A knowledge-enhanced domain-aware large language model agent for atrial fibrillation management

Area of Science:

• Artificial Intelligence in Medicine
• Digital Health
• Chronic Disease Management

Background:

• Large language models (LLMs) offer potential for atrial fibrillation (AF) management but face accuracy challenges.
• Current clinical LLM applications often yield suboptimal results for patient self-management.

Purpose of the Study:

• To develop PULSE, a novel knowledge-enhanced, domain-aware LLM agent for improved AF patient self-management.
• To enhance the accuracy, utility, safety, and readability of LLM-generated medical information for patients.

Main Methods:

• Integrated multimodal inputs, curated clinical knowledge bases, prompt engineering, and retrieval-augmented generation.
• Developed an agent-based architecture for the LLM.
• Evaluated performance against four base LLMs on response quality and readability.

Main Results:

• PULSE significantly improved clinical accuracy, content integrity, utility, and safety (P < 0.05).
• Enhanced empathy and clarity in patient-facing outputs.
• Maintained comparable conciseness to existing models.

Conclusions:

• PULSE demonstrates significant potential for advancing chronic disease self-management through improved LLM performance.
• Agent-driven LLM systems can enhance both factual accuracy and readability of medical information.
• Improved patient self-management can lead to better long-term health outcomes.