Leveraging machine learning: Covariate-adjusted Bayesian adaptive randomization and subgroup discovery in multi-arm survival trials.

Clinical trials evaluate the safety and efficacy of treatments for specific diseases. Ensuring these studies are well-powered is crucial for identifying superior treatments. With the rise of personalized medicine, treatment efficacy may vary based on biomarker profiles. However, researchers often lack prior knowledge about which biomarkers are linked to varied treatment effects. Fixed or response-adaptive designs may not sufficiently account for heterogeneous patient characteristics, such as genetic diversity, potentially reducing the chance of selecting the optimal treatment for individuals. Recent advances in Bayesian nonparametric modeling pave the way for innovative trial designs that not only maintain robust power but also offer the flexibility to identify subgroups deriving greater benefits from specific treatments. Building on this inspiration, we introduce a Bayesian adaptive design for multi-arm trials focusing on time-to-event endpoints. We introduce a covariate-adjusted response adaptive randomization, updating treatment allocation probabilities grounded on causal effect estimates using a random intercept accelerated failure time BART model. After the trial concludes, we suggest employing a multi-response decision tree to pinpoint subgroups with varying treatment impacts. The performance of our design is then assessed via comprehensive simulations.

Examining the Trajectory of Health-Related Quality of Life among Coronavirus Disease Patients.

BACKGROUND: Recent studies have reported a reduction in health-related quality of life (HR-QoL) among post-coronavirus disease 2019 (COVID-19) patients. However, there remains a gap in research examining the heterogeneity and determinants of HR-QoL trajectory in these patients. OBJECTIVE: To describe and identify factors explaining the variability in HR-QoL trajectories among a cohort of patients with history of COVID-19. DESIGN: A prospective study using data from a cohort of COVID-19 patients enrolled into a registry established at a health system in New York City. PARTICIPANTS: Participants were enrolled from July 2020 to June 2022, and completed a baseline evaluation and two follow-up visits at 6 and 12 months. METHODS: We assessed HR-QoL with the 29-item Patient Reported Outcomes Measurement Information System instrument, which was summarized into mental and physical health domains. We performed latent class growth and multinomial logistic regression to examine trajectories of HR-QoL and identify factors associated with specific trajectories. RESULTS: The study included 588 individuals with a median age of 52 years, 65% female, 54% White, 18% Black, and 18% Hispanic. We identified five physical health trajectories and four mental health trajectories. Female gender, having pre-existing hypertension, cardiovascular disease, asthma, and hospitalization for acute COVID-19 were independently associated with lower physical health. In addition, patients with increasing body mass index were more likely to experience lower physical health over time. Female gender, younger age, pre-existing asthma, arthritis and cardiovascular disease were associated with poor mental health. CONCLUSIONS: We found significant heterogeneity of HR-QoL after COVID-19, with women and patients with specific comorbidities at increased risk of lower HR-QoL. Implementation of targeted psychological and physical interventions is crucial for enhancing the quality of life of this patient population.

A new method for clustered survival data: Estimation of treatment effect heterogeneity and variable selection.

We recently developed a new method random-intercept accelerated failure time model with Bayesian additive regression trees (riAFT-BART) to draw causal inferences about population treatment effect on patient survival from clustered and censored survival data while accounting for the multilevel data structure. The practical utility of this method goes beyond the estimation of population average treatment effect. In this work, we exposit how riAFT-BART can be used to solve two important statistical questions with clustered survival data: estimating the treatment effect heterogeneity and variable selection. Leveraging the likelihood-based machine learning, we describe a way in which we can draw posterior samples of the individual survival treatment effect from riAFT-BART model runs, and use the drawn posterior samples to perform an exploratory treatment effect heterogeneity analysis to identify subpopulations who may experience differential treatment effects than population average effects. There is sparse literature on methods for variable selection among clustered and censored survival data, particularly ones using flexible modeling techniques. We propose a permutation-based approach using the predictor's variable inclusion proportion supplied by the riAFT-BART model for variable selection. To address the missing data issue frequently encountered in health databases, we propose a strategy to combine bootstrap imputation and riAFT-BART for variable selection among incomplete clustered survival data. We conduct an expansive simulation study to examine the practical operating characteristics of our proposed methods, and provide empirical evidence that our proposed methods perform better than several existing methods across a wide range of data scenarios. Finally, we demonstrate the methods via a case study of predictors for in-hospital mortality among severe COVID-19 patients and estimating the heterogeneous treatment effects of three COVID-specific medications. The methods developed in this work are readily available in the R ${\textsf {R}}$ package riAFTBART $\textsf {riAFTBART}$ .

