A Human-AI interaction paradigm and its application to rhinocytology

Area of Science:

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

Background:

• Artificial intelligence (AI) tools are increasingly used in medical diagnostics.
• Enhancing user interaction and trust in AI systems is crucial for clinical adoption.
• Current AI systems often lack transparency and user control, limiting their utility.

Purpose of the Study:

• To introduce and evaluate a Human-Centered Artificial Intelligence (HCAI) paradigm for medical cytology.
• To improve the interaction between medical professionals and AI systems, focusing on explainability and user control.
• To redesign an AI-based tool for nasal mucosa analysis using the proposed HCAI interaction paradigm.

Main Methods:

• Developed a Human-AI interaction paradigm based on iterative negotiation.
• Implemented three interaction strategies: Iterative Exploration, Clarification, and Reconfiguration.
• Redesigned an existing AI tool for microscopic analysis of nasal mucosa and tested it with rhinocytologists.

Main Results:

• The redesigned AI tool, incorporating HCAI principles, was evaluated by rhinocytologists.
• Analysis of the evaluation results provided insights into the effectiveness of the interaction strategies.
• The study identified key lessons learned for implementing AI in medical applications.

Conclusions:

• The proposed HCAI interaction paradigm effectively enhances AI usability in medical cytology.
• Explainability and user control are vital for successful AI integration in clinical practice.
• The findings offer valuable guidance for developing and deploying AI tools in medicine.