JMIR Publications

ABSTRACT

Traditional research ethics governance was designed for bounded protocols, identifiable investigators, and temporally limited encounters with human participants. Data science health research (DSHR) disrupts that architecture because health data, biological materials, computational representations, and models persist, travel, combine, and acquire new uses across time. We use DSHR broadly to include data science health research and adjacent secondary uses of health data, models, and biological materials, including learning health systems, public-health surveillance, quality improvement, operations, commercial product development, and artificial intelligence–enabled translational uses that rely on health data or material lineages. We introduce Ethical Governance Continuity Dissolution (EGCD), the progressive and sometimes irreversible loss of domain-specific governance authority across a data, model, or biological-material lineage, such that no coherent set of actors, instruments, and community processes can authorize, constrain, monitor, adjudicate, or remediate current use in relation to the persons and communities of origin. EGCD is distinct from consent staleness, function creep, contextual integrity violation, algorithmic drift, or a single defective data-use agreement. It is the systemic condition in which several such failures together compromise the authority needed to govern current use. Building on our prior work on representational veracity and the continuity trap, we propose a diagnostic architecture that assesses six governance-authority domains, applies three diagnostic criteria, scores each domain from 0 to 3 within a Continuity Authority Matrix, and stages severity from 0 to 4. We then propose an Ethical Continuity Governance and Response Mechanism (ECGRM) comprising a continuity registry, the Continuity Authority Matrix, trigger-based review, a Data Lifecycle Governance Officer function, a Continuity Dissolution Review Board, corrective and preventive action, cross-institutional audit, and community-governance integration. We use a publicly reported Royal Free–DeepMind Streams example to show how domain scores can identify Stage 2 or Stage 3 risk without converting the framework into a retrospective legal judgment. Implementation is proportionate, so that these functions may operate within existing structures, particularly in under-resourced institutions. The score and staging thresholds are conceptual triage aids, not validated metrics or automated determinations of ethical permissibility, and they require empirical validation and inter-rater reliability testing. Their purpose is to prevent the silent loss of governance authority while data, models, and biological materials continue to affect the lives, identities, and community standing of the persons and groups from whom they were derived. Lawful public-health surveillance under a competent authority is a legitimate governance handoff, not EGCD; EGCD arises when authority is not transferred, traceable, accountable, or remediable. This Viewpoint is addressed to research ethics committees and institutional review boards, data access committees, data stewards, biobank and registry leaders, artificial intelligence governance teams, regulators, funders, community-governance bodies, and investigators.