AI in Health Research: WHO Ethics Review and Oversight Guidelines

Subject: GS Paper 4: Ethics

Context: Recently, WHO (World Health Organisation) released a report, Artificial Intelligence-related Health Research: Ethics Review and Oversight, calling for stronger ethical oversight of AI-enabled health research.

UPSC Online Preparation

About AI In Health Research

• Artificial Intelligence (AI) in health research refers to the use of AI technologies to process health data, generate research insights, and support the development and evaluation of health-related technologies and interventions. 
• Key Applications
  • Health Data Analysis: Processing large and complex health datasets to identify patterns and trends.
  • Scientific Discovery: Generating new research insights and hypotheses from health data.
  • Drug and Treatment Development: Supporting the discovery, testing, and evaluation of drugs and treatments.
  • Medical Technology: Developing and evaluating AI-enabled diagnostic and healthcare tools.
  • Research Automation: Assisting researchers with data processing, modeling and prediction.

Major Ethical Concerns Related To AI In Health Research

• Human Dignity and Rights: AI-driven research may risk undermining participants’ autonomy, dignity and fundamental rights, particularly through extensive data collection and automated decision-making.
• Privacy: Large-scale use of health data creates risks of data breaches, re-identification, misuse and unauthorised access to sensitive information.
• Bias and Discrimination: Biased datasets or algorithms may produce discriminatory outcomes. 
  • Example: An AI model trained mainly on data from one population may show lower accuracy for under-represented populations.
• Fairness and Equity: Unequal access to AI technologies, infrastructure and research capacity may widen existing health inequalities.
• Transparency: The complexity or “black-box” nature of some AI systems may make it difficult to understand how they were developed, trained, or arrived at particular outputs.
• Accountability: When AI produces an erroneous research finding or harmful outcome, responsibility may be unclear among researchers, developers, institutions, and regulators.
• Informed Consent: Participants may not fully understand how their health data will be used, analysed or reused by AI systems, weakening meaningful consent.
• Safety: AI systems may generate inaccurate, unreliable or unsafe outputs, creating risks if used without adequate validation and monitoring.
• Public Trust: Opaque, biased or unsafe use of AI can undermine public confidence in health research and healthcare institutions.

Limitations of Existing Ethics Oversight

• One-Time Review: Conventional ethics approval may occur before research begins, while AI models can change through new data, retraining and subsequent deployment.
• Limited AI Expertise: Some Research Ethics Committees may lack expertise to assess algorithmic bias, model validation, data governance and explainability.
• Unclear Responsibility: AI research involves researchers, technology developers, data providers, institutions and regulators, making accountability more complex.
• Beyond Individual Participants: Traditional review may focus on research participants, while AI systems can create wider effects through population-level bias, unequal access and data concentration.
• Rapid Technological Change: Governance frameworks can become outdated as AI capabilities and applications evolve rapidly.

Click to Know UPSC OnlyIAS Coaching Centres

Measures Suggested By WHO To Strengthen Ethical Oversight of AI-Related Health Research

• Strengthen Research Ethics Committees
  • AI Expertise: Equip Research Ethics Committees (RECs) with specialised AI and data expertise.
  • Capacity Building: Provide adequate training and resources to handle complex AI research.
  • Lifecycle Oversight: Extend ethical review from study design to implementation.
  • Risk Assessment: Strengthen assessment of bias, privacy, transparency, accountability and safety.
• Promote Early Ethical Risk Management
  • Researchers: Identify and mitigate ethical risks at the research-design stage.
  • Transparency: Clearly communicate AI use, limitations and potential risks.
  • Societal Impact: Consider the broader implications of AI technologies, beyond individual research participants.
• Establish Multi-Stakeholder Oversight
  • Funders: Integrate ethical safeguards into funding decisions.
  • Scientific Journals: Promote responsible and transparent reporting.
  • Data Governance Bodies: Strengthen responsible health-data governance.
  • Professional Societies: Develop ethical standards and guidance.
  • Regulators: Ensure safety, accountability and compliance.
  • Policymakers: Address systemic risks beyond individual projects.
• Adopt Adaptive And Lifecycle-Based Governance
  • Continuous Oversight: Move beyond one-time ethics approval.
  • Agile Governance: Develop oversight mechanisms capable of responding to rapid technological change.
  • Evidence-Based Approach: Base governance on emerging evidence and evolving AI risks.
  • Future Standards: Use current guidance as a foundation for developing future standards and governance frameworks.
• Promote Equity In AI Health Research
  • LMIC Capacity: Strengthen research and institutional capacity in low- and middle-income countries (LMIC).
  • Local Leadership: Increase local leadership and participation in AI research.
  • Equitable Benefits: Ensure AI advances benefit all populations and countries.
• Multidisciplinary Approach: Bring together ethics, health, AI, data and regulatory expertise.
• Global Cooperation: Encourage collaboration across countries and sectors to develop responsible AI governance.

News Source: WHO

Check Out UPSC CSE Books

Visit PW Store