Harnessing AI in critical care: opportunities, challenges and key steps for success
The DOI matches its Crossref record; 84 of the 85 references it cites were found in a registry or an index.
Abstract
BACKGROUND: The integration of artificial intelligence (AI) into critical care offers significant potential to enhance early diagnosis, predict patient deterioration, personalise treatment and inform clinical decision-making. Despite this promise, AI adoption in the intensive care unit (ICU) faces challenges, as illustrated by the limited number of AI tools which have been approved for clinical use and/or successfully deployed in critical care. METHODS: Aims of the review are to provide a synthesis of research on AI in critical care; assess approved tools; and consider challenges and opportunities, focusing on the different phases of the AI algorithm lifecycle in the ICU, including data collection, modelling, validation, implementation and post-deployment monitoring. Peer-reviewed publications were searched using terms relevant to AI and critical care spanning the years 2000-2025. RESULTS: Research on AI applications in the ICU is characterised by significant limitations including suboptimal data quality, retrospective analyses and a paucity of prospective validation studies. The few AI algorithms that have received Food and Drug Administration approval for use in the ICU have not gained widespread clinical adoption due, in part, to issues such as lack of user trust, integration challenges, unclear clinical impact or performance drift. Overcoming these barriers will require a structured approach that addresses the key challenges identified in the AI lifecycle, including the integration of real-world data, post-deployment performance monitoring, governance and ethical considerations. A successful implementation pathway should consider realistic goal-setting, greater model explainability, improved workflow integration and active end-user involvement. CONCLUSIONS: Advancing critical care with AI will require special attention to interdisciplinary collaboration, robust validation frameworks and adaptive governance models. The need for rigorous scientific evaluation needs to be balanced with the pressure for rapid deployment. Ensuring transparency, safety and alignment with clinical workflows will be critical to achieving meaningful AI integration in critical care.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Citations
Cites
"Why Should I Trust You?": Explaining the Predictions of Any Classifier
Ribeiro, M. T., Singh, S., & Guestrin, C. (2016) · Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining · doi:10.1145/2939672.2939778 · Cited by 17,555
Rudin, C. (2019) · Nature Machine Intelligence · doi:10.1038/s42256-019-0048-x · Cited by 10,561 · cited 1×
“71 72 Achieving high performance and high explainability requires innovative design choices or hybrid models that balance these factors effectively.”
“Why Should I Trust You?”: Explaining the Predictions of Any Classifier
Ribeiro, M., Singh, S., & Guestrin, C. (2016) · Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Demonstrations · doi:10.18653/v1/n16-3020 · Cited by 6,682
A Survey on Bias and Fairness in Machine Learning
Mehrabi, N., Morstatter, F., Saxena, N., Lerman, K., & Galstyan, A. (2022) · ACM Computing Surveys · doi:10.1145/3457607 · Cited by 5,812
Floridi, L., Cowls, J., Beltrametti, M., Chatila, R., Chazerand, P., Dignum, V., Luetge, C., Madelin, R., Pagallo, U., Rossi, F., et al. (2018) · Minds and Machines · doi:10.1007/s11023-018-9482-5 · Cited by 4,774 · cited 2×
“65 Control problems arise from the inherent complexity of AI systems, particularly those employing DL.”
Key challenges for delivering clinical impact with artificial intelligence
Kelly, C. J., Karthikesalingam, A., Suleyman, M., Corrado, G., & King, D. (2019) · BMC Medicine · doi:10.1186/s12916-019-1426-2 · Cited by 3,136
Explainable machine-learning predictions for the prevention of hypoxaemia during surgery
Lundberg, S. M., Nair, B., Vavilala, M. S., Horibe, M., Eisses, M. J., Adams, T., Liston, D. E., Low, D. K.-W., Newman, S.-F., Kim, J., et al. (2018) · Nature Biomedical Engineering · doi:10.1038/s41551-018-0304-0 · Cited by 1,967
Komorowski, M., Celi, L. A., Badawi, O., Gordon, A. C., & Faisal, A. A. (2018) · Nature Medicine · doi:10.1038/s41591-018-0213-5 · Cited by 1,290
A Review on Outlier/Anomaly Detection in Time Series Data
Blázquez-García, A., Conde, A., Mori, U., & Lozano, J. A. (2022) · ACM Computing Surveys · doi:10.1145/3444690 · Cited by 1,102
Vasey, B., Nagendran, M., Campbell, B., Clifton, D. A., Collins, G. S., Denaxas, S., Denniston, A. K., Faes, L., Geerts, B., Ibrahim, M., et al. (2022) · Nature Medicine · doi:10.1038/s41591-022-01772-9 · Cited by 1,039 · cited 2×
“Additionally, frameworks like the DECIDE-AI guidelines 50 77 offer structured evaluation methods, similar to those used for traditional medical devices, ensuring that AI tools are rigorously tested and optimised for clinical safety and effectiveness.”
