3D++ Visualisation of MEBN Graphs and Screen Representations of Patient Models (PIXIE II)

Area of Science:

• Medical Informatics
• Artificial Intelligence in Medicine
• Human-Computer Interaction

Background:

• Probabilistic graphical models, such as Multi-Entity Bayesian Networks (MEBN), are crucial for knowledge representation in medical decision support systems.
• Effective management and comprehension of these complex models are essential for their successful application in clinical settings, particularly in therapy planning.
• Current visualization and information management methods may not adequately address the complexity of MEBNs for end-users like physicians.

Purpose of the Study:

• To investigate methods and tools for transparent, reproducible, and comprehensible information management of patient probabilistic graphical models.
• To address the need for creative user interface design (UID) and graphical user interfaces (GUIs) to simplify the understanding and management of complex models.
• To propose and demonstrate a proof of concept for visualizing MEBNs.

Main Methods:

• Exploration of modeling and visualization techniques for probabilistic graphical models.
• Focus on user interface design (UID) principles for enhancing model comprehension.
• Utilizing the ANTz visualization tool for demonstrating the visualization of MEBNs.

Main Results:

• Identification of key challenges in managing and visualizing complex probabilistic graphical models.
• Proposal of the ANTz tool as a viable solution for visualizing MEBNs.
• Demonstration of a proof of concept for the ANTz visualization tool.

Conclusions:

• Effective visualization and user interface design are critical for the adoption of probabilistic graphical models in clinical decision support.
• The ANTz tool shows promise for improving the transparency, reproducibility, and comprehensibility of MEBNs.
• Further development and validation of visualization tools like ANTz are needed to support physicians in therapy planning.