Multi-response Regression for Block-missing Multi-modal Data without Imputation.

Multi-modal data are prevalent in many scientific fields. In this study, we consider the parameter estimation and variable selection for a multi-response regression using block-missing multi-modal data. Our method allows the dimensions of both the responses and the predictors to be large, and the responses to be incomplete and correlated, a common practical problem in high-dimensional settings. Our proposed method uses two steps to make a prediction from a multi-response linear regression model with block-missing multi-modal predictors. In the first step, without imputing missing data, we use all available data to estimate the covariance matrix of the predictors and the cross-covariance matrix between the predictors and the responses. In the second step, we use these matrices and a penalized method to simultaneously estimate the precision matrix of the response vector, given the predictors, and the sparse regression parameter matrix. Lastly, we demonstrate the effectiveness of the proposed method using theoretical studies, simulated examples, and an analysis of a multi-modal imaging data set from the Alzheimer's Disease Neuroimaging Initiative.

Efficient computation of high-dimensional penalized generalized linear mixed models by latent factor modeling of the random effects.

Modern biomedical datasets are increasingly high-dimensional and exhibit complex correlation structures. Generalized linear mixed models (GLMMs) have long been employed to account for such dependencies. However, proper specification of the fixed and random effects in GLMMs is increasingly difficult in high dimensions, and computational complexity grows with increasing dimension of the random effects. We present a novel reformulation of the GLMM using a factor model decomposition of the random effects, enabling scalable computation of GLMMs in high dimensions by reducing the latent space from a large number of random effects to a smaller set of latent factors. We also extend our prior work to estimate model parameters using a modified Monte Carlo Expectation Conditional Minimization algorithm, allowing us to perform variable selection on both the fixed and random effects simultaneously. We show through simulation that through this factor model decomposition, our method can fit high-dimensional penalized GLMMs faster than comparable methods and more easily scale to larger dimensions not previously seen in existing approaches.

Desensitization and remission after peanut sublingual immunotherapy in 1- to 4-year-old peanut-allergic children: A randomized, placebo-controlled trial.

BACKGROUND: Prior studies of peanut sublingual immunotherapy (SLIT) have suggested a potential advantage with younger age at treatment initiation. OBJECTIVE: We studied the safety and efficacy of SLIT for peanut allergy in 1- to 4-year-old children. METHODS: Peanut-allergic 1- to 4-year-old children were randomized to receive 4 mg peanut SLIT versus placebo. Desensitization was assessed by double-blind, placebo-controlled food challenge (DBPCFC) after 36 months of treatment. Participants desensitized to at least 443 mg peanut protein discontinued therapy for 3 months and then underwent DBPCFC to assess for remission. Biomarkers were measured at baseline and longitudinally during treatment. RESULTS: Fifty participants (25 peanut SLIT, 25 placebo) with a median age of 2.4 years were enrolled across 2 sites. The primary end point of desensitization was met with actively treated versus placebo participants having a significantly greater median cumulative tolerated dose (4443 mg vs 143 mg), higher likelihood of passing the month 36 DBPCFC (60% vs 0), and higher likelihood of demonstrating remission (48% vs 0). The highest rate of desensitization and remission was seen in 1- to 2-year-olds, followed by 2- to 3-year-olds and 3- to 4-year-olds. Longitudinal changes in peanut skin prick testing, peanut-specific IgG4, and peanut-specific IgG4/IgE ratio were seen in peanut SLIT but not placebo participants. Oropharyngeal itching was more commonly reported by peanut SLIT than placebo participants. Skin, gastrointestinal, upper respiratory, lower respiratory, and multisystem adverse events were similar between treatment groups. CONCLUSION: Peanut SLIT safely induces desensitization and remission in 1- to 4-year-old children, with improved outcomes seen with younger age at initiation.

Adaptive Regularized Tri-Factor Non-Negative Matrix Factorization for Cell Type Deconvolution.

