arXiv subjects

Molei Liu

Publications and source records attributed to Molei Liu.

At least 37 records · Page 2Linked to original sources

arXiv2025

RELEAP: Reinforcement-Enhanced Label-Efficient Active Phenotyping for Electronic Health Records

Objective: Electronic health record (EHR) phenotyping often relies on noisy proxy labels, which undermine the reliability of downstream risk prediction. Active learning can reduce annotation costs, but most rely on fixed heuristics and do not ensure that phenotype refinement improves prediction performance. Our goal was to develop a framework that directly uses downstream prediction performance as feedback to guide phenotype correction and sample selection under constrained labeling budgets. Materials and Methods: We propose Reinforcement-Enhanced Label-Efficient Active Phenotyping (RELEAP), a reinforcement learning-based active learning framework. RELEAP adaptively integrates multiple querying strategies and, unlike prior methods, updates its policy based on feedback from downstream models. We evaluated RELEAP on a de-identified Duke University Health System (DUHS) cohort (2014-2024) for incident lung cancer risk prediction, using logistic regression and penalized Cox survival models. Performance was benchmarked against noisy-label baselines and single-strategy active learning. Results: RELEAP consistently outperformed all baselines. Logistic AUC increased from 0.774 to 0.805 and survival C-index from 0.718 to 0.752. Using downstream performance as feedback, RELEAP produced smoother and more stable gains than heuristic methods under the same labeling budget. Discussion: By linking phenotype refinement to prediction outcomes, RELEAP learns which samples most improve downstream discrimination and calibration, offering a more principled alternative to fixed active learning rules. Conclusion: RELEAP optimizes phenotype correction through downstream feedback, offering a scalable, label-efficient paradigm that reduces manual chart review and enhances the reliability of EHR-based risk prediction.

cs.LG↗

arXiv2025

SIM-Shapley: A Stable and Computationally Efficient Approach to Shapley Value Approximation

Explainable artificial intelligence (XAI) is essential for trustworthy machine learning (ML), particularly in high-stakes domains such as healthcare and finance. Shapley value (SV) methods provide a principled framework for feature attribution in complex models but incur high computational costs, limiting their scalability in high-dimensional settings. We propose Stochastic Iterative Momentum for Shapley Value Approximation (SIM-Shapley), a stable and efficient SV approximation method inspired by stochastic optimization. We analyze variance theoretically, prove linear $Q$-convergence, and demonstrate improved empirical stability and low bias in practice on real-world datasets. In our numerical experiments, SIM-Shapley reduces computation time by up to 85% relative to state-of-the-art baselines while maintaining comparable feature attribution quality. Beyond feature attribution, our stochastic mini-batch iterative framework extends naturally to a broader class of sample average approximation problems, offering a new avenue for improving computational efficiency with stability guarantees. Code is publicly available at https://github.com/nliulab/SIM-Shapley.

stat.ML↗

arXiv2025

CEGA: A Cost-Effective Approach for Graph-Based Model Extraction and Acquisition

Graph Neural Networks (GNNs) have demonstrated remarkable utility across diverse applications, and their growing complexity has made Machine Learning as a Service (MLaaS) a viable platform for scalable deployment. However, this accessibility also exposes GNN to serious security threats, most notably model extraction attacks (MEAs), in which adversaries strategically query a deployed model to construct a high-fidelity replica. In this work, we evaluate the vulnerability of GNNs to MEAs and explore their potential for cost-effective model acquisition in non-adversarial research settings. Importantly, adaptive node querying strategies can also serve a critical role in research, particularly when labeling data is expensive or time-consuming. By selectively sampling informative nodes, researchers can train high-performing GNNs with minimal supervision, which is particularly valuable in domains such as biomedicine, where annotations often require expert input. To address this, we propose a node querying strategy tailored to a highly practical yet underexplored scenario, where bulk queries are prohibited, and only a limited set of initial nodes is available. Our approach iteratively refines the node selection mechanism over multiple learning cycles, leveraging historical feedback to improve extraction efficiency. Extensive experiments on benchmark graph datasets demonstrate our superiority over comparable baselines on accuracy, fidelity, and F1 score under strict query-size constraints. These results highlight both the susceptibility of deployed GNNs to extraction attacks and the promise of ethical, efficient GNN acquisition methods to support low-resource research environments.

