A5062212983- Advanced Causal Inference Techniques
- Statistical Methods and Bayesian Inference
- Statistical Methods in Clinical Trials
- Statistical Methods and Inference
- HIV/AIDS Research and Interventions
- Health Systems, Economic Evaluations, Quality of Life
- Radiomics and Machine Learning in Medical Imaging
- HIV Research and Treatment
- HIV, Drug Use, Sexual Risk
- Meta-analysis and systematic reviews
- Global Cancer Incidence and Screening
- MRI in cancer diagnosis
- Machine Learning in Healthcare
- Data-Driven Disease Surveillance
- Colorectal Cancer Screening and Detection
- COVID-19 epidemiological studies
- Medical Imaging Techniques and Applications
- Vaccine Coverage and Hesitancy
- Health disparities and outcomes
- Statistical Methods in Epidemiology
- Bayesian Modeling and Causal Inference
- Influenza Virus Research Studies
- Microfinance and Financial Inclusion
- HIV/AIDS Impact and Responses
- Intensive Care Unit Cognitive Disorders
Brown University
2017-2025
John Brown University
2018-2024
University Hospital Heidelberg
2024
Johns Hopkins University
2016-2024
Fred Hutch Cancer Center
2024
Resonance Research (United States)
2024
University of California, San Francisco
2024
University of Maryland, Baltimore
2024
Cancer Research Center
2024
University of Washington
2024
When treatment effect modifiers influence the decision to participate in a randomized trial, average population represented by individuals will differ from other populations. In this tutorial, we consider methods for extending causal inferences about time‐fixed treatments trial new target of nonparticipants, using data completed and baseline covariate sample population. We examine based on modeling expectation outcome, probability participation, or both (doubly robust). compare simulation...
Despite breast cancer screening rates that exceed those of White women, 26 Black women are diagnosed with more advanced stages cancer.More worrisome, endure higher falsepositive results, an unfavorable outcome implicated in increased risk cancer.Additionally, elevated anxiety, stress, financial burden, [27][28][29][30][31][32] and paradoxically, reduced 33
Purpose To describe the design, conduct, and results of Breast Multiparametric MRI for prediction neoadjuvant chemotherapy Response (BMMR2) challenge. Materials Methods The BMMR2 computational challenge opened on May 28, 2021, closed December 21, 2021. goal was to identify image-based markers derived from multiparametric breast MRI, including diffusion-weighted imaging (DWI) dynamic contrast-enhanced (DCE) along with clinical data predicting pathologic complete response (pCR) following...
Importance The National Lung Screening Trial (NLST) found that screening for lung cancer with low-dose computed tomography (CT) reduced cancer–specific and all-cause mortality compared chest radiography. It is uncertain whether these results apply to a nationally representative target population. Objective To extend inferences about the effects of strategies from NLST population NLST-eligible US adults. Design, Setting, Participants This comparative effectiveness study included data adults...
We take steps toward causally interpretable meta-analysis by describing methods for transporting causal inferences from a collection of randomized trials to new target population, one trial at time and pooling all trials. discuss identifiability conditions average treatment effects in the population provide identification results. show that assumptions allow be transported same have implications law underlying observed data. propose effect estimators rely on different working models code...
This paper proposes a novel paradigm for building regression trees and ensemble learning in survival analysis. Generalizations of the CART Random Forests algorithms general loss functions, latter case more bootstrap procedures, are both introduced. These results, combination with an extension theory censoring unbiased transformations applicable to underpin development two new classes constructing forests: Censoring Unbiased Regression Trees Ensembles. For certain "doubly robust"...
Objectives: National Comprehensive Cancer Network (NCCN) guidelines for stage III colon cancer define low-risk versus high-risk patients based on T (1 to 3 vs. 4) and N 2) status, with some variations in treatment. This study analyzes the impact of tumor deposits (TDs), poor differentiation (PD), perineural invasion (PNI), lymphovascular (LVI) survival. Materials Methods: A retrospective analysis (2010-2015) Database treated both surgery chemotherapy was conducted. Data extracted sex, race,...