Examining the efficacy of a cardio-dance intervention on brain health and the moderating role of ABCA7 in older African Americans: a protocol for a randomized controlled trial.

Introduction: African Americans are two to three times more likely to be diagnosed with Alzheimer's disease (AD) compared to White Americans. Exercise is a lifestyle behavior associated with neuroprotection and decreased AD risk, although most African Americans, especially older adults, perform less than the recommended 150 min/week of moderate-to-vigorous intensity exercise. This article describes the protocol for a Phase III randomized controlled trial that will examine the effects of cardio-dance aerobic exercise on novel AD cognitive and neural markers of hippocampal-dependent function (Aims #1 and #2) and whether exercise-induced neuroprotective benefits may be modulated by an AD genetic risk factor, ABCA7 rs3764650 (Aim #3). We will also explore the effects of exercise on blood-based biomarkers for AD. Methods and analysis: This 6-month trial will include 280 African Americans (≥ 60 years), who will be randomly assigned to 3 days/week of either: (1) a moderate-to-vigorous cardio-dance fitness condition or (2) a low-intensity strength, flexibility, and balance condition for 60 min/session. Participants will complete health and behavioral surveys, neuropsychological testing, saliva and venipuncture, aerobic fitness, anthropometrics and resting-state structural and functional neuroimaging at study entry and 6 months. Discussion: Results from this investigation will inform future exercise trials and the development of prescribed interventions that aim to reduce the risk of AD in African Americans.

Historical Redlining, Social Determinants of Health, and Stroke Prevalence in Communities in New York City.

Importance: Historical redlining was a discriminatory housing policy that placed financial services beyond the reach of residents in inner-city communities. The extent of the impact of this discriminatory policy on contemporary health outcomes remains to be elucidated. Objective: To evaluate the associations among historical redlining, social determinants of health (SDOH), and contemporary community-level stroke prevalence in New York City. Design, Setting, and Participants: An ecological, retrospective, cross-sectional study was conducted using New York City data from January 1, 2014, to December 31, 2018. Data from the population-based sample were aggregated on the census tract level. Quantile regression analysis and a quantile regression forests machine learning model were used to determine the significance and overall weight of redlining in relation to other SDOH on stroke prevalence. Data were analyzed from November 5, 2021, to January 31, 2022. Exposures: Social determinants of health included race and ethnicity, median household income, poverty, low educational attainment, language barrier, uninsurance rate, social cohesion, and residence in an area with a shortage of health care professionals. Other covariates included median age and prevalence of diabetes, hypertension, smoking, and hyperlipidemia. Weighted scores for historical redlining (ie, the discriminatory housing policy in effect from 1934 to 1968) were computed using the mean proportion of original redlined territories overlapped on 2010 census tract boundaries in New York City. Main Outcomes and Measures: Stroke prevalence was collected from the Centers for Disease Control and Prevention 500 Cities Project for adults 18 years and older from 2014 to 2018. Results: A total of 2117 census tracts were included in the analysis. After adjusting for SDOH and other relevant covariates, the historical redlining score was independently associated with a higher community-level stroke prevalence (odds ratio [OR], 1.02 [95% CI, 1.02-1.05]; P < .001). Social determinants of health that were positively associated with stroke prevalence included educational attainment (OR, 1.01 [95% CI, 1.01-1.01]; P < .001), poverty (OR, 1.01 [95% CI, 1.01-1.01]; P < .001), language barrier (OR, 1.00 [95% CI, 1.00-1.00]; P < .001), and health care professionals shortage (OR, 1.02 [95% CI, 1.00-1.04]; P = .03). Conclusions and Relevance: This cross-sectional study found that historical redlining was associated with modern-day stroke prevalence in New York City independently of contemporary SDOH and community prevalence of some relevant cardiovascular risk factors.

Estimating the causal effects of multiple intermittent treatments with application to COVID-19.

