Clinician Trust in AI-Generated Radiology Narratives for Patient Care Decisions
DOI:
https://doi.org/10.61919/gk9qxd94Keywords:
artificial intelligence; radiology narratives; clinician trust; qualitative research; thematic analysis; clinical decision-making; large language models; patient care; clinical governance.Abstract
**Background:** AI-generated radiology narratives are increasingly being considered for drafting, summarising, simplifying and organising radiology report information, but their safe use depends on how clinicians judge trust, uncertainty, accountability and workflow fit when patient-care decisions are involved. **Objective:** This qualitative study explored how clinicians judge the usefulness, limitations and safeguards of AI-generated radiology narratives in clinical decision-making. **Methods:** A qualitative design was used based on semi-structured interviews with ten clinicians from radiology, emergency medicine, oncology, respiratory medicine, orthopaedics, neurology, intensive care, cardiology and general practice. Participants discussed their experiences with radiology reports and their perceptions of AI-generated summaries, impressions, patient-friendly explanations and workflow-integrated narrative outputs. Reflexive thematic analysis was used to identify patterned meanings related to trust, risk, traceability, oversight and governance. **Results:** Five themes were generated: narrative usefulness as cognitive scaffolding rather than replacement; traceability and validation as the foundation of trust; accountability anxiety in higher-risk decisions; workflow fit and interprofessional communication; and provisional trust shaped by governance and learning. Clinicians valued AI-generated narratives when they organised lengthy reports, highlighted key findings and reduced cognitive load, but trust depended on clear labelling, source linkage, uncertainty preservation, radiologist verification, feedback routes, audit trails and local validation. Higher-risk decisions such as cancer staging, emergency discharge, surgical referral and critical-care escalation required stronger human oversight. **Conclusion:** Clinician trust in AI-generated radiology narratives is conditional, provisional and socio-technical. Safe implementation should prioritise transparent, traceable and governed narrative functions that support calibrated reliance rather than replacing radiologist judgement or professional accountability.
References
1. Royal College of Radiologists. Clinical radiology workforce census 2024. London: Royal College of Radiologists; 2025. Available from: https://www.rcr.ac.uk/news-policy/policy-reports-initiatives/clinical-radiology-census-reports/
2. Singh R, et al. How AI is used in FDA-authorized medical devices. NPJ Digit Med. 2025. Available from: https://www.nature.com/articles/s41746-025-01800-1
3. European Society of Radiology. What the radiologist should know about artificial intelligence. Insights Imaging. 2023. Available from: https://doi.org/10.1186/s13244-023-01435-4
4. Chou SHS, et al. Integrating artificial intelligence into radiology workflow: deployment lessons. Radiology. 2022. Available from: https://pubs.rsna.org/
5. Bae MS, et al. Radiology report quality, structured reporting and communication in clinical care. Korean J Radiol. 2021. Available from: https://doi.org/10.3348/kjr.2020.0946
6. Keshavarz P, et al. ChatGPT in radiology: a systematic review of performance, pitfalls and future directions. Clin Imaging. 2024. Available from: https://www.sciencedirect.com/science/article/pii/S2211568424001050
7. Alabed S, et al. Large language models for simplifying radiology reports: systematic review and meta-analysis. Lancet Digit Health. 2026. Available from: https://www.thelancet.com/journals/landig/article/PIIS2589-7500(25)00142-6/fulltext
8. Huang J, et al. Efficiency and quality of generative AI-assisted radiology reporting in clinical care. JAMA Netw Open. 2025. Available from: https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2834943
9. Yu F, et al. Heterogeneity and predictors of the effects of AI assistance on radiologists. Nat Med. 2024. Available from: https://www.nature.com/articles/s41591-024-02850-w
10. McKinney SM, Sieniek M, Godbole V, Godwin J, Antropova N, Ashrafian H, et al. International evaluation of an AI system for breast cancer screening. Nature. 2020;577:89-94. doi:10.1038/s41586-019-1799-6
11. Lång K, et al. Artificial intelligence-supported screen reading versus standard double reading in mammography screening. Lancet Oncol. 2023. Available from: https://doi.org/10.1016/S1470-2045(23)00298-X