cs.LG↗

arXiv2025

Semi-supervised Clustering Through Representation Learning of Large-scale EHR Data

Electronic Health Records (EHR) offer rich real-world data for personalized medicine, providing insights into disease progression, treatment responses, and patient outcomes. However, their sparsity, heterogeneity, and high dimensionality make them difficult to model, while the lack of standardized ground truth further complicates predictive modeling. To address these challenges, we propose SCORE, a semi-supervised representation learning framework that captures multi-domain disease profiles through patient embeddings. SCORE employs a Poisson-Adapted Latent factor Mixture (PALM) Model with pre-trained code embeddings to characterize codified features and extract meaningful patient phenotypes and embeddings. To handle the computational challenges of large-scale data, it introduces a hybrid Expectation-Maximization (EM) and Gaussian Variational Approximation (GVA) algorithm, leveraging limited labeled data to refine estimates on a vast pool of unlabeled samples. We theoretically establish the convergence of this hybrid approach, quantify GVA errors, and derive SCORE's error rate under diverging embedding dimensions. Our analysis shows that incorporating unlabeled data enhances accuracy and reduces sensitivity to label scarcity. Extensive simulations confirm SCORE's superior finite-sample performance over existing methods. Finally, we apply SCORE to predict disability status for patients with multiple sclerosis (MS) using partially labeled EHR data, demonstrating that it produces more informative and predictive patient embeddings for multiple MS-related conditions compared to existing approaches.

stat.ME↗

arXiv2025

Domain Adaptation Optimized for Robustness in Mixture Populations

While domain adaptation methods address data shifts, most assume target populations align with at least one source population, neglecting mixtures that combine sources influenced by factors like demographics. Additional challenges in electronic health record (EHR)-based studies include unobserved outcomes and the need to explain population mixtures using broader clinical characteristics than those in standard risk models. To address these challenges under shifts in both covariate distributions and outcome models, we propose a novel framework: Domain Adaptation Optimized for Robustness in Mixture populations (DORM). Leveraging partially labeled source data, DORM constructs an initial target outcome model under a joint source-mixture assumption. To enhance generalizability to future target populations that may deviate from the joint source-mixture approximation, DORM incorporates a group adversarial learning step to derive a final estimate, optimizing its worst-case performance within a convex uncertainty set built around the initial target model. In addition, this robust domain adaptation procedure is assisted by high-dimensional surrogates that enhance transferability in EHR studies. When a small set of gold-standard or noisy labels is available from the target population, a tuning strategy is implemented to refine the uncertainty set, mitigating conservativeness and further improving performance for the specific target population. Statistical convergence and predictive accuracy of our method are quantified through asymptotic studies. Simulation and real-world studies demonstrate the out-performance of our method over existing approaches.

stat.ME↗

arXiv2025

Knowledge-Guided Wasserstein Distributionally Robust Optimization

Transfer learning is a popular strategy to leverage external knowledge and improve statistical efficiency, particularly with a limited target sample. We propose a novel knowledge-guided Wasserstein Distributionally Robust Optimization (KG-WDRO) framework that adaptively incorporates multiple sources of external knowledge to overcome the conservativeness of vanilla WDRO, which often results in overly pessimistic shrinkage toward zero. Our method constructs smaller Wasserstein ambiguity sets by controlling the transportation along directions informed by the source knowledge. This strategy can alleviate perturbations on the predictive projection of the covariates and protect against information loss. Theoretically, we establish the equivalence between our WDRO formulation and the knowledge-guided shrinkage estimation based on collinear similarity, ensuring tractability and geometrizing the feasible set. This also reveals a novel and general interpretation for recent shrinkage-based transfer learning approaches from the perspective of distributional robustness. In addition, our framework can adjust for scaling differences in the regression models between the source and target and accommodates general types of regularization such as lasso and ridge. Extensive simulations demonstrate the superior performance and adaptivity of KG-WDRO in enhancing small-sample transfer learning.

