Human-in-the-Loop Multi-Agent Ventilator Decision Support with Contextual Bandit Preference Learning

2026-05-22

AI Summary

Existing ventilator decision support systems struggle to simultaneously achieve personalization, controllability, and auditability. To address this challenge, this work proposes the first ventilator decision support system (VDSS) that integrates human-in-the-loop feedback, a modular multi-agent architecture, and a traceable evidence chain. The system employs contract-driven interfaces to coordinate multi-agent collaboration, leverages contextual bandits to learn clinicians’ parameter-tuning preferences online, and incorporates structured rejection feedback to trigger targeted replanning. Evaluations through ICU trajectory replay and expert review demonstrate that VDSS significantly increases recommendation acceptance rates and reduces the number of interaction rounds required to reach feasible ventilation strategies, thereby validating its clinical deployability and potential impact.

Abstract

Ventilator decision support requires sequential decisions that track evolving physiology and disease trajectories while respecting safety boundaries and clinician specific tuning styles. Rule based approaches rarely generalize personalization, and end to end reinforcement learning or single large language model systems remain difficult to control and audit. We propose the Ventilator Decision Support System (VDSS), a human in the loop multi agent framework that coordinates modular decision components through contract driven structured interfaces and produces traceable evidence for review. VDSS performs online preference adaptation with a contextual bandit, updating clinician specific preferences from the final accepted decision at each adjustment cycle and using them to guide subsequent recommendations. Structured rejection feedback triggers targeted replanning to reduce unproductive iterations and improve interaction stability. Retrospective ICU trajectory replay with expert review indicates higher recommendation acceptability and fewer interaction rounds to reach an acceptable plan, supporting clinically deployable human AI collaboration.

Problem

• Ventilator Decision Support
• Human-in-the-Loop
• Contextual Bandit
• Multi-Agent System
• Clinical Personalization

Innovation

• human-in-the-loop
• multi-agent system
• contextual bandit
• preference learning
• ventilator decision support