IdeaDistiller-AI Support for Idea Synthesis in Concept Mapping: Algorithm Development and Validation Study

Area of Science:

• Health Sciences
• Mixed Methods Research

Background:

• Concept mapping (CM) is vital for visualizing complex ideas, particularly in health sciences.
• The idea synthesis phase in CM is a bottleneck: labor-intensive, subjective, and difficult to scale.
• Current CM methods struggle with large datasets due to manual synthesis challenges.

Purpose of the Study:

• Introduce IdeaDistiller, a semiautomated solution for optimizing CM idea synthesis.
• Employ semantic clustering and a human-in-the-loop approach to enhance CM rigor and efficiency.
• Address the scalability and subjectivity issues in the CM idea synthesis phase.

Main Methods:

• Evaluated embedding models, dimensionality reduction, and clustering algorithms on 9 health care datasets (English/Swedish).
• Developed IdeaDistiller to cluster ideas by semantic similarity, identify representative ideas, and provide validation metrics.
• Utilized a human-in-the-loop strategy to maintain methodological rigor during automated synthesis.

Main Results:

• IdeaDistiller significantly reduces manual effort in idea synthesis for CM.
• The approach preserves the quality and transparency of the CM process.
• Human validation remains essential for refining and confirming cluster outputs.

Conclusions:

• Semiautomated methods enhance CM efficiency, scalability, and rigor.
• IdeaDistiller offers a promising approach for handling larger and multilingual CM datasets.
• Future work can integrate this tool into broader CM studies and explore multilingual capabilities.