12. Muehlematter UJ, Daniore P, Vokinger KN. Approval of artificial intelligence and machine learning-based medical devices in the USA and Europe. Lancet Digit Health. 2021. Available from: https://doi.org/10.1016/S2589-7500(20)30292-2
13. van Leeuwen KG, Schalekamp S, Rutten MJCM, van Ginneken B, de Rooij M. Artificial intelligence in radiology: 100 commercially available products and their scientific evidence. Eur Radiol. 2021. Available from: https://doi.org/10.1007/s00330-021-07892-z
14. Reichenpfader D, Muller H, Denecke K. Large language model-based information extraction from radiology reports: a scoping review. NPJ Digit Med. 2024. Available from: https://www.nature.com/npjdigitalmed/
15. Sunshine A, et al. Evaluating the quality and understandability of ChatGPT-generated radiology report summaries. JMIR Form Res. 2025. Available from: https://formative.jmir.org/2025/1/e76097/PDF
16. Agarwal N, et al. RadCliQ: clinical evaluation and quality scoring for radiology report generation. Proc Mach Learn Res. 2023. Available from: https://arxiv.org/abs/2301.06590
17. Gaur S, et al. Radiology report generation and clinical correctness: review of evaluation challenges. IEEE Rev Biomed Eng. 2024. Available from: https://doi.org/10.1109/RBME.2023.3305984
18. Asan O, Bayrak AE, Choudhury A. Artificial intelligence and human trust in healthcare: focus on clinicians. J Med Internet Res. 2020. Available from: https://doi.org/10.2196/15154
19. Glikson E, Woolley AW. Human trust in artificial intelligence: review of empirical research. Acad Manag Ann. 2020. Available from: https://doi.org/10.5465/annals.2018.0057
20. Bucinca Z, Malaya MB, Gajos KZ. To trust or to think: cognitive forcing functions can reduce overreliance on AI. Proc ACM Hum Comput Interact. 2021. Available from: https://doi.org/10.1145/3449287
21. Poursabzi-Sangdeh F, et al. Manipulating and measuring model interpretability and its impact on human-AI decision making. Proceedings of CHI. 2021. Available from: https://doi.org/10.1145/3411764.3445315
22. Tonekaboni S, Joshi S, McCradden MD, Goldenberg A. What clinicians want: contextualizing explainable machine learning for clinical end use. Machine Learning for Healthcare. 2019. Available from: https://arxiv.org/abs/1905.05134
23. Gichoya JW, et al. AI recognition of patient race in medical imaging: a modelling study. Lancet Digit Health. 2022. Available from: https://doi.org/10.1016/S2589-7500(22)00063-2
24. Seyyed-Kalantari L, et al. Underdiagnosis bias of artificial intelligence algorithms applied to chest radiographs in underserved patient populations. Nat Med. 2021. Available from: https://doi.org/10.1038/s41591-021-01595-0
25. World Health Organization. Ethics and governance of artificial intelligence for health. Geneva: World Health Organization; 2021. Available from: https://www.who.int/publications/i/item/9789240029200
26. Geis JR, et al. Ethics of artificial intelligence in radiology: summary of the joint European and North American multisociety statement. Radiology. 2019. Available from: https://doi.org/10.1148/radiol.2019191586
27. European Society of Radiology. Guiding AI in radiology: ESR's recommendations for effective application of the AI Act. Insights Imaging. 2025. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC11825415/
28. Kelly CJ, Karthikesalingam A, Suleyman M, Corrado G, King D. Key challenges for delivering clinical impact with artificial intelligence. Lancet Digit Health. 2019. Available from: https://doi.org/10.1016/S2589-7500(19)30141-4
29. Beede E, et al. A human-centered evaluation of a deep learning system deployed in clinics for the detection of diabetic retinopathy. Proceedings of CHI. 2020. Available from: https://doi.org/10.1145/3313831.3376718
30. Creswell JW, Poth CN. Qualitative inquiry and research design: choosing among five approaches. 4th ed. Thousand Oaks: SAGE Publications; 2018. Available from: https://us.sagepub.com/en-us/nam/qualitative-inquiry-and-research-design/book246896
31. Braun V, Clarke V. Thematic analysis: a practical guide. London: SAGE Publications; 2021. Available from: https://uk.sagepub.com/en-gb/eur/thematic-analysis/book248481
32. Byrne D. A worked example of Braun and Clarke's approach to reflexive thematic analysis. Qual Quant. 2022. Available from: https://doi.org/10.1007/s11135-021-01182-y
Downloads
Published
Issue
Section
License
Copyright (c) 2025 Huang Yuzhe, Gilbert Lister, Armelia Adel Abdullah (Author)
This work is licensed under a Creative Commons Attribution 4.0 International License.
© The Authors. This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0).
.png)
