MétaCan
Menu
Back to cohort
Record W4415749253 · doi:10.2196/81421

Implementation, Experiences, Impact, and Costs of Artificial Intelligence in Chest Diagnostics: Protocol for a Mixed Methods Evaluation

2025· article· en· W4415749253 on OpenAlexvenueno aff
Angus I. G. Ramsay, Chris Sherlaw‐Johnson, Kevin Herbert, Stuti Bagri, Malina Bodea, Nadia Crellin, Holly Elphinstone, Amanda Halliday, Nina Hemmings, R. C. Lawrence, Cyril Lobont, Pei Li Ng, John Lloyd, Efthalia Massou, Raj Mehta, Stephen Morris, Jenny Shand, Holly Walton, Naomi Fulop

Bibliographic record

VenueJMIR Research Protocols · 2025
Typearticle
Languageen
FieldMedicine
TopicCOVID-19 diagnosis using AI
Canadian institutionsnot available
Fundersnot available
KeywordsProtocol (science)Evaluation methodsData collectionProgram evaluationSoftware

Abstract

fetched live from OpenAlex

BACKGROUND: The ability to perform complex tasks has seen artificial intelligence (AI) used to support radiology in clinical settings, including lung cancer detection and diagnosis. Evidence suggests that AI can contribute to accurate diagnosis, reduce errors, and improve efficiency. The National Health Service England (NHSE)-funded Artificial Intelligence Diagnostic Fund (AIDF) is currently supporting 12 National Health Service (NHS) networks to implement AI for chest diagnostic imaging. There is, however, limited evidence on real-world AI implementation and use, including staff, patient, and caregiver experience, and costs and cost-effectiveness. A National Institute for Health and Care Research Rapid Service Evaluation Team Phase 1 evaluation provided insights into the early implementation of these tools and developed a framework for monitoring and evaluation of AI tools for chest diagnostic imaging in practice. OBJECTIVE: This mixed methods evaluation of AI tools for chest diagnostic imaging aims to address previous research gaps by exploring the implementation of AI tools for chest diagnostic imaging, the impact and costs of implementing these service models, and the experiences of patients, caregivers, and staff. METHODS: This study will be a mixed method evaluation of implementation, experiences, impact, and costs of AI for chest diagnostic imaging in NHS services in England, with the evaluation informed by the Major System Change Framework. Trust-level case studies (3 in-depth and up to 9 light-touch) will be performed, including staff member, patient, and caregiver; NHSE AIDF team interviews; meeting observations; and analysis of key relevant documentation. Qualitative data will be analyzed using Rapid Assessment Procedures and inductive thematic analysis, supplemented by in-depth deductive thematic analysis. Data from case study sites and other relevant sources will be used to assess outcomes at the other sites and for comparators. A pragmatic economic model of the chest diagnostic imaging pathway will be developed to estimate key costs and resource use associated with AI tool deployment. Together with input from national stakeholders and staff workshops, the study findings will then be finalized for reporting. RESULTS: As of September 2025, trust-level research and development approvals with participating sites are complete, and data collection has commenced. Results are expected to be reported by the end of February 2026. CONCLUSIONS: The study will provide new insights into the facilitators and barriers to the adoption of AI technology in health care and the perceptions of both the general public and health care staff on its use. It will also inform best practices in approaches for service performance evaluation, for the implementation of AI into existing care pathways, and for the development of models to best support evidence-based decision-making. It will thus establish a framework upon which the greatest benefits of the use of AI in health care can be realized. INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID): DERR1-10.2196/81421.

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 distilled prediction

Teacher imitation

Not 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.

metaresearch head score (Codex)0.008
metaresearch head score (Gemma)0.007
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Other design · Consensus signal: none
GenreCandidate signal: Protocol · Consensus signal: Protocol
Teacher disagreement score0.959
Threshold uncertainty score0.838

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0080.007
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0000.000
Bibliometrics0.0010.001
Science and technology studies0.0000.000
Scholarly communication0.0000.000
Open science0.0000.000
Research integrity0.0000.000
Insufficient payload (model declined to judge)0.0000.000

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.

Opus teacher head0.426
GPT teacher head0.720
Teacher spread0.294 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one teacher head, not a consensus.

The models applied no category: nothing in the taxonomy fit this work.
Study designOther design
Domainnot available
GenreProtocol

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".

Quick stats

Citations0
Published2025
Admission routes1
Has abstractyes

Explore more

Same venueJMIR Research ProtocolsSame topicCOVID-19 diagnosis using AIFrench-language works237,207