AI in medical diagnosis: AI prediction & human judgment

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Clinical Decision Support

Background:

• Artificial intelligence (AI) is increasingly viewed as a tool to augment human capabilities in knowledge work, including medical decision-making.
• However, the integration of AI into clinical practice necessitates understanding the human factors involved in decision-making, which extend beyond algorithmic analysis.
• Factors such as creativity, intuition, emotions, and value judgments are crucial human elements in diagnosis.

Purpose of the Study:

• To explore dermatologists' expectations regarding AI applications in medical diagnosis.
• To identify key dimensions influencing the adoption and beneficial use of AI in clinical settings.
• To inform the development of AI tools that effectively support, rather than replace, medical professionals.

Main Methods:

• Conducted semi-structured, open-ended research interviews with 17 dermatologists.
• Utilized qualitative research methods to gather in-depth insights into user expectations and perceptions.
• Analyzed interview data to identify aggregate dimensions of dermatologists' thinking regarding AI.

Main Results:

• Identified four key dimensions: interaction with AI, responsibility, explainability, and the necessary mindset shift for AI collaboration.
• Dermatologists view AI as a supportive asset, emphasizing the need for AI to complement human expertise.
• Findings highlight the importance of user-centered design for AI in medical diagnosis.

Conclusions:

• AI can be a valuable tool to support physicians in medical diagnosis, but it should not supplant human decision-makers.
• Understanding dermatologists' expectations regarding AI interaction, responsibility, and explainability is crucial for successful implementation.
• The findings offer guidance for physicians and AI vendors to foster beneficial integration of AI in diagnostic processes.