The Evolution of Consent in Biomedical AI: Ensuring Privacy and Equity
Introduction
The evolution of consent in biomedical AI is crucial in maintaining privacy and equity. As AI technologies advance, traditional consent models must adapt to safeguard individuals’ privacy and promote inclusive healthcare. This article examines the complexities of consent in the context of biomedical AI as published in the New England Journal of Medicine (NEJM), highlighting the need for updated frameworks that address modern challenges.
Can Consent Protect Privacy?
Informed consent, as conceived in the 1970s, was designed to protect individuals’ privacy by allowing them to control their data. However, in today’s data-intensive environment, this approach is increasingly ineffective. Algorithms now draw inferences from data, making it possible to infer information about non-consenting individuals. This systemic privacy loss necessitates a reevaluation of consent’s role in protecting privacy.
Consent’s Disparate Impact
Consent rules, although neutral, can have discriminatory impacts. Vulnerable groups, such as migrants or individuals in states with restrictive reproductive laws, are more likely to opt out of data sharing. This underinclusion in AI training data can lead to tools that do not work effectively for these groups. Thus, consent rights, once a bioethical triumph, now risk undermining equity and safety in biomedical AI.
Diminished Public Voice
Consent rights, intended to empower individuals, have fragmented the public’s voice in privacy policy. Data handlers often cater to the least demanding consenters, ignoring broader societal privacy concerns. This individualistic approach contrasts with collective bargaining in labour rights, where collective action amplifies workers’ voices. Current research regulations lack mechanisms for collective public action to influence data use policies.
Failure to Minimise Privacy Risks
Modern research regulations require ethical review boards to minimise risks to human subjects. However, these boards often lack the expertise to manage privacy risks effectively. Consequently, informed consent is sometimes treated as a formality rather than a robust privacy safeguard. This approach, prioritising expediency over trust, risks public backlash that could hinder scientific progress.
Consent and Reciprocity
The UK’s Care.Data project illustrates the pitfalls of ignoring public trust in data initiatives. Despite legal oversight, public backlash led to its termination. Postmortem studies identified reciprocity, nonexploitation, and service of the public good as key principles for public buy-in. Ethical data handling and clear communication with the public are essential for maintaining trust and ensuring the success of biomedical AI projects.
Reciprocity and Open Data Sharing
Open data sharing can coexist with reciprocity if privacy and data use policies are clear and inclusive. Agencies like the National Institutes of Health must establish transparent policies and involve the public in decision-making processes. Biomedical AI, as a collective endeavour, benefits from broad public participation, which top-down policy-setting cannot achieve.
Is Co-Creation of Consent Realistic?
Engaging research participants in co-creating consent is feasible with AI-powered digital tools. Platforms like vTaiwan demonstrate how large groups can deliberate and reach consensus on complex issues. Biomedical AI requires similar tools to involve patients in shaping consent standards, thereby enhancing democratic legitimacy and trust in research.
AI-Powered Bioethics
AI-powered digital tools enable large-scale public engagement in policy-making. Examples like the CrowdLaw Catalog show that governments and institutions can use these tools to gather public input. Biomedical AI must adopt similar approaches to develop ethical frameworks that reflect public values and concerns.
Conclusion
The evolution of consent in biomedical AI is essential for ensuring privacy and equity in healthcare. Traditional consent models must adapt to address modern challenges, involving the public in shaping data use policies. AI-powered bioethics can facilitate this process, fostering trust and inclusivity in biomedical research.
