Nonparametric two-sample tests of high dimensional mean vectors via random integration.

Testing the equality of the means in two samples is a fundamental statistical inferential problem. Most of the existing methods are based on the sum-of-squares or supremum statistics. They are possibly powerful in some situations, but not in others, and they do not work in a unified way. Using random integration of the difference, we develop a framework that includes and extends many existing methods, especially in high-dimensional settings, without restricting the same covariance matrices or sparsity. Under a general multivariate model, we can derive the asymptotic properties of the proposed test statistic without specifying a relationship between the data dimension and sample size explicitly. Specifically, the new framework allows us to better understand the test's properties and select a powerful procedure accordingly. For example, we prove that our proposed test can achieve the power of 1 when nonzero signals in the true mean differences are weakly dense with nearly the same sign. In addition, we delineate the conditions under which the asymptotic relative Pitman efficiency of our proposed test to its competitor is greater than or equal to 1. Extensive numerical studies and a real data example demonstrate the potential of our proposed test.

Enhancing Retinal Fundus Image Quality Assessment With Swin-Transformer-Based Learning Across Multiple Color-Spaces.

Purpose: The assessment of retinal image (RI) quality holds significant importance in both clinical trials and large datasets, because suboptimal images can potentially conceal early signs of diseases, thereby resulting in inaccurate medical diagnoses. This study aims to develop an automatic method for Retinal Image Quality Assessment (RIQA) that incorporates visual explanations, aiming to comprehensively evaluate the quality of retinal fundus images (RIs). Methods: We developed an automatic RIQA system, named Swin-MCSFNet, utilizing 28,792 RIs from the EyeQ dataset, as well as 2000 images from the EyePACS dataset and an additional 1,000 images from the OIA-ODIR dataset. After preprocessing, including cropping black regions, data augmentation, and normalization, a Swin-MCSFNet classifier based on the Swin-Transformer for multiple color-space fusion was proposed to grade the quality of RIs. The generalizability of Swin-MCSFNet was validated across multiple data centers. Additionally, for enhanced interpretability, a Score-CAM-generated heatmap was applied to provide visual explanations. Results: Experimental results reveal that the proposed Swin-MCSFNet achieves promising performance, yielding a micro-receiver operating characteristic (ROC) of 0.93 and ROC scores of 0.96, 0.81, and 0.96 for the "Good," "Usable," and "Reject" categories, respectively. These scores underscore the accuracy of RIQA based on Swin-MCSF in distinguishing among the three categories. Furthermore, heatmaps generated across different RIQA classification scores and various color spaces suggest that regions in the retinal images from multiple color spaces contribute significantly to the decision-making process of the Swin-MCSFNet classifier. Conclusions: Our study demonstrates that the proposed Swin-MCSFNet outperforms other methods in experiments conducted on multiple datasets, as evidenced by the superior performance metrics and insightful Score-CAM heatmaps. Translational Relevance: This study constructs a new retinal image quality evaluation system, which will contribute to the subsequent research of retinal images.

Use of random integration to test equality of high dimensional covariance matrices.

Testing the equality of two covariance matrices is a fundamental problem in statistics, and especially challenging when the data are high-dimensional. Through a novel use of random integration, we can test the equality of high-dimensional covariance matrices without assuming parametric distributions for the two underlying populations, even if the dimension is much larger than the sample size. The asymptotic properties of our test for arbitrary number of covariates and sample size are studied in depth under a general multivariate model. The finite-sample performance of our test is evaluated through numerical studies. The empirical results demonstrate that our test is highly competitive with existing tests in a wide range of settings. In particular, our proposed test is distinctly powerful under different settings when there exist a few large or many small diagonal disturbances between the two covariance matrices.

Data-driven analysis of the simulations of the spread of COVID-19 under different interventions of China.

Since February 2020, COVID-19 has spread rapidly to more than 200 countries in the world. During the pandemic, local governments in China have implemented different interventions to efficiently control the spread of the epidemic. Characterizing transmission of COVID-19 under some typical interventions is essential to help countries develop appropriate interventions. Based on the pre-symptomatic transmission patterns of COVID-19, we established a novel compartmental model: Susceptible-Infectious-Confirmed-Removed (SICR) model, which allowed the effective reproduction number to change over time, thus the effects of policies could be reasonably estimated. Using the epidemic data of Wuhan, Wenzhou, and Shenzhen, we migrated the corresponding estimated policy modes to South Korea, Italy, and the United States and simulated the potential outcomes for these countries when they adopted similar policy strategies to China. We found that the mild interventions implemented in Shenzhen were effective in controlling the epidemic in the early stage, while more stringent policies which were implemented in Wuhan and Wenzhou were necessary if the epidemic became severe and needed to be controlled in a short time.