cs.LG↗

arXiv2024

Multi-source Stable Variable Importance Measure via Adversarial Machine Learning

The quantification and inference of predictive importance for exposure covariates have recently gained significant attention in the context of interpretable machine learning. Contemporary scientific investigations often involve data originating from multiple sources with distributional heterogeneity. It is imperative to introduce a new notation of the variable importance measure that is stable across diverse environments. In this paper, we introduce MIMAL (Multi-source Importance Measure via Adversarial Learning), a novel statistical framework designed to quantify the importance of exposure variables by maximizing the worst-case predictive reward across source mixtures. The proposed framework is adaptable to a broad spectrum of machine learning methodologies for both confounding adjustment and exposure effect characterization. We establish the asymptotic normality of the data-dependent estimator of the multi-source variable importance measure under a general machine learning framework. Our framework requires the similar learning accuracy conditions compared to those required for single-source variable importance analysis. The finite-sample performance of MIMAL is demonstrated through extensive numerical studies encompassing diverse data generation scenarios and machine learning implementations. Furthermore, we illustrate the practical utility of our approach in a real-world case study of air pollution in Beijing, analyzing data collected from multiple locations.

stat.ME↗

arXiv2024

Semi-supervised Triply Robust Inductive Transfer Learning

In this work, we propose a Semi-supervised Triply Robust Inductive transFer LEarning (STRIFLE) approach, which integrates heterogeneous data from a label-rich source population and a label-scarce target population and utilizes a large amount of unlabeled data simultaneously to improve the learning accuracy in the target population. Specifically, we consider a high dimensional covariate shift setting and employ two nuisance models, a density ratio model and an imputation model, to combine transfer learning and surrogate-assisted semi-supervised learning strategies effectively and achieve triple robustness. While the STRIFLE approach assumes the target and source populations to share the same conditional distribution of outcome Y given both the surrogate features S and predictors X, it allows the true underlying model of Y|X to differ between the two populations due to the potential covariate shift in S and X. Different from double robustness, even if both nuisance models are misspecified or the distribution of Y|(S, X) is not the same between the two populations, the triply robust STRIFLE estimator can still partially use the source population when the shifted source population and the target population share enough similarities. Moreover, it is guaranteed to be no worse than the target-only surrogate-assisted semi-supervised estimator with an additional error term from transferability detection. These desirable properties of our estimator are established theoretically and verified in finite samples via extensive simulation studies. We utilize the STRIFLE estimator to train a Type II diabetes polygenic risk prediction model for the African American target population by transferring knowledge from electronic health records linked genomic data observed in a larger European source population.

stat.ME↗

arXiv2024

Bridging Data Gaps in Healthcare: A Scoping Review of Transfer Learning in Biomedical Data Analysis

Clinical and biomedical research in low-resource settings often faces significant challenges due to the need for high-quality data with sufficient sample sizes to construct effective models. These constraints hinder robust model training and prompt researchers to seek methods for leveraging existing knowledge from related studies to support new research efforts. Transfer learning (TL), a machine learning technique, emerges as a powerful solution by utilizing knowledge from pre-trained models to enhance the performance of new models, offering promise across various healthcare domains. Despite its conceptual origins in the 1990s, the application of TL in medical research has remained limited, especially beyond image analysis. In our review of TL applications in structured clinical and biomedical data, we screened 3,515 papers, with 55 meeting the inclusion criteria. Among these, only 2% (one out of 55) utilized external studies, and 7% (four out of 55) addressed scenarios involving multi-site collaborations with privacy constraints. To achieve actionable TL with structured medical data while addressing regional disparities, inequality, and privacy constraints in healthcare research, we advocate for the careful identification of appropriate source data and models, the selection of suitable TL frameworks, and the validation of TL models with proper baselines.

