Bibliographic record
Abstract
BACKGROUND: Artificial intelligence (AI) has many different potential applications for healthcare system delivery, from imagery testing, triage, to the early detection of infectious patterns in large populations. Scientific literature shows that policies, legislation, and quasi-legislation on the question of AI are rapidly emerging and use one of four main regulatory strategies. Each strategy has limitations and can contribute to the creation of ethical and medico-legal risk, which our research contributes to identify. METHODS: We undertook a double-method literature review, first with a scoping literature review. We identified over 1200 documents from ~71 countries (from every continent), as well as international organizations that had adopted a form of regulatory strategy toward AI. We then proceeded to a narrative literature review where we highlighted both the taxonomy of regulatory strategies in use, and their respective risks and challenges to public health objectives. FINDINGS: The legal or quasi-legal framework surrounding the use of AI follows one of four possible strategies: trustworthiness, risk-reduction, ethical/principled, or indirect. The vast majority of the documents surveyed reveal that values and principles of public health are not sufficiently, or even explicitly, built-in the normative framework for health applications of AI (Health AI), and creates a wide array of challenges and risks, which our research helps to identify. CONCLUSIONS: This leads to a detrimental situation and leads to two main consequences: the creation of a "regulatory gap" for Health AI, and also a potential mismatch with already existing medico-legal duties of public health actors and health professionals.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.064 | 0.081 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.008 | 0.007 |
| Science and technology studies | 0.006 | 0.037 |
| Scholarly communication | 0.024 | 0.021 |
| Open science | 0.005 | 0.011 |
| Research integrity | 0.022 | 0.018 |
| Insufficient payload (model declined to judge) | 0.009 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".