{"slug": "unmasking-toxic-mimicry-in-medical-offline-reinforcement-learning-for-icu-sepsis", "title": "Unmasking Toxic Mimicry in Medical Offline Reinforcement Learning for ICU Sepsis Management via Counterfactual Clinical Audits", "summary": "A new study from arXiv (2608.11410v1) introduces the Counterfactual Clinical Audit (CCA) framework, which stress-tests offline reinforcement learning (RL) agents for ICU sepsis management using physiological perturbations based on Surviving Sepsis Campaign guidelines. Auditing the Medical Decision Transformer (MedDT) and Historical Causal Transformer (HCT-RL) on the MIMIC-III database, the authors found that MedDT paradoxically reduces vasopressor dosage as lactate escalates, contradicting resuscitation guidelines, while HCT-RL maintains physiologically consistent responses. The findings expose a systemic misalignment between statistical fit and clinical safety, supporting counterfactual audits as a necessary evaluation standard for medical RL.", "body_md": "arXiv:2608.11410v1 Announce Type: new\nAbstract: Offline reinforcement learning (RL) offers considerable promise for optimizing ICU treatment decisions, yet standard evaluation metrics Mean Squared Error (MSE) and Fitted Q-Evaluation (FQE) assess only behavioral imitation and cannot detect Toxic Mimicry, a failure mode in which agents replicate harmful patterns such as treatment withdrawal during comfort-care transitions. Using the MIMIC-III database, we propose the Counterfactual Clinical Audit (CCA) framework, which stress-tests RL agents through physiological perturbations anchored in Surviving Sepsis Campaign (SSC) guidelines. We audit a Medical Decision Transformer (MedDT) and a Historical Causal Transformer (HCT-RL), the latter employing Causal Action Shielding, propensity-based importance weighting, and Conservative Q-Learning. CCA reveals that MedDT paradoxically reduces vasopressor dosage as lactate escalates, contradicting resuscitation guidelines, while HCT-RL maintains physiologically consistent responses. These findings expose a systemic misalignment between statistical fit and clinical safety, supporting counterfactual audits as a necessary evaluation standard for medical RL.", "url": "https://wpnews.pro/news/unmasking-toxic-mimicry-in-medical-offline-reinforcement-learning-for-icu-sepsis", "canonical_source": "https://arxiv.org/abs/2608.11410", "published_at": "2026-08-13 04:00:00+00:00", "updated_at": "2026-08-13 04:15:07.760259+00:00", "lang": "en", "topics": ["machine-learning", "artificial-intelligence", "ai-safety"], "entities": ["arXiv", "Counterfactual Clinical Audit", "Medical Decision Transformer", "Historical Causal Transformer", "MIMIC-III", "Surviving Sepsis Campaign"], "alternates": {"html": "https://wpnews.pro/news/unmasking-toxic-mimicry-in-medical-offline-reinforcement-learning-for-icu-sepsis", "markdown": "https://wpnews.pro/news/unmasking-toxic-mimicry-in-medical-offline-reinforcement-learning-for-icu-sepsis.md", "text": "https://wpnews.pro/news/unmasking-toxic-mimicry-in-medical-offline-reinforcement-learning-for-icu-sepsis.txt", "jsonld": "https://wpnews.pro/news/unmasking-toxic-mimicry-in-medical-offline-reinforcement-learning-for-icu-sepsis.jsonld"}}