Defining change: Exploring expert views about the regulatory challenges in adaptive artificial intelligence for healthcare
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Continuously learning or adaptive artificial intelligence (AI) applications for screening, diagnostic and other clinical services are yet to be widely deployed. This is partly due to existing device regulation mechanisms that are not fit for purpose regarding the adaptive features of AI. This study aims to identify the challenges in and opportunities for the regulation of adaptive features of AI. We performed in-depth qualitative, semi-structured interviews with a diverse group of 72 experts in high-income countries (Australia, Canada, New Zealand, US, and UK) who are involved in the development, acquisition, deployment and regulation of healthcare AI systems. Our findings revealed perceived challenges in the regulation of adaptive features of machine learning (ML) systems. These challenges include the complexity of AI applications as products subject to regulation; lack of accepted definitions of adaptive changes; diverse approaches to defining significant adaptive change; and lack of clarity about regulation of adaptive change. Our findings reflect potentially competing interests among different stakeholders and diversity of approaches from regulatory bodies and legislators in different jurisdictions across the globe. In addition, our findings highlight the complex regulatory implications of adaptive AI that differ from traditional medical products, drugs or devices. The perceived regulatory challenges raised by adaptive features of AI applications require high-level coordination within a complex regulatory ecosystem that consists of medical device regulators, professional accreditation agencies, professional medical organisations, and healthcare service providers. Regulatory approaches should complement existing safety protocols with new governance mechanisms that specifically take into account the variety of roles and responsibilities that will be required to monitor, evaluate and oversee adaptive changes.
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.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 it