects of yoga using PSQI scores in 16 randomized control trials (RCTs), compared with the control group in improving sleep quality among women using PSQI (SMD = − 0.54; 95% CI = − 0.89 to − 0.19; P  = 0.003). However, three RCTs revealed no effects of yoga compared to the control group in reducing insomnia among women using ISI (SMD = − 0.13; 95% CI = − 0.74 to 0.48; P  = 0.69). Seven RCTs revealed no evidence for effects of yoga compared with the control group in improving sleep quality for women with breast cancer using PSQI (SMD = − 0.15; 95% CI = − 0.31 to 0.01; P  = 0.5). Four RCTs revealed no evidence for the effects of yoga compared with the control group in improving the sleep quality for peri/postmenopausal women using PSQI (SMD = − 0.31; 95% CI = − 0.95 to 0.33; P  = 0.34). Yoga was not associated with any serious adverse events. Discussion This systematic review and meta-analysis demonstrated that yoga intervention in women can be beneficial when compared to non-active control conditions in term of managing sleep problems. The moderator analyses suggest that participants in the non-breast cancer subgroup and participants in the non-peri/postmenopausal subgroup were associated with greater benefits, with a direct correlation of total class time with quality of sleep among other related benefits.

Objectives. The aim of this study was to conduct a systematic review, meta-analysis, and metaregression to determine the current best available evidence of the efficacy and safety of foot reflexology for adult depression, anxiety, and sleep quality. Methods. Electronic databases (PubMed, ClinicalKey, ScienceDirect, EMBASE, PsycINFO, and the Cochrane Library) were searched till August, 10, 2020, and the validity of the eligible studies was critically appraised. Randomized controlled trials comparing foot reflexology groups with control groups for adult depression, anxiety, and sleep quality were included. Twenty-six eligible studies were included to assess the effect of foot reflexology intervention on the reducing symptoms of depression and anxiety and improving quality of sleep, respectively, as the primary outcome. Results. Twenty-six randomized controlled trials involving 2,366 participants met the inclusion criteria. The meta-analyses showed that foot reflexology intervention significantly improved adult depression (Hedges’ g = −0.921; 95% CI: −1.246 to −0.595; P < 0.001), anxiety (Hedges’ g = −1.237; 95% CI −1.682 to −0.791; P < 0.001), and sleep quality (Hedges’ g = −1.665; 95% CI −2.361 to −0.970; P < 0.001). Metaregression reveals that an increase in total foot reflexology time (P = 0.002) and duration (P = 0.01) can significantly improve sleep quality. Conclusions. Foot reflexology may provide additional nonpharmacotherapy intervention for adults suffering from depression, anxiety, or sleep disturbance. However, high quality and rigorous design RCTs in specific population, along with an increase in participants, and a long-term follow-up are recommended in the future.

The associations between the human immunodeficiency virus (HIV) and dementias are as yet to be studied in Taiwan. The aim of this study is to clarify as to whether HIV infections are associated with the risk of dementia. A total of 1,261 HIV-infected patients and 3,783 controls (1:3) matched for age and sex were selected between January 1 and December 31, 2000 from Taiwan's National Health Insurance Research Database (NHIRD). Fine and Gray's survival analysis (competing with mortality) analyzed the risk of dementias during the 15-year follow up. The association between the highly active antiretroviral therapy (HAART) and dementia was analyzed by stratifying the HAART status among the HIV subjects. During the follow-up period, 25 in the HIV group (N= 1,261) and 227 in the control group (N= 3,783) developed dementia (656.25 vs 913.15 per 100,000 person-years). Fine and Gray's survival analysis revealed that the HIV patients were not associated with an increased risk of dementia, with the adjusted hazard ratio (HR) as 0.852 (95% confidence interval [CI]: 0.189-2.886, =0.415) after adjusting for sex, age, comorbidities, geographical region, and the urbanization level of residence. There was no significant difference between the two groups of HIV-infected patients with or without HAART in the risk of dementia. This study found that HIV infections, either with or without HAART, were not associated with increased diagnoses of neurodegenerative dementias in patients older than 50 in Taiwan.

The stent is a major breakthrough in the treatment of coronary artery diseases. The permanent vascular implant of a stent, however, changes the intra-stent blood flow haemodynamics. There is a growing consensus that the stent implant may change the artery wall shear stress distribution and hence trigger the restenosis process. Computational fluid dynamics (CFD) has been widely used to analyse haemodynamics in stented arteries. In this Letter, CFD models were developed to investigate the effects of stent design pattern and strut geometry, respectively, on the wall shear stress distribution in coronary stented arteries. Assessment of the potential restenosis risk was primarily based on the wall shear stress distribution. Results show that the stent design pattern alone does not have a significant impact on the stent haemodynamic behaviour.Wall shear stress is very sensitive to strut thickness, while varying the strut width or crown radius has very little effect. The proposed methodology and findings will provide great insight for future optimisation of a stent design to reduce the risk of restenosis.

