---
title: "Uncertainty-Guided Adaptive Conservative Offline Reinforcement Learning for Safer Mechanical Ventilation"
authors:
  - "Huidong Liu"
  - "Hang Yu"
  - "Qiyang Zhang"
  - "Jiarui Dou"
  - "Xianlei Long"
  - "Jiantao Shi"
  - "Fuqiang Gu"
doi: "10.24963/ijcai.2026/755"
url: "https://ijcai.org/proceedings/2026/755"
pdf: "https://ijcai.org/proceedings/2026/0755.pdf"
venue: "Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence"
year: 2026
pages: "6788-6796"
track: "AI and Health"
keywords:
  - "Precision medicine: Precision medicine"
  - "Medical diagnosis: Medical diagnosis"
  - "Clinical decision support system: Clinical decision support system"
  - "Treatment recommendation: Treatment recommendation"
---

# Uncertainty-Guided Adaptive Conservative Offline Reinforcement Learning for Safer Mechanical Ventilation

Huidong Liu, Hang Yu, Qiyang Zhang, Jiarui Dou, Xianlei Long, Jiantao Shi, Fuqiang Gu

*Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence*, pages 6788-6796. DOI: [10.24963/ijcai.2026/755](https://doi.org/10.24963/ijcai.2026/755)

## Abstract

Mechanical ventilation (MV) is essential in intensive care units (ICUs), yet conventional protocols lack personalization and risk harmful over- or under-ventilation. Offline reinforcement learning (ORL) enables policy optimization from retrospective clinical data without unsafe online interaction, but existing methods are highly sensitive to distributional shift and out-of-distribution (OOD) actions, limiting their reliability in complex clinical settings. To address these challenges, We propose UBER-CQL (Uncertainty-Balanced Exploration and Robust Conservative Q-Learning), a robust ORL algorithm for safe decision-making under dataset shift. UBER-CQL integrates heteroscedastic Bayesian neural networks with conservative Q-learning to model posterior Q-value uncertainty, which is used to adaptively penalize unreliable high-risk actions while maintaining performance within the data support. We further design numerically stable objectives for conservative Bayesian value estimation. Experiments on in-distribution and OOD subsets of MIMIC-III and eICU demonstrate that UBER-CQL outperforms state-of-the-art ORL and clinician baselines, producing safer and more effective MV strategies.