Academic machine learning researchers' ethical perspectives on algorithm development for health care: a qualitative...

Area of Science:

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

Background:

• Ethical considerations are crucial for developing machine learning (ML) tools for clinical care.
• Understanding the perspectives of ML researchers in medicine is vital for responsible innovation.
• The translation of ML tools into medical practice necessitates a thorough examination of ethical challenges.

Purpose of the Study:

• To describe the ethical considerations of academic machine learning researchers involved in developing ML tools for clinical care.
• To identify specific ethical challenges encountered during the development and training of medical ML algorithms.
• To explore researchers' views on necessary support and training for addressing ethical issues in medical AI.

Main Methods:

• Qualitative descriptive study design.
• In-depth, semistructured interviews with 10 ML researchers in medicine.
• Conventional qualitative content analysis to identify emergent themes.

Main Results:

• All interviewed researchers acknowledged the ethical significance of their ML development work.
• Concerns were raised regarding data sampling/labeling (bias, validity, integrity) and algorithm training/testing (reproducibility, target selection).
• Participants emphasized the need for increased interdisciplinary training and coordinated ethics integration.

Conclusions:

• Medical ML researchers face unique ethical standards and technical challenges impacting clinical acceptability.
• Increased support is needed in areas such as bias mitigation, data integrity, and reproducibility.
• Addressing these ethical considerations is key to the successful development and integration of medical ML tools.