VentAgent: When LLMs Learn to Breathe -- Multi-Objective Arbitration for ARDS Ventilation

2026-06-03

AI Summary

This work addresses the challenge of mechanical ventilation in acute respiratory distress syndrome (ARDS), which requires dynamic trade-offs among conflicting objectives such as oxygenation, lung-protective strategies, and acid–base balance. While existing data-driven approaches suffer from imitation bias and reinforcement learning methods lack interpretability, this study proposes VentAgent—a novel hierarchical framework that leverages a large language model (LLM) as a transparent arbitrator. Through perception, planning, and coordination stages, VentAgent explicitly integrates multiple expert policies by harnessing the LLM’s semantic reasoning capabilities to reconcile competing clinical priorities. Evaluated on a high-fidelity physiological simulator, the method significantly outperforms current reinforcement learning and classical control baselines, delivering interpretable, generalizable, and adaptive ventilation control while generating human-readable chains of reasoning to enhance clinical trust and safety.

Abstract

Mechanical ventilation for Acute Respiratory Distress Syndrome (ARDS) requires balancing competing physiological goals, including oxygenation, lung protection, and acid-base homeostasis. However, current data-driven methods, especially those imitating retrospective Electronic Health Records (EHR), often suffer from imitation bias. They may capture superficial correlations from inconsistent clinical demonstrations, such as associating passive ventilator settings with survival because such settings are common in stable patients, and thus fail to generalize to volatile or out-of-distribution phenotypes. Standard Reinforcement Learning (RL) methods also struggle with the adversarial trade-offs of critical care and often produce opaque policies with limited clinical interpretability. To address these limitations, we introduce VentAgent, a hierarchical framework in which Large Language Models (LLMs) act as transparent arbitrators for mechanical ventilation. We reformulate ventilation control as a dynamic Multi-Objective Arbitration process rather than single-objective optimization. VentAgent decomposes decision-making into three interpretable stages: Perception, Planning, and Orchestration. By leveraging the semantic reasoning capabilities of LLMs, it synthesizes strategies from heterogeneous experts and resolves conflicting clinical priorities through an explicit coordination mechanism. Evaluations on a high-fidelity physiological simulator show that VentAgent outperforms state-of-the-art RL and classical control baselines. Moreover, it converts control decisions into human-readable reasoning chains, offering a safer, more interpretable, and adaptable paradigm for critical care automation.

Problem

• ARDS
• mechanical ventilation
• multi-objective arbitration
• imitation bias
• clinical interpretability

Innovation

• Multi-Objective Arbitration
• Large Language Models
• Mechanical Ventilation
• Interpretable AI
• Critical Care Automation