Relationship between Tai Chi and clinical outcomes in elderly patients with COVID-19: a protocol for systematic review and dose-response meta-analysis.

INTRODUCTION: COVID-19 has posed a serious threat to people worldwide, especially the older adults, since its discovery. Tai Chi as a traditional Chinese exercisethat belongs to traditional Chinese medicine has proven its effectiveness against COVID-19. However, no high-quality evidence is found on the dose-response relationships between Tai Chi and clinical outcomes in patients with COVID-19. This study will evaluate and determine the clinical evidence of Tai Chi as a treatment in elderly patients with COVID-19. METHODS AND ANALYSIS: The following electronic bibliographical databases including PubMed, EMBASE, Web of Science, Cochrane Library, China National Knowledge Infrastructure, VIP Database and Wanfang Database will be screened from their inception date to 30 June 2022. All eligible randomised controlled trials or controlled clinical trials related to Tai Chi for elderly patients with COVID-19 will be included. The primary outcomes are forced expiratory volume in 1 s (FEV1), forced vital capacity (FVC) and FEV1/FVC ratio (FEV1%). The secondary outcomes are the time of main symptoms disappearance, length of hospital stay, serum levels of interleukin (IL)-6, IL-1b and tumour necrosis factor-α, and adverse event rate. Two independent reviewers will select the studies, extract the data, and analyse them on EndNote V.X9.0 and Stata V.12.1. The robust error meta-regression model will be used to establish the dose-response relationships between Tai Chi and clinical outcomes. The heterogeneity and variability will be analysed by I2 and τ2 statistics. Risk of bias, subgroup analysis and sensitivity analysis will also be performed. The quality of evidence will be assessed by the Grading of Recommendations Assessment, Development and Evaluation, and the risk of bias will be evaluated by using the Physiotherapy Evidence Database Scale. ETHICS AND DISSEMINATION: This study will review published data; thus, obtaining ethical approval and consent is unnecessary. The results will be disseminated through peer-reviewed publications. PROSPERO REGISTRATION NUMBER: CRD42022327694.

Segmentation ability map: Interpret deep features for medical image segmentation.

Deep convolutional neural networks (CNNs) have been widely used for medical image segmentation. In most studies, only the output layer is exploited to compute the final segmentation results and the hidden representations of the deep learned features have not been well understood. In this paper, we propose a prototype segmentation (ProtoSeg) method to compute a binary segmentation map based on deep features. We measure the segmentation abilities of the features by computing the Dice between the feature segmentation map and ground-truth, named as the segmentation ability score (SA score for short). The corresponding SA score can quantify the segmentation abilities of deep features in different layers and units to understand the deep neural networks for segmentation. In addition, our method can provide a mean SA score which can give a performance estimation of the output on the test images without ground-truth. Finally, we use the proposed ProtoSeg method to compute the segmentation map directly on input images to further understand the segmentation ability of each input image. Results are presented on segmenting tumors in brain MRI, lesions in skin images, COVID-related abnormality in CT images, prostate segmentation in abdominal MRI, and pancreatic mass segmentation in CT images. Our method can provide new insights for interpreting and explainable AI systems for medical image segmentation. Our code is available on: https://github.com/shengfly/ProtoSeg.