cs.LG↗

arXiv2024

Covariate Shift Corrected Conditional Randomization Test

Conditional independence tests are crucial across various disciplines in determining the independence of an outcome variable $Y$ from a treatment variable $X$, conditioning on a set of confounders $Z$. The Conditional Randomization Test (CRT) offers a powerful framework for such testing by assuming known distributions of $X \mid Z$; it controls the Type-I error exactly, allowing for the use of flexible, black-box test statistics. In practice, testing for conditional independence often involves using data from a source population to draw conclusions about a target population. This can be challenging due to covariate shift -- differences in the distribution of $X$, $Z$, and surrogate variables, which can affect the conditional distribution of $Y \mid X, Z$ -- rendering traditional CRT approaches invalid. To address this issue, we propose a novel Covariate Shift Corrected Pearson Chi-squared Conditional Randomization (csPCR) test. This test adapts to covariate shifts by integrating importance weights and employing the control variates method to reduce variance in the test statistics and thus enhance power. Theoretically, we establish that the csPCR test controls the Type-I error asymptotically. Empirically, through simulation studies, we demonstrate that our method not only maintains control over Type-I errors but also exhibits superior power, confirming its efficacy and practical utility in real-world scenarios where covariate shifts are prevalent. Finally, we apply our methodology to a real-world dataset to assess the impact of a COVID-19 treatment on the 90-day mortality rate among patients.

stat.ME↗

arXiv2024

A Semiparametric Approach for Robust and Efficient Learning with Biobank Data

With the increasing availability of electronic health records (EHR) linked with biobank data for translational research, a critical step in realizing its potential is to accurately classify phenotypes for patients. Existing approaches to achieve this goal are based on error-prone EHR surrogate outcomes, assisted and validated by a small set of labels obtained via medical chart review, which may also be subject to misclassification. Ignoring the noise in these outcomes can induce severe estimation and validation bias to both EHR phenotyping and risking modeling with biomarkers collected in the biobank. To overcome this challenge, we propose a novel unsupervised and semiparametric approach to jointly model multiple noisy EHR outcomes with their linked biobank features. Our approach primarily aims at disease risk modeling with the baseline biomarkers, and is also able to produce a predictive EHR phenotyping model and validate its performance without observations of the true disease outcome. It consists of composite and nonparametric regression steps free of any parametric model specification, followed by a parametric projection step to reduce the uncertainty and improve the estimation efficiency. We show that our method is robust to violations of the parametric assumptions while attaining the desirable root-$n$ convergence rates on risk modeling. Our developed method outperforms existing methods in extensive simulation studies, as well as a real-world application in phenotyping and genetic risk modeling of type II diabetes.

stat.ME↗

arXiv2023

Robust and Efficient Semi-supervised Learning for Ising Model

In biomedical studies, it is often desirable to characterize the interactive mode of multiple disease outcomes beyond their marginal risk. Ising model is one of the most popular choices serving for this purpose. Nevertheless, learning efficiency of Ising models can be impeded by the scarcity of accurate disease labels, which is a prominent problem in contemporary studies driven by electronic health records (EHR). Semi-supervised learning (SSL) leverages the large unlabeled sample with auxiliary EHR features to assist the learning with labeled data only and is a potential solution to this issue. In this paper, we develop a novel SSL method for efficient inference of Ising model. Our method first models the outcomes against the auxiliary features, then uses it to project the score function of the supervised estimator onto the EHR features, and incorporates the unlabeled sample to augment the supervised estimator for variance reduction without introducing bias. For the key step of conditional modeling, we propose strategies that can effectively leverage the auxiliary EHR information while maintaining moderate model complexity. In addition, we introduce approaches including intrinsic efficient updates and ensemble, to overcome the potential misspecification of the conditional model that may cause efficiency loss. Our method is justified by asymptotic theory and shown to outperform existing SSL methods through simulation studies. We also illustrate its utility in a real example about several key phenotypes related to frequent ICU admission on MIMIC-III data set.

