arXiv subjects

Anthony Chang

Publications and source records attributed to Anthony Chang.

4 recordsLinked to original sources

arXiv2026

Robust by Design: A Continuous Monitoring and Data Integration Framework for Medical AI

Adaptive medical AI models often face performance drops in dynamic clinical environments due to data drift. We propose an autonomous continuous monitoring and data integration framework that maintains robust performance over time. Focusing on glomerular pathology image classification (proliferative vs. non-proliferative lupus nephritis), our three-stage method uses multi-metric feature analysis and Monte Carlo dropout-based uncertainty gating to decide when to retrain on new data. Only images statistically similar to the training distribution (via Euclidean, cosine, Mahalanobis metrics) and with low predictive entropy are integrated. The model is then incrementally retrained with these images under strict performance safeguards (no metric degradation >5%). In experiments with a ResNet18 ensemble on a multi-center dataset, the framework prevents performance degradation: new images were added without significant change in AUC (~0.92) or accuracy (~89%). This approach addresses data shift and avoids catastrophic forgetting, enabling sustained learning in medical imaging AI.

cs.CV↗

arXiv2026

Shortcut Learning in Glomerular AI: Adversarial Penalties Hurt, Entropy Helps

Stain variability is a pervasive source of distribution shift and potential shortcut learning in renal pathology AI. We ask whether lupus nephritis glomerular lesion classifiers exploit stain as a shortcut, and how to mitigate such bias without stain or site labels. We curate a multi-center, multi-stain dataset of 9,674 glomerular patches (224$\times$224) from 365 WSIs across three centers and four stains (PAS, H&E, Jones, Trichrome), labeled as proliferative vs. non-proliferative. We evaluate Bayesian CNN and ViT backbones with Monte Carlo dropout in three settings: (1) stain-only classification; (2) a dual-head model jointly predicting lesion and stain with supervised stain loss; and (3) a dual-head model with label-free stain regularization via entropy maximization on the stain head. In (1), stain identity is trivially learnable, confirming a strong candidate shortcut. In (2), varying the strength and sign of stain supervision strongly modulates stain performance but leaves lesion metrics essentially unchanged, indicating no measurable stain-driven shortcut learning on this multi-stain, multi-center dataset, while overly adversarial stain penalties inflate predictive uncertainty. In (3), entropy-based regularization holds stain predictions near chance without degrading lesion accuracy or calibration. Overall, a carefully curated multi-stain dataset can be inherently robust to stain shortcuts, and a Bayesian dual-head architecture with label-free entropy regularization offers a simple, deployment-friendly safeguard against potential stain-related drift in glomerular AI.

cs.CV↗

arXiv2026

Mathematical simulations of pediatric hemodynamics in isolated ventricular septal defect

Computer modeling of the cardiovascular system has potential to revolutionize personalized medical care. This is especially promising for congenital heart defects, such as ventricular septal defect (VSD), a hole between the two ventricles of the heart. However, relatively few studies have built computer models for VSD, nor have they considered how natural adaptation to the cardiovascular system with age might interact with the presence of a small, medium, or large size VSD. Here, we combine a lumped parameter model of the cardiovascular system with two key modeling components: a size-dependent resistance dictating shunt flow between the two ventricles and age-dependent scaling relationships for the systemic and pulmonary circulations. Our results provide insight into changes in hemodynamic conditions with various VSD sizes. We investigate the combined effects of VSD size, vascular parameters, and age, showing distinct differences with these three factors. This study lays the necessary foundation for studying VSD and towards building digital shadows and digital twins for managing VSD in pediatrics.

q-bio.TO↗

arXiv2024

Multi-Group Maximum Entropy Method: Modeling Translational Non-Equilibrium

The most rigorous physical description of non-equilibrium gas dynamics is rooted in the numerical solution of the Boltzmann equation. Yet, the large number of degrees of freedom and the wide range of both spatial and temporal scales render these equations intractable for many relevant problems. This study constructs a reduced-order model for the Boltzmann equation, by combining coarse-graining modeling framework with the maximum entropy principle. This is accomplished by projecting the high-dimensional Boltzmann equation onto a carefully chosen lower-dimensional subspace, resulting from the discretization of the velocity space into sub-volumes. Within each sub-volume, the distribution function is reconstructed through the maximum entropy principle, ensuring compliance with the detailed balance. The resulting set of conservation equations comprises mass, momentum, and energy for each sub-volume, allowing for flexibility in the description of the velocity distribution function. This new set of governing equations, while retaining many of the mathematical characteristics of the conventional Navier-Stokes equations far outperforms them in terms of applicability. The proposed methodology is applied to the Bhatnagar, Gross, and Krook (BGK) formulation of the Boltzmann equation. To validate the model's accuracy, we simulate the non-equilibrium relaxation of a gas under spatially uniform conditions. Additionally, the model is used to analyze the shock structure of a 1-D standing shockwave across an extensive range of Mach numbers. Notably, both the non-equilibrium velocity distribution functions and macroscopic metrics derived from our model align remarkably with the direct solutions of the Boltzmann equation.

physics.comp-ph↗