Abstract We present methods for causally interpretable meta-analyses that combine information from multiple randomized trials to draw causal inferences a target population of substantive interest. consider identifiability conditions, derive implications the conditions law observed data, and obtain identification results transporting collection independent new in which experimental data may not be available. propose an estimator potential outcome mean under each treatment studied trials. The...
A crucial component of making individualized treatment decisions is to accurately predict each patient's disease risk. In clinical oncology, risks are often measured through time-to-event data, such as overall survival and progression/recurrence-free survival, subject censoring. Risk prediction models based on recursive partitioning methods becoming increasingly popular largely due their ability handle nonlinear relationships, higher-order interactions, and/or high-dimensional covariates....
Estimating a patient's mortality risk is important in making treatment decisions. Survival trees are useful tool and employ recursive partitioning to separate patients into different groups. Existing 'loss based' procedures that would be used the absence of censoring have previously been extended setting right censored outcomes using inverse probability weighted estimators loss functions. In this paper, we propose new 'doubly robust' extensions these motivated by semiparametric efficiency...
Abstract We considered methods for transporting a prediction model use in new target population, both when outcome and covariate data development are available from source population that has different distribution compared with the (but not data) population. discuss how to tailor account differences between also assess model’s performance (e.g., by estimating mean squared error) provide identifiability results measures of potentially misspecified under sampling design where samples obtained...
Abstract Patient-reported outcomes (PROs) are often collected in cancer clinical trials. Data obtained from trials with PROs essential evaluating participant experiences relating to symptoms, financial toxicity, or health-related quality of life. Although most features trial design, implementation, and analyses apply PROs, several considerations unique. In this paper, we focus on specific issues such as selection the tool, timing frequency assessments, data collection methods. We discuss how...
Models using clinical and MRI-based radiomic features identified from ductal carcinoma in situ lesions improved prediction of disease upstaging at surgery compared with standard information alone.
Deep learning is a class of machine algorithms that are popular for building risk prediction models. When observations censored, the outcomes only partially observed and standard deep cannot be directly applied. We develop new potentially censored. To account censoring, unobservable loss function used in absence censoring replaced by unbiased transformation. The resulting can to estimate both survival probabilities restricted mean survival. show how implemented adapting software uncensored...
Abstract Externally controlled trials have commonly been used when conducting a randomized trial (RCT) is not feasible or ethical. By allowing the study of new treatments, use external controls can lead to accelerated advances in management rare diseases. The controls, however, introduces challenges due potential differences between population are enrolled from and patients on from. Some include, but limited to, how diagnosed treated, case mix underlying populations, ability measure...
Abstract This paper explores the design considerations and hurdles encountered by CHinA National CancEr Screening (CHANCES) Trial Tomosynthesis Mammographic Imaging (TMIST), both aimed at advancing cancer screening research. Before population-based programs are launched, it is important to have confidence that potential benefits of process resulting interventions outweigh harms, an ethical imperative because people actively invited into relatively healthy. Large randomized trials provide...
Abstract Public health interventions guided by clustering of HIV-1 molecular sequences may be impacted choices analytical approaches. We identified commonly-used approaches, applied them to 1886 Rhode Island from 2004–2018, and compared concordance in identifying clusters within between used strict (topological support ≥ 0.95; distance 0.015 substitutions/site) relaxed 0.80–0.95; 0.030–0.045 thresholds reflect different epidemiological scenarios. found that differed method threshold depended...
Summary Causally interpretable meta-analysis combines information from a collection of randomized controlled trials to estimate treatment effects in target population which experimentation may not be possible but covariate can obtained. In such analyses, key practical challenge is the presence systematically missing data when some have collected on one or more baseline covariates, other not, that for all participants latter. this article, we provide identification results potential...
We propose Causal Interaction Trees for identifying subgroups of participants that have enhanced treatment effects using observational data. extend the Classification and Regression Tree algorithm by splitting criteria focus on maximizing between-group effect heterogeneity based subgroup-specific estimators to dictate decision-making in algorithm. derive properties three account nature data -- inverse probability weighting, g-formula doubly robust estimators. study performance proposed...