Browsing by Author "Yang, Qing"
Results Per Page
Sort Options
Item Open Access A latent factor linear mixed model for high-dimensional longitudinal data analysis.(Statistics in medicine, 2013-10) An, Xinming; Yang, Qing; Bentler, Peter MHigh-dimensional longitudinal data involving latent variables such as depression and anxiety that cannot be quantified directly are often encountered in biomedical and social sciences. Multiple responses are used to characterize these latent quantities, and repeated measures are collected to capture their trends over time. Furthermore, substantive research questions may concern issues such as interrelated trends among latent variables that can only be addressed by modeling them jointly. Although statistical analysis of univariate longitudinal data has been well developed, methods for modeling multivariate high-dimensional longitudinal data are still under development. In this paper, we propose a latent factor linear mixed model (LFLMM) for analyzing this type of data. This model is a combination of the factor analysis and multivariate linear mixed models. Under this modeling framework, we reduced the high-dimensional responses to low-dimensional latent factors by the factor analysis model, and then we used the multivariate linear mixed model to study the longitudinal trends of these latent factors. We developed an expectation-maximization algorithm to estimate the model. We used simulation studies to investigate the computational properties of the expectation-maximization algorithm and compare the LFLMM model with other approaches for high-dimensional longitudinal data analysis. We used a real data example to illustrate the practical usefulness of the model.Item Open Access Algorithm-hardware co-optimization for neural network efficiency improvement(2020) Yang, QingDeep neural networks (DNNs) are tremendously applied in the artificial intelligence field. While the performance of DNNs is continuously improved by more complicated and deeper structures, the feasibility of deployment on edge devices remains a critical problem. In this thesis, we present algorithm-hardware co-optimization approaches to address the challenges of efficient DNN deployments from three aspects: 1) save computational cost, 2) save memory cost, and 3) save data movements.
First, we present a joint regularization technique to advance the compression beyond the weights to neuron activations. By distinguishing and leveraging the significant difference among neuron responses and connections during learning, the jointly pruned network, namely JPnet, optimizes the sparsity of activations and weights. Second, to structurally regulate the dynamic activation sparsity (DAS), we propose a generic low-cost approach based on winners-take-all (WTA) dropout technique. The network enhanced by the proposed WTA dropout, namely DASNet, features structured activation sparsity with an improved sparsity level, which can be easily utilized to achieve acceleration on conventional embedded systems. The effectiveness of JPNet and DASNet has been thoroughly evaluated through various network models with different activation functions and on different datasets. Third, we propose BitSystolic, a neural processing unit based on a systolic array structure, to fully support the mixed-precision inference. In BitSystolic, the numerical precision of both weights and activations can be configured in the range of 2b~8b, fulfilling different requirements across mixed-precision models and tasks. Moreover, the design can support various data flows presented in different types of neural layers and adaptively optimize the data reuse by switching between the matrix-matrix mode and vector-matrix mode. We designed and fabricated the proposed BitSystolic in the 65nm process. Our measurement results show that BitSystolic features the unified power efficiency of up to 26.7 TOPS/W with 17.8 mW peak power consumption across various layer types. In the end, we will have a glance at computing-in-memory architectures based on resistive random-access memory (ReRAM) which realizes in-place storage and computation. A quantized training method is proposed to enhance the accuracy of neuromorphic systems based on ReRAM by alleviating the impact of limited parameter precision.
