arXiv subjects

Rachel Wong

Publications and source records attributed to Rachel Wong.

5 recordsLinked to original sources

arXiv2026

A Comparative Study of Feature Selection Methods for EHR Diagnosis Codes in Opioid Use Disorder Prediction

Feature selection is a critical step in electronic health record (EHR)-based predictive modeling, where input variables are often high-dimensional, sparse, noisy, and redundant. Large feature sets not only increase computational burden and overfitting risk, but also make model interpretation difficult, leading to limited usefulness in clinical settings. In this study, we focus on diagnosis-related features and compare five feature selection paradigms for opioid use disorder (OUD) prediction: recurrence enrichment, NTK-motivated early gradient sensitivity, LightGBM-SHAP, Elastic Net, and large language model (LLM)-guided semantic selection. We use a unified preprocessing and evaluation framework and assess each method by downstream predictive performance, resampling stability, and representation of infrequent diagnosis codes. Our results demonstrate that performance improves with larger feature budgets with diminishing returns beyond a moderate size. NTK sensitivity provides the best overall balance of accuracy and stability, and LLM-guided selection contributes complementary clinically meaningful signals despite lower standalone performance.

cs.LG↗

arXiv2025

Tidal Synchronization of TESS Eclipsing Binaries

Tidal synchronization plays a fundamental role in the evolution of binary star systems. However, key details such as the timescale of synchronization, efficiency of tidal dissipation, rotational period, and dependence on stellar mass are not well constrained. We present a catalog of rotation periods, orbital periods, and eccentricities from eclipsing binaries (EBs) that can be used to study the role of tides in the rotational evolution of low-mass dwarf (FGKM spectral type) binaries. This study presents the largest catalog of EB orbital and rotational periods (Porb and Prot) measured from the Transiting Exoplanet Satellite Survey (TESS). We first classify 4584 light curves from the TESS Eclipsing Binary Catalog according to out-of-eclipse stellar variability type: starspot modulation, ellipsoidal variability, non-periodic variability, and "other" variability (e.g. pulsations). We then manually validate each light curve classification, resulting in a sample of 1039 candidates with 584 high-confidence EBs that exhibit detectable star-spot modulation. From there, we measure and compare the rotation period of each starspot-modulated EB using three methods: Lomb-Scargle periodograms, autocorrelation function, and phase dispersion minimization. We find that our period distributions are consistent with previous work that used a sample of 816 starspot EBs from Kepler to identify two populations: a synchronous population (with Porb~Prot) and a subsynchronous population (with 8Porb~7Prot). Using Bayesian model comparison, we find that a bimodal distribution is a significantly better fit than a unimodal distribution for Kepler and TESS samples, both individually or combined, confirming that the subsynchronous population is statistically significant.

astro-ph.SR↗

arXiv2023

Enhancing Clinical Predictive Modeling through Model Complexity-Driven Class Proportion Tuning for Class Imbalanced Data: An Empirical Study on Opioid Overdose Prediction

Class imbalance problems widely exist in the medical field and heavily deteriorates performance of clinical predictive models. Most techniques to alleviate the problem rebalance class proportions and they predominantly assume the rebalanced proportions should be a function of the original data and oblivious to the model one uses. This work challenges this prevailing assumption and proposes that links the optimal class proportions to the model complexity, thereby tuning the class proportions per model. Our experiments on the opioid overdose prediction problem highlight the performance gain of tuning class proportions. Rigorous regression analysis also confirms the advantages of the theoretical framework proposed and the statistically significant correlation between the hyperparameters controlling the model complexity and the optimal class proportions.

cs.LG↗

arXiv2022

A Multimodal Transformer: Fusing Clinical Notes with Structured EHR Data for Interpretable In-Hospital Mortality Prediction

Deep-learning-based clinical decision support using structured electronic health records (EHR) has been an active research area for predicting risks of mortality and diseases. Meanwhile, large amounts of narrative clinical notes provide complementary information, but are often not integrated into predictive models. In this paper, we provide a novel multimodal transformer to fuse clinical notes and structured EHR data for better prediction of in-hospital mortality. To improve interpretability, we propose an integrated gradients (IG) method to select important words in clinical notes and discover the critical structured EHR features with Shapley values. These important words and clinical features are visualized to assist with interpretation of the prediction outcomes. We also investigate the significance of domain adaptive pretraining and task adaptive fine-tuning on the Clinical BERT, which is used to learn the representations of clinical notes. Experiments demonstrated that our model outperforms other methods (AUCPR: 0.538, AUCROC: 0.877, F1:0.490).

cs.CL↗

arXiv2022

A method for comparing multiple imputation techniques: a case study on the U.S. National COVID Cohort Collaborative

Healthcare datasets obtained from Electronic Health Records have proven to be extremely useful to assess associations between patients' predictors and outcomes of interest. However, these datasets often suffer from missing values in a high proportion of cases and the simple removal of these cases may introduce severe bias. For these reasons, several multiple imputation algorithms have been proposed to attempt to recover the missing information. Each algorithm presents strengths and weaknesses, and there is currently no consensus on which multiple imputation algorithms works best in a given scenario. Furthermore, the selection of each algorithm parameters and data-related modelling choices are also both crucial and challenging. In this paper, we propose a novel framework to numerically evaluate strategies for handling missing data in the context of statistical analysis, with a particular focus on multiple imputation techniques. We demonstrate the feasibility of our approach on a large cohort of type-2 diabetes patients provided by the National COVID Cohort Collaborative (N3C) Enclave, where we explored the influence of various patient characteristics on outcomes related to COVID-19. Our analysis included classic multiple imputation techniques as well as simple complete-case Inverse Probability Weighted models. The experiments presented here show that our approach could effectively highlight the most valid and performant missing-data handling strategy for our case study. Moreover, our methodology allowed us to gain an understanding of the behavior of the different models and of how it changed as we modified their parameters. Our method is general and can be applied to different research fields and on datasets containing heterogeneous types.

cs.AI↗