ICMR Releases Ethical Guidelines for AI in Healthcare and Biomedical research

In March 2023, the Indian Council of Medical Research (“ICMR”) released the “Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare” (“AI Guidelines”). These guidelines provide a comprehensive framework addressing ethical principles, governance, and informed consent when deploying AI-based solutions in biomedical research and healthcare.

Key ethical principles outlined in the AI Guidelines include:

• Autonomy: Patients should have oversight and control over medical decision-making involving AI technologies, and a consent process must provide patients the autonomy to reject AI technology.
• Safety and Risk Minimization: Stakeholders are responsible for the safety of AI technologies, requiring robust control mechanisms, secure systems, pre-emptive design, anonymization of data, and compliance with legal and data protection requirements.
• Trustworthiness: Diagnostic or prognostic tools in AI technologies must be lawful, ethical, reliable, valid, explainable, transparent, and independent of a physician’s views.
• Data Privacy: Emphasis on privacy protection throughout AI technology development and deployment, including anonymization of patient data, adherence to privacy laws, and user control over collected data.
• Accountability and Liability: AI technologies should undergo audits, involve health sector representatives, and assign responsibility based on the facts of the case, holding designers or manufacturers responsible for malfunctions.
• Optimization of Data Quality: Ensuring unbiased training data representation of the Indian population to eliminate errors in AI-based technology.
• Accessibility, Equity, and Inclusiveness: Fair distribution and equal access to AI technology, particularly for socially and economically disadvantaged groups, with the use of local languages in the user interface.
• Collaboration: Encouraging collaboration with AI experts, both interdisciplinary and international, to keep pace with the evolving nature of AI.
• Non-discrimination and Fairness: Eliminating biases in training datasets and promoting equality, along with providing a safe mechanism to address concerns regarding AI technology.
• Validity: Clinical and field validation of AI technology before application to ensure patient safety and efficacy.

The AI Guidelines also provide detailed guiding principles for the development, validation, and deployment phases. Ethical review procedures in medical AI are recommended, involving experts from the legal field, and researchers are required to obtain written informed consent from study participants.