Item Open Access Are quit attempts among U.S. female nurses who smoke different from female smokers in the general population? An analysis of the 2006/2007 tobacco use supplement to the current population survey.(BMC Womens Health, 2012-03-19) Sarna, Linda; Bialous, Stella Aguinaga; Nandy, Karabi; Yang, QingBACKGROUND: Smoking is a significant women's health issue. Examining smoking behaviors among occupational groups with a high prevalence of women may reveal the culture of smoking behavior and quit efforts of female smokers. The purpose of this study was to examine how smoking and quitting characteristics (i.e., ever and recent quit attempts) among females in the occupation of nursing are similar or different to those of women in the general population. METHODS: Cross-sectional data from the Tobacco Use Supplement of the Current Population Survey 2006/2007 were used to compare smoking behaviors of nurses (n = 2, 566) to those of non-healthcare professional women (n = 93, 717). Smoking characteristics included years of smoking, number of cigarettes, and time to first cigarette with smoking within the first 30 minutes as an indicator of nicotine dependence. Logistic regression models using replicate weights were used to determine correlates of ever and previous 12 months quit attempts. RESULTS: Nurses had a lower smoking prevalence than other women (12.1% vs 16.6%, p < 0.0001); were more likely to have ever made a quit attempt (77% vs 68%, p = 0.0002); but not in the previous 12 months (42% vs 43%, p = 0.77). Among those who ever made a quit attempt, nurses who smoked within 30 minutes of waking, were more likely to have made a quit attempt compared to other women (OR = 3.1, 95% CI: 1.9, 5.1). When considering quit attempts within the last 12 months, nurses whose first cigarette was after 30 minutes of waking were less likely to have made a quit attempt compared to other females (OR = 0.69, 95% CI: 0.49, 0.98). There were no other significant differences in ever/recent quitting. CONCLUSIONS: Smoking prevalence among female nurses was lower than among women who were not in healthcare occupations, as expected. The lack of difference in recent quit efforts among female nurses as compared to other female smokers has not been previously reported. The link between lower level of nicotine dependence, as reflected by the longer time to first cigarette, and lower quit attempts among nurses needs further exploration.Item Open Access Characterizing epigenetic aging in an adult sickle cell disease cohort.(Blood advances, 2024-01) Lê, Brandon M; Hatch, Daniel; Yang, Qing; Shah, Nirmish; Luyster, Faith S; Garrett, Melanie E; Tanabe, Paula; Ashley-Koch, Allison E; Knisely, Mitchell RAbstract
Sickle cell disease (SCD) affects ∼100 000 predominantly African American individuals in the United States, causing significant cellular damage, increased disease complications, and premature death. However, the contribution of epigenetic factors to SCD pathophysiology remains relatively unexplored. DNA methylation (DNAm), a primary epigenetic mechanism for regulating gene expression in response to the environment, is an important driver of normal cellular aging. Several DNAm epigenetic clocks have been developed to serve as a proxy for cellular aging. We calculated the epigenetic ages of 89 adults with SCD (mean age, 30.64 years; 60.64% female) using 5 published epigenetic clocks: Horvath, Hannum, PhenoAge, GrimAge, and DunedinPACE. We hypothesized that in chronic disease, such as SCD, individuals would demonstrate epigenetic age acceleration, but the results differed depending on the clock used. Recently developed clocks more consistently demonstrated acceleration (GrimAge, DunedinPACE). Additional demographic and clinical phenotypes were analyzed to explore their association with epigenetic age estimates. Chronological age was significantly correlated with epigenetic age in all clocks (Horvath, r = 0.88; Hannum, r = 0.89; PhenoAge, r = 0.85; GrimAge, r = 0.88; DunedinPACE, r = 0.34). The SCD genotype was associated with 2 clocks (PhenoAge, P = .02; DunedinPACE, P < .001). Genetic ancestry, biological sex, β-globin haplotypes, BCL11A rs11886868, and SCD severity were not associated. These findings, among the first to interrogate epigenetic aging in adults with SCD, demonstrate epigenetic age acceleration with recently developed epigenetic clocks but not older-generation clocks. Further development of epigenetic clocks may improve their predictive ability and utility for chronic diseases such as SCD.Item Open Access EXpanding Technology-Enabled, Nurse-Delivered Chronic Disease Care (EXTEND): Protocol and Baseline Data for a Randomized Trial.(Contemporary clinical trials, 2024-08) German, Jashalynn; Yang, Qing; Hatch, Daniel; Lewinski, Allison; Bosworth, Hayden B; Kaufman, Brystana G; Chatterjee, Ranee; Pennington, Gina; Matters, Doreen; Lee, Donghwan; Urlichich, Diana; Kokosa, Sarah; Canupp, Holly; Gregory, Patrick; Roberson, Cindy Leslie; Smith, Benjamin; Huber, Sherry; Doukellis, Katheryn; Deal, Tammi; Burns, Rose; Crowley, Matthew J; Shaw, Ryan JBackground