stat.ME↗

arXiv2023

Maxway CRT: Improving the Robustness of the Model-X Inference

The model-X conditional randomization test (CRT) is a flexible and powerful testing procedure for the conditional independence hypothesis: X is independent of Y conditioning on Z. Though having many attractive properties, the model-X CRT relies on the model-X assumption that we have perfect knowledge of the distribution of X | Z. If there is an error in modeling the distribution of X | Z, this approach may lose its validity. This problem is even more severe when the adjustment covariates Z are of high dimensionality, in which situation precise modeling of X against Z can be hard. In response to this, we propose the Maxway (Model and Adjust X With the Assistance of Y) CRT, which learns the distribution of Y | Z, and uses it to calibrate the resampling distribution of X to gain robustness to the error in modeling X. We prove that the type-I error inflation of the Maxway CRT can be controlled by the learning error for the low-dimensional adjusting model plus the product of learning errors for X | Z and Y | Z, which could be interpreted as an "almost doubly robust" property. Based on this, we develop implementing algorithms of the Maxway CRT in practical scenarios including (surrogate-assisted) semi-supervised learning and transfer learning where valid information about Y | Z can be potentially provided by some auxiliary or external data. Through extensive simulation studies under different scenarios, we demonstrate that the Maxway CRT achieves significantly better type-I error control than existing model-X inference approaches while preserving similar powers. Finally, we apply our methodology to two real examples, including (1) studying obesity paradox with electronic health record (EHR) data assisted by surrogate variables; (2) inferring the side effect of statins among the ethnic minority group via transferring knowledge from the majority group.

stat.ME↗

arXiv2023

Improve Efficiency of Doubly Robust Estimator when Propensity Score is Misspecified

Doubly robust (DR) estimation is a crucial technique in causal inference and missing data problems. We propose a novel Propensity score Augmentved Doubly robust (PAD) estimator to enhance the commonly used DR estimator for average treatment effect on the treated (ATT), or equivalently, the mean of the outcome under covariate shift. Our proposed estimator attains a lower asymptotic variance than the conventional DR estimator when the propensity score (PS) model is misspecified and the outcome regression (OR) model is correct while maintaining the double robustness property that it is valid when either the PS or OR model is correct. These are realized by introducing some properly calibrated adjustment covariates to linearly augment the PS model and solving a restricted weighted least square (RWLS) problem to minimize the variance of the augmented estimator. Both the asymptotic analysis and simulation studies demonstrate that PAD can significantly reduce the estimation variance compared to the standard DR estimator when the PS model is wrong and the OR is correct, and maintain close performance to DR when the PS model is correct. We further applied our method to study the effects of eligibility for 401(k) plan on the improvement of net total financial assets using data from the Survey of Income and Program Participation of 1991.

stat.ME↗

arXiv2022

Doubly Robust Augmented Model Accuracy Transfer Inference with High Dimensional Features

Due to label scarcity and covariate shift happening frequently in real-world studies, transfer learning has become an essential technique to train models generalizable to some target populations using existing labeled source data. Most existing transfer learning research has been focused on model estimation, while there is a paucity of literature on transfer inference for model accuracy despite its importance. We propose a novel $\mathbf{D}$oubly $\mathbf{R}$obust $\mathbf{A}$ugmented $\mathbf{M}$odel $\mathbf{A}$ccuracy $\mathbf{T}$ransfer $\mathbf{I}$nferen$\mathbf{C}$e (DRAMATIC) method for point and interval estimation of commonly used classification performance measures in an unlabeled target population using labeled source data. Specifically, DRAMATIC derives and evaluates the risk model for a binary response $Y$ against some low dimensional predictors $\mathbf{A}$ on the target population, leveraging $Y$ from source data only and high dimensional adjustment features $\mathbf{X}$ from both the source and target data. The proposed estimators are doubly robust in the sense that they are $n^{1/2}$ consistent when at least one model is correctly specified and certain model sparsity assumptions hold. Simulation results demonstrate that the point estimation have negligible bias and the confidence intervals derived by DRAMATIC attain satisfactory empirical coverage levels. We further illustrate the utility of our method to transfer the genetic risk prediction model and its accuracy evaluation for type II diabetes across two patient cohorts in Mass General Brigham (MGB) collected using different sampling mechanisms and at different time points.