Motivation: Accurate deconvolution of cell types from bulk gene expression is crucial for understanding cellular compositions and uncovering cell-type specific differential expression and physiological states of diseased tissues. Existing deconvolution methods have limitations, such as requiring complete cellular gene expression signatures or neglecting partial biological information. Moreover, these methods often overlook varying cell-type mRNA amounts, leading to biased proportion estimates. Additionally, they do not effectively utilize valuable reference information from external studies, such as means and ranges of population cell-type proportions. Results: To address these challenges, we introduce an Adaptive Regularized Tri-factor non-negative matrix factorization approach for deconvolution (ARTdeConv). We rigorously establish the numerical convergence of our algorithm. Through benchmark simulations, we demonstrate the superior performance of ARTdeConv compared to state-of-the-art reference-free methods. In a real-world application, our method accurately estimates cell proportions, as evidenced by the nearly perfect Pearson's correlation between ARTdeConv estimates and flow cytometry measurements in a dataset from a trivalent influenza vaccine study. Moreover, our analysis of ARTdeConv estimates in COVID-19 patients reveals patterns consistent with important immunological phenomena observed in other studies. Availability and implementation: The proposed method, ARTdeConv, is implemented as an R package and can be accessed on GitHub for researchers and practitioners at https://github.com/gr8lawrence/ARTDeConv .

Adaptive Supervised Learning on Data Streams in Reproducing Kernel Hilbert Spaces with Data Sparsity Constraint.

Data are generated at an unprecedented rate and scale these days across many disciplines. The field of streaming data analysis has emerged as a result of new data collection and storage technologies in various areas, such as air pollution monitoring, detection of traffic congestion, disease surveillance, and recommendation systems. In this paper, we consider the problem of model estimation for data streams in reproducing kernel Hilbert spaces. We propose an adaptive supervised learning method with a data sparsity constraint that uses limited storage spaces and can handle non-stationary models. We demonstrate the competitive performance of the proposed method using simulations and analysis of the bike sharing dataset.

HIGH-DIMENSIONAL FACTOR REGRESSION FOR HETEROGENEOUS SUBPOPULATIONS.

In modern scientific research, data heterogeneity is commonly observed owing to the abundance of complex data. We propose a factor regression model for data with heterogeneous subpopulations. The proposed model can be represented as a decomposition of heterogeneous and homogeneous terms. The heterogeneous term is driven by latent factors in different subpopulations. The homogeneous term captures common variation in the covariates and shares common regression coefficients across subpopulations. Our proposed model attains a good balance between a global model and a group-specific model. The global model ignores the data heterogeneity, while the group-specific model fits each subgroup separately. We prove the estimation and prediction consistency for our proposed estimators, and show that it has better convergence rates than those of the group-specific and global models. We show that the extra cost of estimating latent factors is asymptotically negligible and the minimax rate is still attainable. We further demonstrate the robustness of our proposed method by studying its prediction error under a mis-specified group-specific model. Finally, we conduct simulation studies and analyze a data set from the Alzheimer's Disease Neuroimaging Initiative and an aggregated microarray data set to further demonstrate the competitiveness and interpretability of our proposed factor regression model.

Should All Patients With Pulmonary Hypertension Undergoing Non-Cardiac Surgery Be Managed by Cardiothoracic Fellowship-Trained Anesthesiologists?

OBJECTIVES: To identify differences in practice patterns and outcomes related to the induction of general anesthesia for patients with pulmonary hypertension (PH) performed by anesthesiologists who have completed a cardiothoracic fellowship (CTA group) vs those who have not (non-CTA group). DESIGN: Retrospective study with propensity score matching. SETTING: Operating room. PARTICIPANTS: All adult patients with PH undergoing general anesthesia requiring intubation at a single academic center over 5 years. INTERVENTIONS: Patient baseline characteristics, peri-induction management variables, post-induction mean arterial pressure (MAP), and other outcomes were compared between CTA and non-CTA groups. METHODS AND MAIN RESULTS: Following propensity scoring matching, 402 patients were included in the final model, 100 in the CTA group and 302 in the non-CTA group. Also following matching, only cases of mild to moderate PH without right ventricular dysfunction remained in the analysis. Matched groups were overall statistically similar with respect to baseline characteristics; however, there was a greater incidence of higher ASA class (P = .025) and cardiology and thoracic procedures (P < .001) being managed by the CTA group. No statistical differences were identified in practice patterns or outcomes related to the induction of anesthesia between groups, except for longer hospital length of stay in the CTA group (P = .008). CONCLUSIONS: These results provide early evidence to suggest the induction of general anesthesia of patients with non-severe PH disease can be comparably managed by either anesthesiologists with or without a cardiothoracic fellowship. However, these findings should be confirmed in a prospective study.

Development and Feasibility of a Primary Care Provider Training Intervention to Improve Atrial Fibrillation Management.