Approximately 10-15 % of individuals with type 2 diabetes have persistently poorly-controlled diabetes mellitus (PPDM) despite receiving available care, and frequently have comorbid hypertension. Mobile monitoring-enabled telehealth has the potential to improve outcomes in treatment-resistant chronic disease by supporting self-management and facilitating patient-clinician contact but must be designed in a manner amenable to real-world use.Methods
Expanding Technology-Enabled, Nurse-Delivered Chronic Disease Care (EXTEND) is an ongoing randomized trial comparing two 12-month interventions for comorbid PPDM and hypertension: 1) EXTEND, a mobile monitoring-enabled self-management intervention; and 2) EXTEND Plus, a comprehensive, nurse-delivered telehealth program incorporating mobile monitoring, self-management support, and pharmacist-supported medication management. Both arms leverage a novel platform that uses existing technological infrastructure to enable transmission of patient-generated health data into the electronic health record. The primary study outcome is difference in HbA1c change from baseline to 12 months. Secondary outcomes include blood pressure, weight, implementation barriers/facilitators, and costs.Results
Enrollment concluded in June 2023 following randomization of 220 patients. Baseline characteristics are similar between arms; mean age is 54.5 years, and the cohort is predominantly female (63.6 %) and Black (68.2 %), with a baseline HbA1c of 9.81 %.Conclusion
The EXTEND trial is evaluating two mobile monitoring-enabled telehealth approaches that seek to improve outcomes for patients with PPDM and hypertension. Critically, these approaches are designed around existing infrastructure, so may be amenable to implementation and scaling. This study will promote real-world use of telehealth to maximize benefits for those with high-risk chronic disease.Item Open Access Joint Inference for Competing Risks Survival Data(Journal of the American Statistical Association, 2016-07-02) Li, Gang; Yang, Qing© 2016 American Statistical Association. This article develops joint inferential methods for the cause-specific hazard function and the cumulative incidence function of a specific type of failure to assess the effects of a variable on the time to the type of failure of interest in the presence of competing risks. Joint inference for the two functions are needed in practice because (i) they describe different characteristics of a given type of failure, (ii) they do not uniquely determine each other, and (iii) the effects of a variable on the two functions can be different and one often does not know which effects are to be expected. We study both the group comparison problem and the regression problem. We also discuss joint inference for other related functions. Our simulation shows that our joint tests can be considerably more powerful than the Bonferroni method, which has important practical implications to the analysis and design of clinical studies with competing risks data. We illustrate our method using a Hodgkin disease data and a lymphoma data. Supplementary materials for this article are available online.Item Open Access Predicting health outcomes with intensive longitudinal data collected by mobile health devices: a functional principal component regression approach.(BMC medical research methodology, 2024-03) Yang, Qing; Jiang, Meilin; Li, Cai; Luo, Sheng; Crowley, Matthew J; Shaw, Ryan JBackground
Intensive longitudinal data (ILD) collected in near real time by mobile health devices provide a new opportunity for monitoring chronic diseases, early disease risk prediction, and disease prevention in health research. Functional data analysis, specifically functional principal component analysis, has great potential to abstract trends in ILD but has not been used extensively in mobile health research.Objective
To introduce functional principal component analysis (fPCA) and demonstrate its potential applicability in estimating trends in ILD collected by mobile heath devices, assessing longitudinal association between ILD and health outcomes, and predicting health outcomes.Methods
fPCA and scalar-to-function regression models were reviewed. A case study was used to illustrate the process of abstracting trends in intensively self-measured blood glucose using functional principal component analysis and then predicting future HbA1c values in patients with type 2 diabetes using a scalar-to-function regression model.Results
Based on the scalar-to-function regression model results, there was a slightly increasing trend between daily blood glucose measures and HbA1c. 61% of variation in HbA1c could be predicted by the three preceding months' blood glucose values measured before breakfast (P < 0.0001, [Formula: see text]).Conclusions
