Providing Care: Intrinsic Human-Machine Teams and Data.

Area of Science:

• Artificial Intelligence in Healthcare
• Machine Learning
• Clinical Decision Support Systems

Background:

• Human-machine teaming is crucial in healthcare AI, yet integrating human expertise with quantitative data remains a challenge.
• Existing methods struggle to effectively combine qualitative insights with numerical health data.
• Lack of explainability and trust hinders the adoption of AI in clinical settings.

Purpose of the Study:

• To propose a novel method for incorporating qualitative expert perspectives into machine learning training data.
• To develop a critical clinical event (CCE) vector that merges quantitative data with human insights.
• To enhance explainability, understandability, and trust in AI-based clinical decision support systems (CDSS).

Main Methods:

• Implementation of an entropy-based consensus construct to manage qualitative data.
• Development of a CCE vector to combine qualitative and quantitative data, addressing small sample sizes, non-normal distributions, and ordinal Likert scales.
• Encoding human considerations directly into machine learning models.

Main Results:

• The CCE vector effectively minimizes challenges associated with qualitative data, enabling its integration with quantitative measures.
• Incorporating human perspectives leads to machine learning models that encode expert considerations.
• The proposed method provides a foundation for increased explainability and trust in AI CDSS.

Conclusions:

• The developed method offers a robust approach to integrating human expertise into AI for healthcare.
• This integration is key to improving human-machine teaming in clinical decision support.
• The CCE vector facilitates the development of more trustworthy and understandable AI systems in medicine.