Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2607.19020

Freezing the Physiological Encoder: Explanation Stability Under Bounded Updates of an ICU Model

Fatema Ferdous Tamanna·K. M. Merajul Arefin·Md. Abdul Masud
View original source ↗Read source document ↗Export metadata ↓

Abstract

Clinical prediction models deployed in intensive care units may require model updating when data distributions shift, yet unconstrained adaptation can alter model behavior in ways that are difficult to audit. We propose a structurally bounded updating framework that separates physiological dynamics from treatment context and restricts post-drift adaptation to the treatment pathway and fusion head, while leaving the physiological encoder unchanged. Rather than assuming that physiological information remains stable, we investigate how this predefined update boundary affects model explanations after distribution shift. Using 84,792 MIMIC-IV ICU stays across four temporal transitions, we compare selective adaptation with full model adaptation under treatment-side distributional and performance drift. Selective adaptation produces more stable physiological attribution ordering than full adaptation, with rank correlation of 0.875 versus 0.812 and top-5 feature agreement of 0.674 versus 0.552, while retrieval stability also improves (Jaccard similarity 0.614 versus 0.517). Importantly, freezing does not make explanations globally invariant; instead, it constrains where model changes can occur, redirecting explanatory changes toward the treatment pathway and fusion component. Predictive performance remains task-dependent, with selective adaptation outperforming full adaptation for some outcomes while showing a slight disadvantage for intubation prediction. These results suggest that explanation behavior after model updating is influenced not simply by whether a component is frozen, but by the structural boundary defining which components are permitted to absorb adaptation. Such predefined boundaries provide a practical basis for auditable and controlled updating of clinical prediction models under distribution shift.

Explore related subjects

cs.LGcs.AIcs.IRq-bio.QM

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Fatema Ferdous Tamanna, K. M. Merajul Arefin, Md. Abdul Masud. 2026-09-13. Freezing the Physiological Encoder: Explanation Stability Under Bounded Updates of an ICU Model. https://arxiv.org/abs/2607.19020

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Rollout Total Correlation for Deep Reinforcement Learning

Learning task-relevant representations is crucial for reinforcement learning. Recent approaches aim to learn such representations by improving the temporal consistency in the observed transitions. However, they only consider individual transitions and can fail to achieve long-term consistency. Instead, we argue that capturing aspects of the state that correlate with other states and actions of the trajectory---even more distant in the future---could further help in extracting task-relevant information. Hence, in this paper we investigate how to learn representations by maximizing the rollout total correlation, the correlation among all learned representations and actions within the trajectories produced by the agent. For improving rollout total correlation, we propose to combine two complementary lower bounds based on a generative and a discriminative model, combined with a simple and effective technique of chunk-wise mini-batching. Furthermore, we propose an intrinsic reward based on the learned representation for better exploration. Experimental evaluations on a set of challenging image-based simulated control tasks show that our method achieves better sample efficiency, and robustness to both white noise and natural video backgrounds compared to leading baselines.

cs.LG↗
arXiv2026

Reward Shaping to Mitigate Reward Hacking in RLHF

Reinforcement learning from human feedback (RLHF) is widely used to align large language models (LLMs) with human preferences. However, RLHF remains vulnerable to \emph{reward hacking}, whereby a policy exploits imperfections in the reward function instead of learning the intended behavior, thereby undermining alignment. Although reward shaping can stabilize RLHF training and partially mitigate reward hacking, shaping methods and their underlying design principles have not been systematically investigated. To address this gap, we conduct a comprehensive study of prevalent reward-shaping techniques. Our analysis identifies two key design principles: (1) the reinforcement-learning reward should be bounded, and (2) it should grow rapidly at first and then gradually saturate. Motivated by these principles, we propose Preference as Reward (PAR), a novel method that uses the latent preferences encoded in the reward model as the reinforcement-learning signal. We further show that PAR possesses two variance-reduction properties that stabilize RLHF training and substantially widen the practical window for early stopping. Our evaluation consists of two parts. First, we compare PAR with several other reward-shaping strategies using Proximal Policy Optimization (PPO) as the reinforcement-learning algorithm and Gemma2-2B as the base model. Second, we compare PAR with the vanilla baseline (i.e., unshaped reward) across four base models and four reinforcement-learning algorithms. In the first set of experiments, PAR consistently outperforms other reward-shaping methods and also reflects high data efficiency and robustness. The second set of experiments shows that PAR is particularly effective for actor-critic RL algorithms when value estimates become unstable and demonstrates its effectiveness across different base models. The code is available at https://github.com/PorUna-byte/PAR.

cs.LG↗
arXiv2026

Trajectory Entropy Reinforcement Learning for Robust Robot Motor Skill Learning

Simplicity is a critical inductive bias for designing data-driven controllers, especially when robustness is important. Despite the impressive results of deep reinforcement learning in complex control tasks, it is prone to capturing intricate and spurious correlations between observations and actions, leading to failure under slight perturbations to the environment. To tackle this problem, in this work we introduce a novel inductive bias towards simple policies in reinforcement learning. The simplicity inductive bias is introduced by minimizing the entropy of entire action trajectories, corresponding to the number of bits required to describe information in action trajectories after the agent observes state trajectories. Our reinforcement learning agent, Trajectory Entropy Reinforcement Learning, is optimized to minimize the trajectory entropy while maximizing rewards. We show that the trajectory entropy can be effectively estimated by learning a variational parameterized action prediction model, and use the prediction model to construct an information-regularized reward function. Furthermore, we construct a practical algorithm that enables the joint optimization of models, including the policy and the prediction model. Experimental evaluations on several high-dimensional locomotion tasks show that our learned policies produce more cyclical and consistent action trajectories, and achieve superior performance, and robustness to noise and dynamic changes than the state-of-the-art.

cs.LG↗