Sex-specific relationships between prenatal exposure to metal mixtures and birth weight in a Chinese birth cohort.

Birth weight is an indicator linking intrauterine environmental exposures to later-life diseases, and intrauterine metal exposure may affect birth weight in a sex-specific manner. We investigated sex-specific associations between prenatal exposure to metal mixtures and birth weight in a Chinese birth cohort. The birth weight of 1296 boys and 1098 girls were recorded, and 10 metals in maternal urine samples collected during pregnancy were measured using inductively coupled plasma mass spectrometry. Bayesian Kernel Machine Regression was used to estimate the association of individual metals or metal mixtures and birth weight for gestational age (BW for GA). The model showed a sex-specific relationship between prenatal exposure to metal mixtures and BW for GA with a significant negative association in girls and a non-significant positive association in boys. Cadmium (Cd) and nickel (Ni) were positively and negatively associated with BW for GA in girls, respectively. Moreover, increasing thallium (Tl) concentration lowered the positive association between Cd and BW for GA and enhanced the negative association between Ni and BW for GA in girls. When exposure to other metals increased, the positive association with Cd diminished, whereas the negative association with Ni or Tl increased. Our findings provide evidence supporting the complex effects of intrauterine exposure to metal mixtures on the birth weight of girls and further highlight the sex heterogeneity in fetal development influenced by intrauterine environmental factors.

Computational Analysis of Pathological Image Enables Interpretable Prediction for Microsatellite Instability.

Background: Microsatellite instability (MSI) is associated with several tumor types and has become increasingly vital in guiding patient treatment decisions; however, reasonably distinguishing MSI from its counterpart is challenging in clinical practice. Methods: In this study, interpretable pathological image analysis strategies are established to help medical experts to identify MSI. The strategies only require ubiquitous hematoxylin and eosin-stained whole-slide images and perform well in the three cohorts collected from The Cancer Genome Atlas. Equipped with machine learning and image processing technique, intelligent models are established to diagnose MSI based on pathological images, providing the rationale of the decision in both image level and pathological feature level. Findings: The strategies achieve two levels of interpretability. First, the image-level interpretability is achieved by generating localization heat maps of important regions based on deep learning. Second, the feature-level interpretability is attained through feature importance and pathological feature interaction analysis. Interestingly, from both the image-level and feature-level interpretability, color and texture characteristics, as well as their interaction, are shown to be mostly contributed to the MSI prediction. Interpretation: The developed transparent machine learning pipeline is able to detect MSI efficiently and provide comprehensive clinical insights to pathologists. The comprehensible heat maps and features in the intelligent pipeline reflect extra- and intra-cellular acid-base balance shift in MSI tumor.

An Interaction-Based Method for Refining Results From Gene Set Enrichment Analysis.

Purpose: To demonstrate an interaction-based method for the refinement of Gene Set Enrichment Analysis (GSEA) results. Method: Intravitreal injection of miR-124-3p antagomir was used to knockdown the expression of miR-124-3p in mouse retina at postnatal day 3 (P3). Whole retinal RNA was extracted for mRNA transcriptome sequencing at P9. After preprocessing the dataset, GSEA was performed, and the leading-edge subsets were obtained. The Apriori algorithm was used to identify the frequent genes or gene sets from the union of the leading-edge subsets. A new statistic d was introduced to evaluate the frequent genes or gene sets. Reverse transcription quantitative PCR (RT-qPCR) was performed to validate the expression trend of candidate genes after the knockdown of miR-124-3p. Results: A total of 115,140 assembled transcript sequences were obtained from the clean data. With GSEA, the NOD-like receptor signaling pathway, C-type-like lectin receptor signaling pathway, phagosome, necroptosis, JAK-STAT signaling pathway, Toll-like receptor signaling pathway, leukocyte transendothelial migration, chemokine signaling pathway, NF-kappa B signaling pathway and RIG-I-like signaling pathway were identified as the top 10 enriched pathways, and their leading-edge subsets were obtained. After being refined by the Apriori algorithm and sorted by the value of the modulus of d , Prkcd, Irf9, Stat3, Cxcl12, Stat1, Stat2, Isg15, Eif2ak2, Il6st, Pdgfra, Socs4 and Csf2ra had the significant number of interactions and the greatest value of d to downstream genes among all frequent transactions. Results of RT-qPCR validation for the expression of candidate genes after the knockdown of miR-124-3p showed a similar trend to the RNA-Seq results. Conclusion: This study indicated that using the Apriori algorithm and defining the statistic d was a novel way to refine the GSEA results. We hope to convey the intricacies from the computational results to the low-throughput experiments, and to plan experimental investigations specifically.

Segmentation of Laser Marks of Diabetic Retinopathy in the Fundus Photographs Using Lightweight U-Net.