To draw real-world evidence about the comparative effectiveness of multiple time-varying treatments on patient survival, we develop a joint marginal structural survival model and a novel weighting strategy to account for time-varying confounding and censoring. Our methods formulate complex longitudinal treatments with multiple start/stop switches as the recurrent events with discontinuous intervals of treatment eligibility. We derive the weights in continuous time to handle a complex longitudinal dataset without the need to discretize or artificially align the measurement times. We further use machine learning models designed for censored survival data with time-varying covariates and the kernel function estimator of the baseline intensity to efficiently estimate the continuous-time weights. Our simulations demonstrate that the proposed methods provide better bias reduction and nominal coverage probability when analyzing observational longitudinal survival data with irregularly spaced time intervals, compared to conventional methods that require aligned measurement time points. We apply the proposed methods to a large-scale COVID-19 dataset to estimate the causal effects of several COVID-19 treatments on the composite of in-hospital mortality and ICU admission.

Using Tree-Based Machine Learning for Health Studies: Literature Review and Case Series.

Tree-based machine learning methods have gained traction in the statistical and data science fields. They have been shown to provide better solutions to various research questions than traditional analysis approaches. To encourage the uptake of tree-based methods in health research, we review the methodological fundamentals of three key tree-based machine learning methods: random forests, extreme gradient boosting and Bayesian additive regression trees. We further conduct a series of case studies to illustrate how these methods can be properly used to solve important health research problems in four domains: variable selection, estimation of causal effects, propensity score weighting and missing data. We exposit that the central idea of using ensemble tree methods for these research questions is accurate prediction via flexible modeling. We applied ensemble trees methods to select important predictors for the presence of postoperative respiratory complication among early stage lung cancer patients with resectable tumors. We then demonstrated how to use these methods to estimate the causal effects of popular surgical approaches on postoperative respiratory complications among lung cancer patients. Using the same data, we further implemented the methods to accurately estimate the inverse probability weights for a propensity score analysis of the comparative effectiveness of the surgical approaches. Finally, we demonstrated how random forests can be used to impute missing data using the Study of Women's Health Across the Nation data set. To conclude, the tree-based methods are a flexible tool and should be properly used for health investigations.

A Flexible Approach for Assessing Heterogeneity of Causal Treatment Effects on Patient Survival Using Large Datasets with Clustered Observations.

Personalized medicine requires an understanding of treatment effect heterogeneity. Evolving toward causal evidence for scenarios not studied in randomized trials necessitates a methodology using real-world evidence. Herein, we demonstrate a methodology that generates causal effects, assesses the heterogeneity of the effects and adjusts for the clustered nature of the data. This study uses a state-of-the-art machine learning survival model, riAFT-BART, to draw causal inferences about individual survival treatment effects, while accounting for the variability in institutional effects; further, it proposes a data-driven approach to agnostically (as opposed to a priori hypotheses) ascertain which subgroups exhibit an enhanced treatment effect from which intervention, relative to global evidence-average treatment effects measured at the population level. Comprehensive simulations show the advantages of the proposed method in terms of bias, efficiency and precision in estimating heterogeneous causal effects. The empirically validated method was then used to analyze the National Cancer Database.

A flexible approach for causal inference with multiple treatments and clustered survival outcomes.

When drawing causal inferences about the effects of multiple treatments on clustered survival outcomes using observational data, we need to address implications of the multilevel data structure, multiple treatments, censoring, and unmeasured confounding for causal analyses. Few off-the-shelf causal inference tools are available to simultaneously tackle these issues. We develop a flexible random-intercept accelerated failure time model, in which we use Bayesian additive regression trees to capture arbitrarily complex relationships between censored survival times and pre-treatment covariates and use the random intercepts to capture cluster-specific main effects. We develop an efficient Markov chain Monte Carlo algorithm to draw posterior inferences about the population survival effects of multiple treatments and examine the variability in cluster-level effects. We further propose an interpretable sensitivity analysis approach to evaluate the sensitivity of drawn causal inferences about treatment effect to the potential magnitude of departure from the causal assumption of no unmeasured confounding. Expansive simulations empirically validate and demonstrate good practical operating characteristics of our proposed methods. Applying the proposed methods to a dataset on older high-risk localized prostate cancer patients drawn from the National Cancer Database, we evaluate the comparative effects of three treatment approaches on patient survival, and assess the ramifications of potential unmeasured confounding. The methods developed in this work are readily available in the R $$ \mathsf{R}\kern.15em $$ package riAFTBART $$ \mathsf{riAFTBART} $$ .

Correlates of cancer prevalence across census tracts in the United States: A Bayesian machine learning approach.