The disparities in atrial fibrillation (AF) care are partially attributed to inadequate access to providers with specialized training in AF. Primary care providers (PCPs) are often the sole providers of AF care in under-resourced regions. As such, we sought to create a virtual education intervention for PCPs and to evaluate its impact on the use of stroke risk reduction strategies in patients with AF. A multidisciplinary team mentored PCPs on AF management over 6 months using a virtual case-based training format. Surveys of participant knowledge and confidence in AF care were compared before and after the intervention. Hierarchical logistic regression modeling was used to evaluate change in oral anticoagulation (OAC) therapy in the patients seen by participants before or after training. Of 41 participants trained, 49% worked in family medicine, 41% internal medicine, and 10% general cardiology. Participants attended a mean of 14 1-hour sessions. Overall, the appropriate use of OAC (for CHA2DS2-VASc score ≥1 man, ≥2 women) increased from 37% to 46% (p <0.001) comparing the patients seen before (n = 1,739) versus after (n = 610) intervention. The factors independently associated with appropriate OAC use included participant training (odds ratio [OR] 1.4, p = 0.002) and participant competence in AF management. The factors associated with decreased OAC use included patient age (OR 0.8 per 10 year, p = 0.008) and nonwhite race (OR 0.7, p = 0.028). Provider knowledge and confidence in AF care improved (p <0.001). In conclusion, we show that a virtual PCP training intervention improves the use of stroke risk reduction therapy in outpatients with AF and could be a widely scalable intervention to improve AF care in under-resourced communities.

Uncovering Diverse Mechanistic Spreading Pathways in Disease Progression of Alzheimer's Disease.

Background: The AT[N] research framework focuses on three major biomarkers in Alzheimer's disease (AD): amyloid-ß deposition (A), pathologic tau (T), and neurodegeneration [N]. Objective: We hypothesize that the diverse mechanisms such as A⟶T and A⟶[N] pathways from one brain region to others, may underlie the wide variation in clinical symptoms. We aim to uncover the causal-like effect of regional AT[N] biomarkers on cognitive decline as well as the interaction with non-modifiable risk factors such as age and APOE4. Methods: We apply multi-variate statistical inference to uncover all possible mechanistic spreading pathways through which the aggregation of an upstream biomarker (e.g., increased amyloid level) in a particular brain region indirectly impacts cognitive decline, via the cascade build-up of a downstream biomarker (e.g., reduced metabolism level) in another brain region. Furthermore, we investigate the survival time for each identified region-to-region pathological pathway toward the AD onset. Results: We have identified a collection of critical brain regions on which the amyloid burdens exert an indirect effect on the decline in memory and executive function (EF) domain, being mediated by the reduction of metabolism level at other brain regions. APOE4 status has been found not only involved in many A⟶N mechanistic pathways but also significantly contributes to the risk of developing AD. Conclusion: Our major findings include 1) the region-to-region A⟶N⟶MEM and A⟶N⟶MEM pathways exhibit distinct spatial patterns; 2) APOE4 is significantly associated with both direct and indirect effects on the cognitive decline while sex difference has not been identified in the mediation analysis.

An Efficient Greedy Search Algorithm for High-dimensional Linear Discriminant Analysis.

High-dimensional classification is an important statistical problem that has applications in many areas. One widely used classifier is the Linear Discriminant Analysis (LDA). In recent years, many regularized LDA classifiers have been proposed to solve the problem of high-dimensional classification. However, these methods rely on inverting a large matrix or solving large-scale optimization problems to render classification rules-methods that are computationally prohibitive when the dimension is ultra-high. With the emergence of big data, it is increasingly important to develop more efficient algorithms to solve the high-dimensional LDA problem. In this paper, we propose an efficient greedy search algorithm that depends solely on closed-form formulae to learn a high-dimensional LDA rule. We establish theoretical guarantee of its statistical properties in terms of variable selection and error rate consistency; in addition, we provide an explicit interpretation of the extra information brought by an additional feature in a LDA problem under some mild distributional assumptions. We demonstrate that this new algorithm drastically improves computational speed compared with other high-dimensional LDA methods, while maintaining comparable or even better classification performance.

Open-label study of the efficacy, safety, and durability of peanut sublingual immunotherapy in peanut-allergic children.