Diabetic retinopathy (DR) is a prevalent vision-threatening disease worldwide. Laser marks are the scars left after panretinal photocoagulation, a treatment to prevent patients with severe DR from losing vision. In this study, we develop a deep learning algorithm based on the lightweight U-Net to segment laser marks from the color fundus photos, which could help indicate a stage or providing valuable auxiliary information for the care of DR patients. We prepared our training and testing data, manually annotated by trained and experienced graders from Image Reading Center, Zhongshan Ophthalmic Center, publicly available to fill the vacancy of public image datasets dedicated to the segmentation of laser marks. The lightweight U-Net, along with two postprocessing procedures, achieved an AUC of 0.9824, an optimal sensitivity of 94.16%, and an optimal specificity of 92.82% on the segmentation of laser marks in fundus photographs. With accurate segmentation and high numeric metrics, the lightweight U-Net method showed its reliable performance in automatically segmenting laser marks in fundus photographs, which could help the AI assist the diagnosis of DR in the severe stage.

Risk factors associated with mortality of COVID-19 in 3125 counties of the United States.

BACKGROUND: The number of cumulative confirmed cases of COVID-19 in the United States has risen sharply since March 2020. A county health ranking and roadmaps program has been established to identify factors associated with disparity in mobility and mortality of COVID-19 in all counties in the United States. The risk factors associated with county-level mortality of COVID-19 with various levels of prevalence are not well understood. METHODS: Using the data obtained from the County Health Rankings and Roadmaps program, this study applied a negative binomial design to the county-level mortality counts of COVID-19 as of August 27, 2020 in the United States. In this design, the infected counties were categorized into three levels of infections using clustering analysis based on time-varying cumulative confirmed cases from March 1 to August 27, 2020. COVID-19 patients were not analyzed individually but were aggregated at the county-level, where the county-level deaths of COVID-19 confirmed by the local health agencies. Clustering analysis and Kruskal-Wallis tests were used in our statistical analysis. RESULTS: A total of 3125 infected counties were assigned into three classes corresponding to low, median, and high prevalence levels of infection. Several risk factors were significantly associated with the mortality counts of COVID-19, where higher level of air pollution (0.153, P < 0.001) increased the mortality in the low prevalence counties and elder individuals were more vulnerable in both the median (0.049, P < 0.001) and high (0.114, P < 0.001) prevalence counties. The segregation between non-Whites and Whites (low: 0.015, P < 0.001; median:0.025, P < 0.001; high: 0.019, P = 0.005) and higher Hispanic population (low and median: 0.020, P < 0.001; high: 0.014, P = 0.009) had higher likelihood of risk of the deaths in all infected counties. CONCLUSIONS: The mortality of COVID-19 depended on sex, race/ethnicity, and outdoor environment. The increasing awareness of the impact of these significant factors may help decision makers, the public health officials, and the general public better control the risk of pandemic, particularly in the reduction in the mortality of COVID-19.

Identification of key miRNAs and genes for mouse retinal development using a linear model.

MicroRNAs (miRNAs) are upstream regulators of gene expression and are involved in several biological processes. The purpose of the present study was to obtain a detailed spatiotemporal miRNA expression profile in mouse retina, to identify one or more miRNAs that are key to mouse retinal development and to investigate the roles and mechanisms of these miRNAs. The miRNA expression pattern of the developing mouse retina was acquired from Locked Nucleic Acid microarrays. Data were processed to identify differentially expressed miRNAs (DE­miRNAs) using the linear model in Python 3.6. Following bioinformatics analysis and reverse transcription­quantitative polymerase chain reaction validation, 8 miRNAs (miR­9­5p, miR­130a­3p, miR­92a­3p, miR­20a­5p, miR­93­5p, miR­9­3p, miR­709 and miR­124) were identified as key DE­miRNAs with low variability during mouse retinal development. Gene Ontology analysis revealed that the target genes of the DE­miRNAs were enriched in cellular metabolic processes. Kyoto Encyclopedia of Genes and Genomes analysis demonstrated that the target genes of the DE­miRNAs were significantly enriched in PI3K/AKT/mTOR, class O of forkhead box transcription factors, mitogen­activated protein kinase (MAPK), neurotrophin and transforming growth factor (TGF)­ß signaling, as well as focal adhesion and the axon guidance pathway. PI3K, AKT, PTEN, MAPK1, Son of Sevenless, sphingosine­1­phosphate receptor 1, BCL­2L11, TGF­ß receptor type 1/2 and integrin α (ITGA)/ITGAB, which are key components of the aforementioned pathways and were revealed to be target genes of several of the DE­miRNAs. The present study used a linear model to identify several DE­miRNAs, as well as their target genes and associated pathways, which may serve crucial roles in mouse retinal development. Therefore, the results obtained in the present study may provide the groundwork for further experiments.