Preventive measures, health behaviors, environmental exposures, and sociodemographic characteristics affect individual-level cancer risks. It is unclear how they influence neighborhood-level cancer risks. We developed a large-scale neighborhood health dataset for 72,337 census tracts in the United States by combining data from three publicly available sources. We used Bayesian additive regression trees to identify the most important predictors of tract-level cancer prevalence among adults (age ≥18 years), and examined their impact on cancer prevalence using partial dependence plots. The five most important census tract-level correlates of cancer prevalence were the proportion of population who were aged 65 years and older, had routine checkup and were non-Hispanic White, the proportion of houses built before 1960, and the proportion of population living below the poverty line. The identified predictors of neighborhood-level cancer prevalence may inform public health practitioners and policymakers to prioritize the improvement of environmental and neighborhood factors in reducing the cancer burden.

Immunotherapy Outcomes in Individuals With Non-Small Cell Lung Cancer and Poor Performance Status.

BACKGROUND: Patients with non-small cell lung cancer (NSCLC) treated in real-world practice typically have worse performance status (PS) compared with clinical trial patients, and the effectiveness of immunotherapy in this population in unknown. In this study, we assessed the effectiveness of standard of care immunotherapy for the first-line treatment of stage IV patients with NSCLC with Eastern Cooperative Oncology Group (ECOG) PS greater than or equal to 2. METHODS: We selected ECOG PS greater than or equal to 2 patients from real-world oncology data from a deidentified database and included them if they were diagnosed with stage IV NSCLC and had documented Programmed death-ligand 1 [PD-(L)1] expression greater than 0. Patients with tumor PD-(L)1 expression of at least 50% treated with pembrolizumab monotherapy were compared with those who did not have any documented treatment. Patients with tumor PD-(L)1 expression less than 50% treated with pembrolizumab and chemotherapy were compared with those treated with pembrolizumab monotherapy and those without documented treatment. RESULTS: In our propensity score-adjusted analysis, patients with ECOG PS of at least 2 and tumor PD-(L)1 expression of at least 50% treated with pembrolizumab monotherapy had statistically significantly better real-world overall survival compared with those without documented treatment (adjusted hazard ratio [HR] = 0.39, 95% confidence internal [CI] = 0.32 to 0.47). For patients with tumor PD-(L)1 expression less than 50%, there was also a statistically significant real-world overall survival benefit for those who received treatment either with combination pembrolizumab plus chemotherapy (adjusted HR = 0.39, 95% CI = 0.32 to 0.46) or pembrolizumab monotherapy (adjusted HR = 0.55, 95% CI = 0.41 to 0.70) compared with patients receiving no documented treatment. CONCLUSIONS: Among a highly representative sample of patients with advanced NSCLC and poor PS, our findings suggest that immunotherapy may provide an important survival benefit in individuals with high PD-(L)1-expressing tumors and in conjunction with chemotherapy in tumors with low PD-(L)1 expression.

A flexible approach for variable selection in large-scale healthcare database studies with missing covariate and outcome data.

BACKGROUND: Prior work has shown that combining bootstrap imputation with tree-based machine learning variable selection methods can provide good performances achievable on fully observed data when covariate and outcome data are missing at random (MAR). This approach however is computationally expensive, especially on large-scale datasets. METHODS: We propose an inference-based method, called RR-BART, which leverages the likelihood-based Bayesian machine learning technique, Bayesian additive regression trees, and uses Rubin's rule to combine the estimates and variances of the variable importance measures on multiply imputed datasets for variable selection in the presence of MAR data. We conduct a representative simulation study to investigate the practical operating characteristics of RR-BART, and compare it with the bootstrap imputation based methods. We further demonstrate the methods via a case study of risk factors for 3-year incidence of metabolic syndrome among middle-aged women using data from the Study of Women's Health Across the Nation (SWAN). RESULTS: The simulation study suggests that even in complex conditions of nonlinearity and nonadditivity with a large percentage of missingness, RR-BART can reasonably recover both prediction and variable selection performances, achievable on the fully observed data. RR-BART provides the best performance that the bootstrap imputation based methods can achieve with the optimal selection threshold value. In addition, RR-BART demonstrates a substantially stronger ability of detecting discrete predictors. Furthermore, RR-BART offers substantial computational savings. When implemented on the SWAN data, RR-BART adds to the literature by selecting a set of predictors that had been less commonly identified as risk factors but had substantial biological justifications. CONCLUSION: The proposed variable selection method for MAR data, RR-BART, offers both computational efficiency and good operating characteristics and is utilitarian in large-scale healthcare database studies.

