Randomised controlled trials in medical AI: ethical considerations

Area of Science:

• Medical Informatics
• Artificial Intelligence
• Clinical Trials

Background:

• Artificial intelligence (AI) shows promise for medical diagnosis and prognosis.
• Current studies often lack evidence of clinical benefit for AI systems.
• There is a growing consensus for AI's clinical utility to be validated through clinical trials.

Purpose of the Study:

• To systematically examine the ethical considerations specific to conducting randomized controlled trials (RCTs) for AI in medicine.
• To address the methodological adjustments and ethical challenges posed by AI RCTs compared to traditional drug RCTs.
• To stimulate further discussion on the research ethics of medical AI.

Main Methods:

• Focus on core ethical principles: clinical equipoise, informed consent, and fairness.
• Analysis of ethical challenges unique to AI randomized controlled trials (RCTs).
• Systematic account development for the ethics of AI RCTs.

Main Results:

• AI RCTs necessitate distinct methodological adjustments compared to drug RCTs.
• These adjustments introduce unique ethical challenges.
• Key ethical principles require careful consideration in the context of AI clinical validation.

Conclusions:

• The ethical landscape of AI clinical trials requires careful navigation.
• Addressing informed consent, fairness, and clinical equipoise is crucial for ethical AI research.
• Further discourse is needed to establish robust ethical guidelines for medical AI development and deployment.