Aligning diffusion models with user preferences has been a key challenge. Existing methods for aligning diffusion models either require retraining or are limited to differentiable reward functions. To address these limitations, we propose a stochastic optimization approach, dubbed Demon, to guide the denoising process at inference time without backpropagation through reward functions or model retraining. Our approach works by controlling noise distribution in denoising steps to concentrate density on regions corresponding to high rewards through stochastic optimization. We provide comprehensive theoretical and empirical evidence to support and validate our approach, including experiments that use non-differentiable sources of rewards such as Visual-Language Model (VLM) APIs and human judgements. To the best of our knowledge, the proposed approach is the first inference-time, backpropagation-free preference alignment method for diffusion models. Our method can be easily integrated with existing diffusion models without further training. Our experiments show that the proposed approach significantly improves the average aesthetics scores for text-to-image generation.

Current Large Language Models (LLMs) are not only limited to some maximum context length, but also are not able to robustly consume long inputs. To address these limitations, we propose ReadAgent, an LLM agent system that increases effective context length up to 20x in our experiments. Inspired by how humans interactively read long documents, we implement ReadAgent as a simple prompting system that uses the advanced language capabilities of LLMs to (1) decide what content to store together in a memory episode, (2) compress those memory episodes into short episodic memories called gist memories, and (3) take actions to look up passages in the original text if ReadAgent needs to remind itself of relevant details to complete a task. We evaluate ReadAgent against baselines using retrieval methods, using the original long contexts, and using the gist memories. These evaluations are performed on three long-document reading comprehension tasks: QuALITY, NarrativeQA, and QMSum. ReadAgent outperforms the baselines on all three tasks while extending the effective context window by 3.5-20x.

Imitation learning often needs a large demonstration set in order to handle the full range of situations that an agent might find itself in during deployment. However, collecting expert demonstrations can be expensive. Recent work in vision, reinforcement learning, and NLP has shown that auxiliary representation learning objectives can reduce the need for large amounts of expensive, task-specific data. Our Empirical Investigation of Representation Learning for Imitation (EIRLI) investigates whether similar benefits apply to imitation learning. We propose a modular framework for constructing representation learning algorithms, then use our framework to evaluate the utility of representation learning for imitation across several environment suites. In the settings we evaluate, we find that existing algorithms for image-based representation learning provide limited value relative to a well-tuned baseline with image augmentations. To explain this result, we investigate differences between imitation learning and other settings where representation learning has provided significant benefit, such as image classification. Finally, we release a well-documented codebase which both replicates our findings and provides a modular framework for creating new representation learning algorithms out of reusable components.

Grounding language to visual relations is critical to various language-and-vision applications. In this work, we tackle two fundamental language-and-vision tasks: image-text matching and image captioning, and demonstrate that neural scene graph generators can learn effective visual relation features to facilitate grounding language to visual relations and subsequently improve the two end applications. By combining relation features with the state-of-the-art models, our experiments show significant improvement on the standard Flickr30K and MSCOCO benchmarks. Our experimental results and analysis show that relation features improve downstream models' capability of capturing visual relations in end vision-and-language applications. We also demonstrate the importance of learning scene graph generators with visually relevant relations to the effectiveness of relation features.

In this paper, we study the problem of image-text matching. Inferring the latent semantic alignment between objects or other salient stuff (e.g. snow, sky, lawn) and the corresponding words in sentences allows to capture fine-grained interplay between vision and language, and makes image-text matching more interpretable. Prior work either simply aggregates the similarity of all possible pairs of regions and words without attending differentially to more and less important words or regions, or uses a multi-step attentional process to capture limited number of semantic alignments which is less interpretable. In this paper, we present Stacked Cross Attention to discover the full latent alignments using both image regions and words in a sentence as context and infer image-text similarity. Our approach achieves the state-of-the-art results on the MS-COCO and Flickr30K datasets. On Flickr30K, our approach outperforms the current best methods by 22.1% relatively in text retrieval from image query, and 18.2% relatively in image retrieval with text query (based on Recall@1). On MS-COCO, our approach improves sentence retrieval by 17.8% relatively and image retrieval by 16.6% relatively (based on Recall@1 using the 5K test set). Code has been made available at: https://github.com/kuanghuei/SCAN.

Animals have evolved various agile locomotion strategies, such as sprinting, leaping, and jumping. There is a growing interest in developing legged robots that move like their biological counterparts and show various agile skills to navigate complex environments quickly. Despite the interest, the field lacks systematic benchmarks to measure the performance of control policies and hardware in agility. We introduce the Barkour benchmark, an obstacle course to quantify agility for legged robots. Inspired by dog agility competitions, it consists of diverse obstacles and a time based scoring mechanism. This encourages researchers to develop controllers that not only move fast, but do so in a controllable and versatile way. To set strong baselines, we present two methods for tackling the benchmark. In the first approach, we train specialist locomotion skills using on-policy reinforcement learning methods and combine them with a high-level navigation controller. In the second approach, we distill the specialist skills into a Transformer-based generalist locomotion policy, named Locomotion-Transformer, that can handle various terrains and adjust the robot's gait based on the perceived environment and robot states. Using a custom-built quadruped robot, we demonstrate that our method can complete the course at half the speed of a dog. We hope that our work represents a step towards creating controllers that enable robots to reach animal-level agility.