stat.ME↗

arXiv2022

Assessing the Most Vulnerable Subgroup to Type II Diabetes Associated with Statin Usage: Evidence from Electronic Health Record Data

There have been increased concerns that the use of statins, one of the most commonly prescribed drugs for treating coronary artery disease, is potentially associated with the increased risk of new-onset type II diabetes (T2D). Nevertheless, to date, there is no robust evidence supporting as to whether and what kind of populations are indeed vulnerable for developing T2D after taking statins. In this case study, leveraging the biobank and electronic health record data in the Partner Health System, we introduce a new data analysis pipeline and a novel statistical methodology that address existing limitations by (i) designing a rigorous causal framework that systematically examines the causal effects of statin usage on T2D risk in observational data, (ii) uncovering which patient subgroup is most vulnerable for developing T2D after taking statins, and (iii) assessing the replicability and statistical significance of the most vulnerable subgroup via a bootstrap calibration procedure. Our proposed approach delivers asymptotically sharp confidence intervals and debiased estimate for the treatment effect of the most vulnerable subgroup in the presence of high-dimensional covariates. With our proposed approach, we find that females with high T2D genetic risk are at the highest risk of developing T2D due to statin usage.

stat.AP↗

arXiv2022

Augmented Transfer Regression Learning with Semi-non-parametric Nuisance Models

In contemporary statistical learning, covariate shift correction plays an important role in transfer learning when distribution of the testing data is shifted from the training data. Importance weighting, as a natural and principle strategy to adjust for covariate shift, has been commonly used in the field of transfer learning. However, this strategy is not robust to model misspecification or excessive estimation error. In this paper, we propose an augmented transfer regression learning (ATReL) approach that introduces an imputation model for the targeted response, and uses it to augment the importance weighting equation. With novel semi-non-parametric constructions and calibrated moment estimating equations for the two nuisance models, our ATReL method is less prone to (i) the curse of dimensionality compared to nonparametric approaches, and (ii) model mis-specification than parametric approaches. We show that our ATReL estimator is root-n-consistent when at least one nuisance model is correctly specified, estimation for the parametric part of the nuisance models achieves parametric rate, and the nonparametric components are rate doubly robust. Simulation studies demonstrate that our method is more robust and efficient than existing parametric and fully nonparametric (machine learning) estimators under various configurations. We also examine the utility of our method through a real example about transfer learning of phenotyping algorithm for rheumatoid arthritis across different time windows. Finally, we propose ways to enhance the intrinsic efficiency of our estimator and to incorporate modern machine learning methods with our proposed framework.

stat.ME↗

arXiv2021

Efficient Estimation and Evaluation of Prediction Rules in Semi-Supervised Settings under Stratified Sampling

In many contemporary applications, large amounts of unlabeled data are readily available while labeled examples are limited. There has been substantial interest in semi-supervised learning (SSL) which aims to leverage unlabeled data to improve estimation or prediction. However, current SSL literature focuses primarily on settings where labeled data is selected randomly from the population of interest. Non-random sampling, while posing additional analytical challenges, is highly applicable to many real world problems. Moreover, no SSL methods currently exist for estimating the prediction performance of a fitted model under non-random sampling. In this paper, we propose a two-step SSL procedure for evaluating a prediction rule derived from a working binary regression model based on the Brier score and overall misclassification rate under stratified sampling. In step I, we impute the missing labels via weighted regression with nonlinear basis functions to account for nonrandom sampling and to improve efficiency. In step II, we augment the initial imputations to ensure the consistency of the resulting estimators regardless of the specification of the prediction model or the imputation model. The final estimator is then obtained with the augmented imputations. We provide asymptotic theory and numerical studies illustrating that our proposals outperform their supervised counterparts in terms of efficiency gain. Our methods are motivated by electronic health records (EHR) research and validated with a real data analysis of an EHR-based study of diabetic neuropathy.

stat.ML↗