Prospective Cohort Study to Investigate the Safety of Preoperative Tumor Necrosis Factor Inhibitor Exposure in Patients With Inflammatory Bowel Disease Undergoing Intra-abdominal Surgery.

BACKGROUND & AIMS: Whether preoperative treatment of inflammatory bowel disease (IBD) with tumor necrosis factor inhibitors (TNFis) increases the risk of postoperative infectious complications remains controversial. The primary aim of this study was to determine whether preoperative exposure to TNFis is an independent risk factor for postoperative infectious complications within 30 days of surgery. METHODS: We conducted a multicenter prospective observational study of patients with IBD undergoing intra-abdominal surgery across 17 sites from the Crohn's & Colitis Foundation Clinical Research Alliance. Infectious complications were categorized as surgical site infections (SSIs) or non-SSIs. Current TNFi exposure was defined as use within 12 weeks of surgery, and serum was collected for drug-level analyses. Multivariable models for occurrence of the primary outcome, any infection, or SSI were adjusted by predefined covariates (age, sex, preoperative steroid use, and disease type), baseline variables significantly associated (P < .05) with any infection or SSI separately, and TNFi exposure status. Exploratory models used TNFi exposure based on serum drug concentration. RESULTS: A total of 947 patients were enrolled from September 2014 through June 2017. Current TNFi exposure was reported by 382 patients. Any infection (18.1% vs 20.2%, P = .469) and SSI (12.0% vs 12.6%, P = .889) rates were similar in patients currently exposed to TNFis and those unexposed. In multivariable analysis, current TNFi exposure was not associated with any infection (odds ratio, 1.050; 95% confidence interval, 0.716-1.535) or SSI (odds ratio, 1.249; 95% confidence interval, 0.793-1.960). Detectable TNFi drug concentration was not associated with any infection or SSI. CONCLUSIONS: Preoperative TNFi exposure was not associated with postoperative infectious complications in a large prospective multicenter cohort.

A Role for CXCR3 Ligands as Biomarkers of Post-Operative Crohn's Disease Recurrence.

BACKGROUND AND AIMS: Crohn's disease [CD] recurrence following ileocolic resection [ICR] is common. We sought to identify blood-based biomarkers associated with CD recurrence. METHODS: CD patients undergoing ICR were recruited across six centres. Serum samples were obtained at post-operative colonoscopy. A multiplex immunoassay was used to analyse 92 inflammation-related proteins [Olink Proteomics]. Bayesian analysis was used to identify proteins associated with increasing Rutgeerts score. Identified proteins were used in receiver operating characteristic [ROC] analysis to examine the ability to identify CD recurrence [Rutgeerts score ≥i2]. Existing single cell data were interrogated to further elucidate the role of the identified proteins. RESULTS: Data from 276 colonoscopies in 213 patients were available. Median time from surgery to first and second colonoscopy was 7 (interquartile range [IQR] 6-9) and 19 [IQR 16-23] months, respectively. Disease recurrence was evident at 60 [30%] first and 36 [49%] second colonoscopies. Of 14 proteins significantly associated with Rutgeerts score, the strongest signal was seen for CXCL9 and MMP1. Among patients on anti-tumour necrosis factor drugs, CXCL9 and CXCL11 were most strongly associated with Rutgeerts score. Both are CXCR3 ligands. Incorporation of identified proteins into ROC analysis improved the ability to identify disease recurrence as compared to C-reactive protein alone: area under the curve [AUC] 0.75 (95% confidence interval [CI]: 0.66-0.82] vs 0.64 [95% CI 0.56-0.72], p = 0.012. Single cell transcriptomic data provide evidence that innate immune cells are the primary source of the identified proteins. CONCLUSIONS: CXCR3 ligands are associated with CD recurrence following ICR. Incorporation of novel blood-based candidate biomarkers may aid in identification of CD recurrence.

Determinants of Total End-of-Life Health Care Costs of Medicare Beneficiaries: A Quantile Regression Forests Analysis.