BACKGROUND: Studies on the efficacy of peanut sublingual immunotherapy (SLIT) are limited. The durability of desensitization after SLIT has not been well described. OBJECTIVE: We sought to evaluate the efficacy and safety of 4-mg peanut SLIT and persistence of desensitization after SLIT discontinuation. METHODS: Challenge-proven peanut-allergic 1- to 11-year-old children were treated with open-label 4-mg peanut SLIT for 48 months. Desensitization after peanut SLIT was assessed by a 5000-mg double-blind, placebo-controlled food challenge (DBPCFC). A novel randomly assigned avoidance period of 1 to 17 weeks was followed by the DBPCFC. Skin prick test results immunoglobulin levels, basophil activation test results, TH1, TH2, and IL-10 cytokines were measured longitudinally. Safety was assessed through patient-reported home diaries. RESULTS: Fifty-four participants were enrolled and 47 (87%) completed peanut SLIT and the 48-month DBPCFC per protocol. The mean successfully consumed dose (SCD) during the DBPCFC increased from 48 to 2723 mg of peanut protein after SLIT (P < .0001), with 70% achieving clinically significant desensitization (SCD > 800 mg) and 36% achieving full desensitization (SCD = 5000 mg). Modeled median time to loss of clinically significant desensitization was 22 weeks. Peanut skin prick test; peanut-specific IgE, IgG4, and IgG4/IgE ratio; and peanut-stimulated basophil activation test, IL-4, IL-5, IL-13, IFN-γ, and IL-10 changed significantly compared with baseline, with changes seen as early as 6 months. Median rate of reaction per dose was 0.5%, with transient oropharyngeal itching being the most common, and there were no dosing symptoms requiring epinephrine. CONCLUSIONS: In this open-label, prospective study, peanut SLIT was safe and induced clinically significant desensitization in most of the children, lasting more than 17 weeks after discontinuation of therapy.

Testing generalized linear models with high-dimensional nuisance parameter.

Generalized linear models often have a high-dimensional nuisance parameters, as seen in applications such as testing gene-environment interactions or gene-gene interactions. In these scenarios, it is essential to test the significance of a high-dimensional sub-vector of the model's coefficients. Although some existing methods can tackle this problem, they often rely on the bootstrap to approximate the asymptotic distribution of the test statistic, and thus are computationally expensive. Here, we propose a computationally efficient test with a closed-form limiting distribution, which allows the parameter being tested to be either sparse or dense. We show that under certain regularity conditions, the type I error of the proposed method is asymptotically correct, and we establish its power under high-dimensional alternatives. Extensive simulations demonstrate the good performance of the proposed test and its robustness when certain sparsity assumptions are violated. We also apply the proposed method to Chinese famine sample data in order to show its performance when testing the significance of gene-environment interactions.

Mapping lesion-specific response and progression dynamics and inter-organ variability in metastatic colorectal cancer.

Achieving systemic tumor control across metastases is vital for long-term patient survival but remains intractable in many patients. High lesion-level response heterogeneity persists, conferring many dissociated responses across metastatic lesions. Most studies of metastatic disease focus on tumor molecular and cellular features, which are crucial to elucidating the mechanisms underlying lesion-level variability. However, our understanding of lesion-specific heterogeneity on the macroscopic level, such as lesion dynamics in growth, response, and progression during treatment, remains rudimentary. This study investigates lesion-specific response heterogeneity through analyzing 116,542 observations of 40,612 lesions in 4,308 metastatic colorectal cancer (mCRC) patients. Despite significant differences in their response and progression dynamics, metastatic lesions converge on four phenotypes that vary with anatomical site. Importantly, we find that organ-level progression sequence is closely associated with patient long-term survival, and that patients with the first lesion progression in the liver often have worse survival. In conclusion, our study provides insights into lesion-specific response and progression heterogeneity in mCRC and creates impetus for metastasis-specific therapeutics.

Decomposition of variation of mixed variables by a latent mixed Gaussian copula model.

Many biomedical studies collect data of mixed types of variables from multiple groups of subjects. Some of these studies aim to find the group-specific and the common variation among all these variables. Even though similar problems have been studied by some previous works, their methods mainly rely on the Pearson correlation, which cannot handle mixed data. To address this issue, we propose a latent mixed Gaussian copula (LMGC) model that can quantify the correlations among binary, ordinal, continuous, and truncated variables in a unified framework. We also provide a tool to decompose the variation into the group-specific and the common variation over multiple groups via solving a regularized M-estimation problem. We conduct extensive simulation studies to show the advantage of our proposed method over the Pearson correlation-based methods. We also demonstrate that by jointly solving the M-estimation problem over multiple groups, our method is better than decomposing the variation group by group. We also apply our method to a Chlamydia trachomatis genital tract infection study to demonstrate how it can be used to discover informative biomarkers that differentiate patients.

glmmPen: High Dimensional Penalized Generalized Linear Mixed Models.

