No abstract

Most recent semantic segmentation methods adopt a fully-convolutional network (FCN) with an encoderdecoder architecture. The encoder progressively reduces the spatial resolution and learns more abstract/semantic visual concepts with larger receptive fields. Since context modeling is critical for segmentation, the latest efforts have been focused on increasing the receptive field, through either dilated/atrous convolutions or inserting attention modules. However, the encoder-decoder based FCN architecture remains unchanged. In this paper, we aim to provide an alternative perspective by treating semantic segmentation as a sequence-to-sequence prediction task. Specifically, we deploy a pure transformer (i.e., without convolution and resolution reduction) to encode an image as a sequence of patches. With the global context modeled in every layer of the transformer, this encoder can be combined with a simple decoder to provide a powerful segmentation model, termed SEgmentation TRansformer (SETR). Extensive experiments show that SETR achieves new state of the art on ADE20K (50.28% mIoU), Pascal Context (55.83% mIoU) and competitive results on Cityscapes. Particularly, we achieve the first (44.42% mIoU) position in the highly competitive ADE20K test server leaderboard.

No abstract

Nationwide surveys are urgently needed to further explore the prevalence of alcohol-related disorders in China.

The objective of this cross-sectional study was to identify the factors associated with health-related quality of life (HRQOL) of family caregivers of Chinese stroke patients. One hundred and twenty-three stroke patients consecutively admitted to a stroke clinic and their 123 family caregivers were recruited. The caregivers' HRQOLs were assessed with the Short Form 36 (SF-36). Two subscores of the SF-36, the Physical Component Summary and the Mental Component Summary Scores served as the dependent variables. Independent variables comprised patients and caregivers' demographic data and physical and psychological conditions in relation to caregivers' HRQOL. The severity of coexisting illnesses of both the patients and their caregivers was assessed using the Cumulative Illness Rating Scale. The Geriatric Depression Scale (GDS) was administered to rate both the patients and their caregivers' depressive symptoms. The extent of the patients' functional disability was assessed using the Barthel Index and the Instrumental Activities of Daily Living. Global cognition of the patients was measured using the Mini Mental State Examination. The level of social support and recent life events of the caregivers were rated using Lubben Social Network Scale and the Modified Life Event Scale, respectively. The multiple linear regression revealed caregivers' Cumulative Illness Rating Scale and GDS score were associated with Physical Component Summary Score of SF-36. The patients and caregivers' GDS score were associated with Mental Component Summary Score of SF-36. Severity of depressive symptoms in Chinese stroke patients and their caregivers is an important factor associated with the caregivers' HRQOL. Further studies evaluating interventions on caregivers' HRQOL should include the assessment and management of mood disorders.

ImportanceArea-level factors have been identified as important social determinants of health (SDoH) that impact many health-related outcomes. Less is known about how the social vulnerability index (SVI), as a scalable composite score, can multidimensionally explain the population-based cancer screening program uptake at a county level.ObjectiveTo examine the geographic variation of US Preventive Services Task Force (USPSTF)–recommended breast, cervical, and colorectal cancer screening rates and the association between county-level SVI and the 3 screening rates.Design, Setting, and ParticipantsThis population-based cross-sectional study used county-level information from the Centers for Disease Control and Prevention’s PLACES and SVI data sets from 2018 for 3141 US counties. Analyses were conducted from October 2021 to February 2022.ExposuresSocial vulnerability index score categorized in quintiles.Main Outcomes and MeasuresThe main outcome was county-level rates of USPSTF guideline-concordant, up-to-date breast, cervical, and colorectal screenings. Odds ratios were calculated for each cancer screening by SVI quintile as unadjusted (only accounting for eligible population per county) or adjusted for urban-rural status, percentage of uninsured adults, and primary care physician rate per 100 000 residents.ResultsAcross 3141 counties, county-level cancer screening rates showed regional disparities ranging from 54.0% to 81.8% for breast cancer screening, from 69.9% to 89.7% for cervical cancer screening, and from 39.8% to 74.4% for colorectal cancer screening. The multivariable regression model showed that a higher SVI was significantly associated with lower odds of cancer screening, with the lowest odds in the highest SVI quintile. When comparing the highest quintile of SVI (SVI-Q5) with the lowest quintile of SVI (SVI-Q1), the unadjusted odds ratio was 0.86 (95% posterior credible interval [CrI], 0.84-0.87) for breast cancer screening, 0.80 (95% CrI, 0.79-0.81) for cervical cancer screening, and 0.72 (95% CrI, 0.71-0.73) for colorectal cancer screening. When fully adjusted, the odds ratio was 0.92 (95% CrI, 0.90-0.93) for breast cancer screening, 0.87 (95% CrI, 0.86-0.88) for cervical cancer screening, and 0.86 (95% CrI, 0.85-0.88) for colorectal cancer screening, showing slightly attenuated associations.Conclusions and RelevanceIn this cross-sectional study, regional disparities were found in cancer screening rates at a county level. Quantifying how SVI associates with each cancer screening rate could provide insight into the design and focus of future interventions targeting cancer prevention disparities.

Effective measures should be provided to prevent falls in psychiatric wards. A prospective study exploring predictors of falls is warranted.

Insomnia symptoms have a multi-factorial origin in stroke. Besides depressive symptoms, frontal lobe infarction and diabetes mellitus predict insomnia symptoms.