Hospital AI Lab Practical information on medical AI for Indian hospitals

Updated:

ICMR’s ten AI principles, read by a hospital administrator

ICMR’s 2023 guidelines list ten ethical principles for AI in healthcare. What each one means when you are reading a vendor proposal.

ICMR’s Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare (2023) were prepared by the DHR-ICMR Artificial Intelligence Cell. They are ethical guidelines for research and healthcare, and this note treats them as guidance, not as a statute. They are worth an administrator’s time because their principles map onto questions you already ask a vendor.

The ten principles

Section 1 lists: autonomy; safety and risk minimization; trustworthiness; data privacy; accountability and liability; optimization of data quality; accessibility, equity and inclusiveness; collaboration; non-discrimination and fairness; and validity.

How to read them for a proposal

  • Autonomy. Does the tool leave the patient and the clinician in charge, or does it act on its own?
  • Safety and risk minimization. What can go wrong, how would you notice, and who can switch it off?
  • Trustworthiness and validity. Was it tested on patients like yours, and by whom?
  • Data privacy and data quality. Where does the data go, and how is its quality kept up?
  • Accountability and liability. Who is answerable when it errs: the vendor, the hospital, the clinician?
  • Accessibility, equity, non-discrimination and fairness. Does it work for your languages and regions, and for every patient group?

Beyond the principles

Section 2 gives guiding principles for the development, validation and clinical deployment phases. Section 3 covers ethical review, Section 4 informed consent and Section 5 governance of AI use, and the document ends with an ethics checklist. The introduction also describes medical AI for hospital management systems, such as help with scheduling and with coordination between departments, and names ethical concerns including algorithmic transparency and explainability, clarity on liability, accountability and oversight, and bias and discrimination.

Key takeaways

  • ICMR’s 2023 guidelines list ten ethical principles for AI in biomedical research and healthcare.
  • Each maps to an ordinary vendor question: who decides, what can go wrong, who is answerable, who is left out.
  • They are guidance for ethics review and governance, and end with an ethics checklist you can borrow from.

Sources: ICMR, Ethical Guidelines for Application of Artificial Intelligence in Biomedical Research and Healthcare (2023)

This article is part of a series

All series

← Blog