Cited by
Recent approaches of artificial intelligence in intensive care unit: A review
Havelikar, U. A., Bembde, A., Patil, D., Shinde, A., Sharma, R. K., Patel, V. P., & Chaudhari, N. B. (2025) · Intelligent Hospital · doi:10.1016/j.inhs.2025.100030 · Cited by 17
Żerdziński, K., Janiec, J., Jóźwik, K., Łajczak, P., & Krzych, Ł. (2025) · Journal of Clinical Medicine · doi:10.3390/jcm15010185 · Cited by 11
Artificial intelligence-driven triage and decision-making in trauma systems/settings
Kalimouttou, A., Perkins, Z., & Gauss, T. (2026) · Current Opinion in Anaesthesiology · doi:10.1097/aco.0000000000001619 · Cited by 6
Self-citationJing, S., Tie, X., Wei, Y., Zhou, R., Zou, T., Qin, Y., Zeng, X., Li, Y., & Yin, W. (2025) · Frontiers in Medicine · doi:10.3389/fmed.2025.1622604 · Cited by 9
Artificial Intelligence in Cardiovascular Medicine: A Giant Step in Personalized Medicine?
Jankauskas, S. S., Varzideh, F., Kansakar, U., & Santulli, G. (2026) · Journal of Personalized Medicine · doi:10.3390/jpm16040192 · Cited by 11
Metadata
- Authors
- Kalimouttou, A., Stevens, R. D., & Pirracchio, R.
- Affiliations
- Centre Hospitalier Universitaire de Grenoble (FR)
- Johns Hopkins University (US)
- University of California, San Francisco (US)
- Corresponding author
- Pirracchio, Romain
- Venue
- Thorax
- Publisher
- BMJ
- Published
- 2025-06-15
- Volume
- 81
- Issue
- 2
- Pages
- 183-192
- ISSN
- 0040-6376 (ISSN-L)
- 1468-3296
- OpenAlex
- W4411319594
- Semantic Scholar
- cb8e489f624e8bfa396def854e9ee866a3278492
- Cited by
- 12
- References
- 85
- Citations by year
- 2026: 8 · 2025: 4
- FWCI
- 0.71
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- Thorax (publishedVersion)
- PubMed (publishedVersion)
- Funders
- Common Fund (R01EB035028-01)
- Patient-Centered Outcomes Research Institute (P0562155)
- Amomed Pharma (no grant number)
- Topics
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning in Healthcare
- Health Sciences › Medicine › Surgery › Healthcare Technology and Patient Monitoring
- Keywords
- intensive care unit, AI adoption barriers, model validation, clinical decision support, clinical deterioration prediction, early diagnosis, model explainability, post-deployment monitoring, performance drift, AI governance, ethical considerations, real-world data, clinical workflow integration, user trust
- MeSH terms
- Algorithms, Artificial Intelligence, Critical Care, Humans, Intensive Care Units
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Harnessing AI in critical care: opportunities, challenges and key steps for success. https://app.cochat.ai/papers/10.1136/thorax-2024-222125
- The paper
- Kalimouttou, A., Stevens, R. D., & Pirracchio, R. (2026). Harnessing AI in critical care: opportunities, challenges and key steps for success. Thorax. https://doi.org/10.1136/thorax-2024-222125
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.