Generalized linear mixed models (GLMMs) are widely used in research for their ability to model correlated outcomes with non-Gaussian conditional distributions. The proper selection of fixed and random effects is a critical part of the modeling process, where model misspecification may lead to significant bias. However, the joint selection of fixed and random effects has historically been limited to lower dimensional GLMMs, largely due to the use of criterion-based model selection strategies. Here we present the R package glmmPen, one of the first to select fixed and random effects in higher dimension using a penalized GLMM modeling framework. Model parameters are estimated using a Monte Carlo expectation conditional minimization (MCECM) algorithm, which leverages Stan and RcppArmadillo for increased computational efficiency. Our package supports the Binomial, Gaussian, and Poisson families and multiple penalty functions. In this manuscript we discuss the modeling procedure, estimation scheme, and software implementation through application to a pancreatic cancer subtyping study. Simulation results show our method has good performance in selecting both the fixed and random effects in high dimensional GLMMs.

Quality of Recovery After Rotator Cuff Repair With Interscalene Liposomal Bupivacaine Versus Interscalene Nerve Catheter.

Background: Interscalene nerve catheters have been proven to be effective in managing pain after rotator cuff repair (RCR) surgery. Liposomal bupivacaine is a newer approved therapy for use around the interscalene brachial plexus, but its analgesic efficacy has limited supporting data in various patient populations. Purpose/Hypothesis: The purpose of this study was to investigate the quality of recovery after arthroscopic RCR in patients who received either single-injection interscalene liposomal bupivacaine or an interscalene peripheral nerve catheter. It was hypothesized that interscalene peripheral nerve catheters would provide more reliable analgesia and improved patient satisfaction 48 hours after surgery. Study Design: Cohort study; Level of evidence, 2. Methods: Enrolled were 93 consecutive patients who underwent arthroscopic rotator cuff surgery at a single ambulatory surgery center between October 2020 and June 2021. Of these patients, 13 were lost to follow-up; thus, 80 patients were included in statistical analysis. One group of patients (n = 48) received a preoperative interscalene nerve block placed with 10 mL 0.5% bupivacaine and 10 mL 1.3% liposomal bupivacaine. The second group (n = 32) received a preoperative interscalene catheter with an initial bolus of 20 mL 0.25% bupivacaine and a 0.2% ropivacaine infusion by an elastomeric pump set at 10 mL/hr for 48 hours. The primary outcome was the difference between preoperative and 48-hour postoperative quality of recovery-15 (QoR-15) scores. Secondary outcomes included visual analog pain scores, opioid use, and patient satisfaction. Complications and adverse effects were also noted. The Kruskal-Wallis test was used to analyze means and standard deviations for continuous endpoints; Fisher exact test was used to analyze counts and proportions for categorical endpoints. Results: The liposomal bupivacaine group had a mean reduction of 3.9 in their postoperative QoR-15 scores, and the catheter group had a mean reduction of 25.1 in their postoperative QoR-15 scores, indicating a significantly worse functional recovery period compared with liposomal bupivacaine within the first 48 hours (P < .001). Patients who received liposomal bupivacaine also had significantly lower pain scores on the second postoperative day, improved quality of sleep, and improved satisfaction with analgesia (P < .05 for all). Conclusion: The use of interscalene liposomal bupivacaine demonstrated significantly improved quality of recovery when compared with interscalene nerve catheter after RCR.

Predictors of Post-induction Hypotension for Patients With Pulmonary Hypertension.

