SURGENT: A Surgical Multi-Agent Assistance System Across the Perioperative Workflow

2026-05-28 · Source: Computation and Language · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Clinical Care & Medical Practice, Medical Devices & Health Technology · Depth: Expert, quick

Summary

SURGENT is a surgical multi-agent assistance system designed to overcome the limitations of web-based Large Language Models in complex surgical care. It integrates a Tree-of-Thought planner, multi-department collaboration agents, and retrieval-augmented reasoning, leveraging clinical guidelines and biomedical literature. A novel memory design in SURGENT manages both long-term patient histories and short-term working summaries, facilitating contextualized and consistent reasoning. Evaluated across five perioperative tasks—case analysis, surgical plan simulation, safety monitoring, complication risk assessment, and rehabilitation guidance—SURGENT demonstrated superior performance compared to baseline LLMs and existing medical multi-agent frameworks, providing recommendations better aligned with patient histories. Ablation studies identified DeepSeek as an effective locally deployable backbone model, enabling privacy-preserving deployment without relying on centralized services, positioning SURGENT as a practical and trustworthy advancement for surgical assistance.

Key takeaway

For AI Scientists and medical technology developers building surgical assistance systems, SURGENT demonstrates a robust architecture for enhancing clinical decision support. You should consider integrating multi-agent frameworks with specialized memory management and retrieval-augmented reasoning to overcome standard LLM limitations. Prioritize locally deployable models like DeepSeek to ensure patient data privacy and enable trustworthy, equitable, and secure deployments in healthcare settings.

Key insights

SURGENT is a multi-agent system enhancing surgical decision support through advanced memory and collaborative reasoning.

Principles

• Combine Tree-of-Thought with multi-agent systems.
• Integrate long-term and short-term memory for context.
• Prioritize local deployment for privacy.

Method

SURGENT employs a Tree-of-Thought planner, multi-department collaboration agents, and retrieval-augmented reasoning, supported by a dual-layer memory system for patient data.

In practice

• Use DeepSeek for privacy-preserving LLM deployment.
• Apply multi-agent systems to perioperative tasks.
• Enhance LLM reasoning with structured memory.

Topics

• Surgical Assistance Systems
• Multi-Agent Systems
• Large Language Models
• Perioperative Workflow
• Retrieval-Augmented Generation
• DeepSeek

Best for: CTO, AI Architect, AI Product Manager, AI Scientist, Research Scientist, Domain Expert

Related on AIssential

• SEMA-RAG: A Self-Evolving Multi-Agent Retrieval-Augmented Generation Framework for Medical Reasoning — Multi-agent RAG with self-evolving, sufficiency-driven retrieval significantly improves medical question answering accuracy.
• Promptimus: Improving already good LLM prompts with zero manual engineering — Promptimus automates prompt optimization for LLMs by focusing on specific failure points and generating targeted, iterative refinements.
• not much happened today — Agentic AI success increasingly relies on robust harnesses and infrastructure, not just base model capabilities.
• Perplexity Max Debuts Multi-AI Agent Tool — Perplexity's new agent tool unifies 19 AI models for complex workflows, targeting professionals with a multimodal strategy.
• Reinforced Agent: Inference-Time Feedback for Tool-Calling Agents — Proactive, in-execution review by a specialized agent improves tool-calling agent reliability and performance.
• How Anthropic uses Surge AI to Train and Evaluate Claude — High-quality human feedback is critical for developing safe and capable large language models.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Computation and Language.