Functional data analysis, specifically fPCA, offers a unique tool to capture patterns in ILD collected by mobile health devices. It is particularly useful in assessing longitudinal dynamic association between repeated measures and outcomes, and can be easily integrated in prediction models to improve prediction precision.Item Open Access Relationship between hospital performance measures and outcomes in patients with acute ischaemic stroke: a prospective cohort study.(BMJ open, 2018-08) Zhang, Xinmiao; Li, Zixiao; Zhao, Xingquan; Xian, Ying; Liu, Liping; Wang, Chunxue; Wang, Chunjuan; Li, Hao; Prvu Bettger, Janet; Yang, Qing; Wang, David; Jiang, Yong; Bao, Xiaolei; Yang, Xiaomeng; Wang, Yilong; Wang, YongjunOBJECTIVE:Evidence-based performance measures have been increasingly used to evaluate hospital quality of stroke care, but their impact on stroke outcomes has not been verified. We aimed to evaluate the correlations between hospital performance measures and outcomes among patients with acute ischaemic stroke in a Chinese population. METHODS:Data were derived from a prospective cohort, which included 120 hospitals participating in the China National Stroke Registry between September 2007 and August 2008. Adherence to nine evidence-based performance measures was examined, and the composite score of hospital performance measures was calculated. The primary stroke outcomes were hospital-level, 30-day and 1-year risk-standardised mortality (RSM). Associations of individual performance measures and composite score with stroke outcomes were assessed using Spearman correlation coefficients. RESULTS:One hundred and twenty hospitals that recruited 12 027 patients with ischaemic stroke were included in our analysis. Among 12 027 patients, 61.59% were men, and the median age was 67 years. The overall composite score of performance measures was 63.3%. The correlation coefficients between individual performance measures ranged widely from 0.01 to 0.66. No association was observed between the composite score and 30-day RSM. The composite score was modestly associated with 1-year RSM (Spearman correlation coefficient, 0.34; p<0.05). The composite score explained only 2.53% and 10.18% of hospital-level variation in 30-day and 1-year RSM for patients with acute stroke. CONCLUSIONS:Adherence to evidence-based performance measures for acute ischaemic stroke was suboptimal in China. There were various correlations among hospital individual performance measures. The hospital performance measures had no correlations with 30-day RSM rate and modest correlations with 1-year RSM rate.Item Open Access Results of the CHlorhexidine Gluconate Bathing implementation intervention to improve evidence-based nursing practices for prevention of central line associated bloodstream infections Study (CHanGing BathS): a stepped wedge cluster randomized trial.(Implementation science : IS, 2021-04-26) Reynolds, Staci S; Woltz, Patricia; Keating, Edward; Neff, Janice; Elliott, Jennifer; Hatch, Daniel; Yang, Qing; Granger, Bradi BBackground
Central line-associated bloodstream infections (CLABSIs) result in approximately 28,000 deaths and approximately $2.3 billion in added costs to the U.S. healthcare system each year, and yet, many of these infections are preventable. At two large health systems in the southeast United States, CLABSIs continue to be an area of opportunity. Despite strong evidence for interventions to prevent CLABSI and reduce associated patient harm, such as use of chlorhexidine gluconate (CHG) bathing, the adoption of these interventions in practice is poor. The primary objective of this study was to assess the effect of a tailored, multifaceted implementation program on nursing staff's compliance with the CHG bathing process and electronic health record (EHR) documentation in critically ill patients. The secondary objectives were to examine the (1) moderating effect of unit characteristics and cultural context, (2) intervention effect on nursing staff's knowledge and perceptions of CHG bathing, and (3) intervention effect on CLABSI rates.Methods
A stepped wedged cluster-randomized design was used with units clustered into 4 sequences; each sequence consecutively began the intervention over the course of 4 months. The Grol and Wensing Model of Implementation helped guide selection of the implementation strategies, which included educational outreach visits and audit and feedback. Compliance with the appropriate CHG bathing process and daily CHG bathing documentation were assessed. Outcomes were assessed 12 months after the intervention to assess for sustainability.Results