Purpose The purpose is to identify predictors of post-induction hypotension (PIH) during general anesthesia in a population of patients with varying degrees of pulmonary hypertension (PH). Methods This is a single-center, retrospective, observational study of perioperative data obtained via electronic health records from patients with PH undergoing surgery over a five-year period. Baseline patient characteristics, peri-induction management variables, and pre-induction mean arterial pressure (MAP) were statistically analyzed using Kruskal-Wallis rank sum tests, Pearson's chi-squared tests, and logistic regression analysis to identify risk factors for PIH. We further assessed the relationship between PH and PIH using propensity score matching. Primary outcomes include a percent decrease in post-induction blood pressure as well as a post-induction nadir with a threshold of 55 mm Hg. Results Eight hundred fifty-seven patients in the cohort stratified by severity of PH reveal that advanced age (p < 0.001), higher BMI (P = 0.002), higher American Society of Anesthesiologists (ASA) score (P = 0.001), and renal and cardiac comorbidities (P < 0.001) are associated with PH severity. None of our tested parameters were significantly predictive for PIH in patients with PH. Right heart failure was found to be weakly and non-significantly predictive of PIH in patients with PH (P = 0.052, odds ratio [OR] = 1.116). Diabetes (P = 0.007, OR = 0.919) and maintenance of spontaneous ventilation (P = 0.012, OR = 0.925) were associated with decreased rates of PIH. Conclusion Hypotension after induction of general anesthesia in patients with PH is a serious problem, yet statistically significant risk factors were not identified. History of diabetes and preservation of spontaneous ventilation had a significant but weak effect of decreasing rates of PIH. This pilot study was limited by retrospective design and warrants further analysis with a prospective cohort.

Hypothermia and Prolonged Time From Procedure End to Extubation After Endovascular Thoracic Aortic Surgery.

OBJECTIVE: Perioperative hypothermia (core temperature <36°C) occurs in 50%-to-80% of patients recovering from thoracic aortic surgery, though its effects have not been described fully in this context. The authors, therefore, sought to characterize the incidence of perioperative hypothermia and its association with time from procedure end to extubation in endovascular aortic surgical patients. DESIGN: A retrospective cohort study. SETTING: At a single academic tertiary center. PARTICIPANTS: Patients recovering from thoracic aortic surgery with lumbar drains. INTERVENTION: None. MEASUREMENTS AND MAIN RESULTS: A total of 196 patients were included in this study, 55 of whom were hypothermic with temperatures <35.0°C at the end of surgery. Though the unadjusted time to extubation was not statistically different in the hypothermic group (median 8 minutes, IQR 5-13.5 minutes) compared to the normothermic group (median 7 minutes, IQR 4-12 minutes; p = 0.062), multivariate predictors of increased time from procedure end to extubation included hypothermia (p = 0.011), age (p = 0.009), diabetes (p = 0.015), history of carotid disease (p = 0.040), and crystalloid volume (p = 0.019). CONCLUSIONS: Hypothermia in patients recovering from endovascular aortic surgery was associated with prolonged time from procedure end to extubation. Because of the retrospective observational nature of the authors' analysis, it was not possible to determine the extent to which prolonged mechanical ventilation was influenced by low temperature.

Sleep, cardiovascular risk factors, and kidney function: The Multi-Ethnic Study of Atherosclerosis (MESA).

OBJECTIVES: Examine the associations of sleep measures with kidney function changes over time among individuals from a community-based study. METHODS: The sample includes 1657 participants (287 with chronic kidney disease [CKD]) in the Multi-Ethnic Study of Atherosclerosis Sleep Cohort (mean age: 57.7 years, male: 46.0%). We examined associations between a large set of sleep variables (polysomnography, actigraphy, and questionnaires) and cardiovascular disease risk factors and changes in estimated glomerular filtration rate (eGFR) and urinary albumin-to-creatinine ratio over approximately 5 years using high-dimensional regression. We investigated the modifying effect of sleep on the associations between cardiovascular disease risk factors and kidney function. RESULTS: Sleep metrics predicted kidney function decline only among individuals with baseline CKD. Among this group, eGFR decline was associated with decreased stage N3 sleep (0.32 mL/min/1.73 m2/y per 10% decrease in N3, p < .001); increased actigraphy napping frequency (beta: -0.20 [-0.30, -0.07]); and actigraphy sleep midpoint trajectory in early morning (ref: midnight, beta: -0.84 [-1.19, -0.50]). Urinary albumin-to-creatinine ratio increase was associated with high wake bouts trajectory (ref: low, beta: 0.97 [0.28, 1.67]) and increased sleep-related hypoxemia (oxygen saturation %time<90 [≥5%], beta: 2.17 [1.26, 3.08]). Sleep metrics--N3 sleep, naps, and midpoint trajectory--significantly modified associations between hemoglobin A1C and eGFR decline. CONCLUSIONS: Reduced deep sleep, daytime napping, increased wake bouts, delayed sleep rhythms, and overnight hypoxemia are associated with longitudinal kidney function decline, with effects most apparent in individuals with CKD. Deep sleep, napping, and sleep timing modified the association between hemoglobin A1C and kidney function.