BACKGROUND: To identify and rank the importance of key determinants of end-of-life (EOL) health care costs, and to understand how the key factors impact different percentiles of the distribution of health care costs. METHOD: We applied a principled, machine learning-based variable selection algorithm, using Quantile Regression Forests, to identify key determinants for predicting the 10th (low), 50th (median), and 90th (high) quantiles of EOL health care costs, including costs paid for by Medicare, Medicaid, Medicare Health Maintenance Organizations (HMOs), private HMOs, and patient's out-of-pocket expenditures. RESULTS: Our sample included 7 539 Medicare beneficiaries who died between 2002 and 2017. The 10th, 50th, and 90th quantiles of EOL health care cost are $5 244, $35 466, and $87 241, respectively. Regional characteristics, specifically, the EOL-Expenditure Index, a measure for regional variation in Medicare spending driven by physician practice, and the number of total specialists in the hospital referral region were the top 2 influential determinants for predicting the 50th and 90th quantiles of EOL costs but were not determinants of the 10th quantile. Black race and Hispanic ethnicity were associated with lower EOL health care costs among decedents with lower total EOL health care costs but were associated with higher costs among decedents with the highest total EOL health care costs. CONCLUSIONS: Factors associated with EOL health care costs varied across different percentiles of the cost distribution. Regional characteristics and decedent race/ethnicity exemplified factors that did not impact EOL costs uniformly across its distribution, suggesting the need to use a "higher-resolution" analysis for examining the association between risk factors and health care costs.

De Novo Ostomy Placement Is Associated with Increased Outpatient Opioid Use In Patients with Inflammatory Bowel Disease.

BACKGROUND/AIMS: Opioid use is associated with poor outcomes in patients with inflammatory bowel disease (IBD). We aimed to identify novel factors associated with increased outpatient opioid (OPRx) use following IBD-related hospitalization. METHODS: This was a retrospective cohort study of IBD patients ≥ 18 years old, hospitalized during 2018. The primary outcome was receiving ≥ 1(OPRx) in the year following index hospitalization (IH), excluding prescriptions written within 2 weeks of discharge. Secondary outcomes included having 1-2 vs ≥ 3 OPRx and rates of healthcare utilization. Univariate and multivariate analyses tested associations with OPRx. RESULTS: Of 526 patients analyzed, 209 (40%) received at least 1 OPRx; with a median of 2 [1-3] OPRx. Presence or placement of ostomy at IH, exposure to opioids during IH, ulcerative colitis (UC), mental health comorbidities, admission for surgery and managed on the surgical service, and IBD surgery within 1 year prior to IH were associated with ≥ 1 OPRx on univariate analysis. On multivariable analysis, UC, ostomy placement during IH, anxiety, and inpatient opioid exposure were independently associated with ≥ 1 OPRx. A majority (> 70%) of both inpatient and outpatient opioid prescriptions were written by surgeons. Patients requiring ≥ 3 OPRx had the highest rates of unplanned IBD surgery (56% p = 0.04), all-cause repeat hospitalization (81%, p = 0.003), and IBD-related repeat hospitalization (77%, p = 0.007) in the year following IH. CONCLUSIONS: A multimodal approach to pain management for IBD patients, as well as increased recognition that any patient with a de novo ostomy is at particular risk of opioid use, is needed.

A FLEXIBLE SENSITIVITY ANALYSIS APPROACH FOR UNMEASURED CONFOUNDING WITH MULTIPLE TREATMENTS AND A BINARY OUTCOME WITH APPLICATION TO SEER-MEDICARE LUNG CANCER DATA.

In the absence of a randomized experiment, a key assumption for drawing causal inference about treatment effects is the ignorable treatment assignment. Violations of the ignorability assumption may lead to biased treatment effect estimates. Sensitivity analysis helps gauge how causal conclusions will be altered in response to the potential magnitude of departure from the ignorability assumption. However, sensitivity analysis approaches for unmeasured confounding in the context of multiple treatments and binary outcomes are scarce. We propose a flexible Monte Carlo sensitivity analysis approach for causal inference in such settings. We first derive the general form of the bias introduced by unmeasured confounding, with emphasis on theoretical properties uniquely relevant to multiple treatments. We then propose methods to encode the impact of unmeasured confounding on potential outcomes and adjust the estimates of causal effects in which the presumed unmeasured confounding is removed. Our proposed methods embed nested multiple imputation within the Bayesian framework, which allow for seamless integration of the uncertainty about the values of the sensitivity parameters and the sampling variability, as well as use of the Bayesian Additive Regression Trees for modeling flexibility. Expansive simulations validate our methods and gain insight into sensitivity analysis with multiple treatments. We use the SEER-Medicare data to demonstrate sensitivity analysis using three treatments for early stage non-small cell lung cancer. The methods developed in this work are readily available in the R package SAMTx.