Among the 14 clinical units participating, 8 were in a university hospital setting and 6 were in community hospital settings. CHG bathing process compliance and nursing staff's knowledge and perceptions of CHG bathing significantly improved after the intervention (p = .009, p = .002, and p = .01, respectively). CHG bathing documentation compliance and CLABSI rates did not significantly improve; however, there was a clinically significant 27.4% decrease in CLABSI rates.Conclusions
Using educational outreach visits and audit and feedback implementation strategies can improve adoption of evidence-based CHG bathing practices.Trial registration
ClinicalTrials.gov, NCT03898115 , Registered 28 March 2019.Item Open Access Sample size determination for jointly testing a cause-specific hazard and the all-cause hazard in the presence of competing risks.(Stat Med, 2017-12-27) Yang, Qing; Fung, Wing K; Li, GangThis article considers sample size determination for jointly testing a cause-specific hazard and the all-cause hazard for competing risks data. The cause-specific hazard and the all-cause hazard jointly characterize important study end points such as the disease-specific survival and overall survival, which are commonly used as coprimary end points in clinical trials. Specifically, we derive sample size calculation methods for 2-group comparisons based on an asymptotic chi-square joint test and a maximum joint test of the aforementioned quantities, taking into account censoring due to lost to follow-up as well as staggered entry and administrative censoring. We illustrate the application of the proposed methods using the Die Deutsche Diabetes Dialyse Studies clinical trial. An R package "powerCompRisk" has been developed and made available at the CRAN R library.Item Open Access Symptom Clusters of Midlife Menopausal Women with Metabolic Syndrome(2022) Min, Se HeeBackground: Midlife menopausal women with metabolic syndrome experience co-occurring symptoms that adversely affect their health outcomes. The purposes of this dissertation were to describe the symptom experience and presence of symptom clusters in midlife menopausal women with metabolic syndrome; to identify the number and types of symptom clusters and key symptoms based on symptom occurrence and severity dimension; and to identify the subgroups of midlife menopausal women with metabolic syndrome at high-risk for greater symptom cluster burden over time and their associated characteristics.Methods: A scoping review and two quantitative studies with cross-sectional and longitudinal approach using secondary data analysis were used in this dissertation. The Joanna Briggs Institute (JBI) Scoping Review methodology served as a guide for the scoping review. A total of eight articles were included and systematically evaluated. Network analysis was used to identify symptom clusters and key symptoms. Multi-trajectory analysis using latent class growth analysis was conducted to identify the high-risk subgroup of midlife menopausal women with metabolic syndrome for greater symptom cluster burden over time. Descriptive statistics was used to explain the demographic characteristics of each symptom cluster burden subgroup and bivariate analysis (analysis of variance, chi-square test) was conducted to examine the association between each symptom cluster burden subgroup and demographic characteristics. Results: Midlife menopausal women with metabolic syndrome experienced urogenital symptoms, vasomotor symptoms, psychological symptoms, sleep symptoms, and somatic symptoms. Urogenital symptoms were the most frequently assessed while sleep and somatic symptoms were the least frequently assessed. However, there were no current studies that examined the presence of symptom clusters in this population. The cross-sectional study using network analysis found that midlife menopausal women with metabolic syndrome experienced the psychological/somatic/genital cluster (key symptom: frequent mood change), the sleep/urinary cluster (sleep disturbance), and the vasomotor cluster (cold sweat) in the symptom occurrence dimension. In addition, they experienced the psychological/somatic/sexual cluster (anxiety), the sleep/urinary cluster (sleep disturbance), and the vasomotor/genital cluster (night sweat) in the symptom severity dimension. A total of four classes were identified with Class 1 (low symptom cluster burden), Class 2 and Class 3 (moderate symptom cluster burden), and Class 4 (high symptom cluster burden). Social support was a significant predictor of high symptom cluster burden subgroup. Conclusions and Implications: This dissertation is the first to identify the symptom clusters and key symptoms in midlife menopausal women with metabolic syndrome. In addition, this dissertation identified four subgroups of midlife menopausal women with metabolic syndrome based on their symptom cluster trajectory over time. This has allowed for an understanding of a high-risk subgroup for greater symptom cluster burden. Clinicians need to routinely assess symptom clusters and offer targeted symptom cluster interventions in clinical settings.