Clinical Needs Assessment of a Machine Learning-Based Asthma Management Tool: User-Centered Design Approach

Area of Science:

• Health Informatics
• Clinical Decision Support Systems
• User Experience Research

Background:

• Personalized asthma management is hindered by fragmented electronic health record data.
• Machine learning (ML) tools offer potential solutions but face implementation barriers.

Purpose of the Study:

• Explore clinicians' experiences with current asthma management systems.
• Identify user requirements for ML-based clinical decision support tools.
• Determine barriers to implementing ML in asthma care.

Main Methods:

• Employed a user-centered design approach (double-diamond framework).
• Conducted semi-structured interviews with 14 clinicians managing pediatric asthma.
• Utilized shadowing and synthesis analysis to inform research questions.

Main Results:

• Identified key themes: usability, workflow, algorithm explainability, and implementation barriers.
• Clinicians require cohesive information views and proactive risk identification.
• Algorithm transparency, validity documentation, and proper training are crucial for adoption.

Conclusions:

• A user-centered, informatics-based approach is vital for developing effective ML tools in clinical practice.
• Engaging clinicians early in the design process ensures technology meets practical needs.