Variable selection with missing data in both covariates and outcomes: Imputation and machine learning.

Variable selection in the presence of both missing covariates and outcomes is an important statistical research topic. Parametric regression are susceptible to misspecification, and as a result are sub-optimal for variable selection. Flexible machine learning methods mitigate the reliance on the parametric assumptions, but do not provide as naturally defined variable importance measure as the covariate effect native to parametric models. We investigate a general variable selection approach when both the covariates and outcomes can be missing at random and have general missing data patterns. This approach exploits the flexibility of machine learning models and bootstrap imputation, which is amenable to nonparametric methods in which the covariate effects are not directly available. We conduct expansive simulations investigating the practical operating characteristics of the proposed variable selection approach, when combined with four tree-based machine learning methods, extreme gradient boosting, random forests, Bayesian additive regression trees, and conditional random forests, and two commonly used parametric methods, lasso and backward stepwise selection. Numeric results suggest that, extreme gradient boosting and Bayesian additive regression trees have the overall best variable selection performance with respect to the F1 score and Type I error, while the lasso and backward stepwise selection have subpar performance across various settings. There is no significant difference in the variable selection performance due to imputation methods. We further demonstrate the methods via a case study of risk factors for 3-year incidence of metabolic syndrome with data from the Study of Women's Health Across the Nation.

Estimating heterogeneous survival treatment effect in observational data using machine learning.

Methods for estimating heterogeneous treatment effect in observational data have largely focused on continuous or binary outcomes, and have been relatively less vetted with survival outcomes. Using flexible machine learning methods in the counterfactual framework is a promising approach to address challenges due to complex individual characteristics, to which treatments need to be tailored. To evaluate the operating characteristics of recent survival machine learning methods for the estimation of treatment effect heterogeneity and inform better practice, we carry out a comprehensive simulation study presenting a wide range of settings describing confounded heterogeneous survival treatment effects and varying degrees of covariate overlap. Our results suggest that the nonparametric Bayesian Additive Regression Trees within the framework of accelerated failure time model (AFT-BART-NP) consistently yields the best performance, in terms of bias, precision, and expected regret. Moreover, the credible interval estimators from AFT-BART-NP provide close to nominal frequentist coverage for the individual survival treatment effect when the covariate overlap is at least moderate. Including a nonparametrically estimated propensity score as an additional fixed covariate in the AFT-BART-NP model formulation can further improve its efficiency and frequentist coverage. Finally, we demonstrate the application of flexible causal machine learning estimators through a comprehensive case study examining the heterogeneous survival effects of two radiotherapy approaches for localized high-risk prostate cancer.

Estimating heterogeneous survival treatment effects of lung cancer screening approaches: A causal machine learning analysis.

The National Lung Screening Trial (NLST) found that low-dose computed tomography (LDCT) screening provided lung cancer (LC) mortality benefit compared to chest radiography (CXR). Considerable research concerns identifying the differential treatment effects that may exist in certain subpopulations. We shed light on several important issues in existing research and highlight the need for further investigation of the heterogeneous comparative effect of LDCT versus CXR, using more flexible and rigorous statistical approaches. We used a high-performance Bayesian machine learning approach designed for censored survival data, accelerated failure time Bayesian additive regression trees model (AFT-BART), to flexibly capture the relationships between the failure time and predictors. We then used the counterfactual framework to draw Markov chain Monte Carlo samples of the individual treatment effect for each participant. Using these posterior samples, we explored the possible treatment effect heterogeneity via a stepwise binary tree approach. When re-analyzed with AFT-BART, LDCT did not have a statistically significant LC or overall mortality benefit compared to CXR. The Asian and Black (particularly those with pack-year ≥ 37 years and without emphysema) NLST population were shown to have enhanced overall mortality benefit from LDCT than the population average. Although inconclusive for LC mortality benefit, Asians, Blacks and Whites with history of chronic obstructive pulmonary disease showed a small trend towards benefit from LDCT. Causal inference with flexible machine learning modeling can provide valuable knowledge for informing treatment decision and planning targeted clinical trials emphasizing personalized medicine approaches.