Exploring the Concept of an AI-Enabled Multi-Agentic Workflow to Support Telehealth Disease Management

Area of Science:

• Digital Health
• Health Informatics
• Telemedicine

Background:

• Telehealth disease management programs are effective but resource-intensive and difficult to scale.
• Existing telehealth models face sustainability challenges due to high resource demands.

Purpose of the Study:

• To propose a digital care assistant, adapted from the HerzMobil program, to enhance telehealth disease management.
• To improve scalability and provide staff support by automating routine tasks for patients and healthcare professionals.

Main Methods:

• Identification of key areas within telehealth workflows suitable for automation.
• Description of a message handling workflow and the datasets utilized.
• Development of a multi-agentic framework comprising five specialized agents.

Main Results:

• A novel multi-agentic framework with five distinct, specialized agents was developed.
• Evaluation approaches for both the overall system performance and individual agent efficacy were defined.

Conclusions:

• The modular architecture facilitates adaptation to new use cases and ensures privacy via separated responsibilities.
• Selecting appropriate evaluation metrics is crucial for enhancing system reliability, transparency, and user acceptance.