Artificial Intelligence to Facilitate SEP-1 Measure Compliance and Fluid Management in Sepsis
Open accessThe DOI matches its Crossref record; 80 of the 80 references it cites were found in a registry or an index; 93 of the 116 sources its claims cite were checked for support.
Abstract
Sepsis remains a leading cause of preventable morbidity and mortality worldwide, and adherence to the Centers for Medicare & Medicaid Services Severe Sepsis and Septic Shock Early Management Bundle (SEP-1) remains modest and variable across institutions. Simultaneously, controversy persists regarding fixed-volume fluid resuscitation mandates, particularly given the increasing emphasis on individualized, physiology-guided management. Artificial intelligence (AI) has emerged as a potential strategy to address both operational and clinical gaps in sepsis care. This review examines the current state of SEP-1 implementation, key barriers to compliance, and ongoing debates surrounding early fluid administration. We then discuss contemporary evidence on AI-enabled tools designed to accelerate bundle processes and support personalized fluid management. Early warning systems, natural language processing-augmented models, and telemedicine-integrated platforms have demonstrated improvements in process measures such as time-to-antibiotics and bundle component completion when embedded within defined clinical workflows. Reinforcement learning, causal machine learning, and predictive models offer promise for individualized fluid strategies, although most data remain retrospective and hypothesis-generating. Successful integration will require prospective validation, clinician-in-the-loop oversight, governance frameworks, and continuous monitoring for safety, equity, and model drift. AI should augment-rather than replace-clinical judgment to improve reliability, timeliness, and personalization in sepsis management.
Overview
CoChat Summary
TL;DR
This is a narrative literature review on how AI might improve SEP-1 compliance and personalize fluid management in sepsis, while warning that most evidence is retrospective and implementation-dependent. The paper argues AI should augment clinician judgment, not replace it, until prospective validation shows safety and patient-centered benefit.
Key findings
- National SEP-1 performance is described as modest and variable: mean hospital compliance is approximately 49%, and one ED network reported mean compliance of approximately 54%.
- SEP-1 abstraction is unreliable in practice: in a study across three hospitals, abstractors agreed on “time zero” in only 36% of cases.
- The review cites AI-enabled sepsis tools that improved process measures, including an end-to-end telesepsis intervention across five emergency departments and over 56,000 ED encounters with 1,233 confirmed sepsis cases, improving SEP-1 compliance from 68.4% to 78.3% (p = 0.002).
- For fluid management, the paper highlights multiple AI studies with promising performance, including models with AUROC ~0.83–0.85 for fluid responsiveness, 77.8% predictive accuracy for individualized fluid needs, and a sepsis early risk assessment model with AUROC up to 0.94 that increased early detection by 32% while reducing false positives by up to 17%.
Methods
- Narrative review, not a formal systematic review: the authors searched PubMed, Google Scholar, and Scopus, prioritizing studies from 2020 to the present, and also screened reference lists.
- Included studies on AI-based sepsis prediction, decision support, and hemodynamic optimization, with emphasis on SEP-1/sepsis bundle compliance and fluid resuscitation.
- Synthesis focused on reported workflow integration, process measures, and patient-centered outcomes across heterogeneous study designs.
Limitations
- The paper states that a formal systematic review was not performed because of heterogeneity in study methods and outcomes.
- The authors repeatedly note that most available evidence is retrospective and hypothesis-generating, with limited prospective randomized evidence; this is stated in the paper.
- Evident from the text: findings are drawn from a narrative synthesis of selected studies, so conclusions are vulnerable to selection bias and cannot support pooled effect estimates.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures
![Figure 1 Artificial intelligence lifecycle in critical care medicine highlighting key failure points and safeguards [35,37,38,39,40,41,43,44].](https://figures.cochat.ai/paper-figures/7f2f6c818f4b436a8c996c8a456b3f24/artificial-intelligence-to-facilitate-sep-1-measure-compliance-and-fluid-figure-1.jpg)
![Figure 2 Taxonomy of AI approaches for fluid and hemodynamic management in sepsis patients. Predictive models estimate fluid requirement or physiologic response. Prescriptive models recommend time-varying fluid and/or vasopressor strategies. Control-oriented models aim to titrate therapy to physiologic targets [61,62,63,64,65,66,67,68,69,70,73,76].](https://figures.cochat.ai/paper-figures/189dde783979433c9a5e1d7da1383f75/artificial-intelligence-to-facilitate-sep-1-measure-compliance-and-fluid-figure-2.jpg)
![Figure 3 AI sepsis tool implementation pathway with workflow phases (planning, implementation, and sustainment) across socio-technical domains (data, clinical workflow, governance, and training/operations). Checkpoint callouts summarize common barriers and paired mitigation strategies reported in real-world implementations and systematic reviews [10,33,34,46,48,77,78].](https://figures.cochat.ai/paper-figures/64f578b2e4354afba0f6cf5c7e859247/artificial-intelligence-to-facilitate-sep-1-measure-compliance-and-fluid-figure-3.jpg)
Verification
Claims the paper makes
64 Valid · 1 Not valid · 9 Not checked · 33 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“While previous studies examining restricted versus liberal fluid strategies had patients enrolled after an approximately 30 mL/kg of crystalloid fluid administration, fixed-volume mandates embedded in SEP-1 may conflict with individualized, physiology-driven care, particularly for patients at elevated risk of fluid-related harm [22,23].”
§ Artificial Intelligence to Facilitate SEP-1 Measure Compliance and Fluid Management in Sepsis > 4. Ongoing Fluid Resuscitation Controversies in Septic Shock
What the sources say
- Partly supportedcites [22] Meyhoff, 2022, Restriction of Intravenous Fluid in ICU Patients with Septic Shock, N. Engl. J. Med. sourceExperimental — accuracy not yet measured
The Interventions section says patients had received at least 1 liter of intravenous fluid before screening, and Table 1 reports median pre-randomization volumes of 3000–3200 ml. But the Statistical Analysis table note defines those volumes as including crystalloids, colloids, and blood products; the paper does not establish that patients were enrolled after approximately 30 ml/kg of crystalloid specifically. (LLM, conf=0.86)
- Partly supportedcites [23] Shapiro, 2023, Early Restrictive or Liberal Fluid Management for Sepsis-Induced Hypotension, N. Engl. J. Med. sourceExperimental — accuracy not yet measured
The Methods describe enrollment after initial treatment with 1–3 liters of intravenous fluid, and Characteristics of the Patients reports a median of 2050 ml before randomization in each group (page not available in our copy). This supports enrollment after fluids but does not establish the claimed approximately 30 ml/kg crystalloid dose. (LLM, conf=0.82)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Additionally, some mandated elements lack strong outcomes evidence [7,8].”
§ Artificial Intelligence to Facilitate SEP-1 Measure Compliance and Fluid Management in Sepsis > 3. Current State of SEP-1 Sepsis Measure Compliance and Implementation Barriers
What the sources say
- Supportedcites [7] Wang, 2020, Driving blind: Instituting SEP-1 without high quality outcomes data, J. Thorac. Dis. source
The Abstract says systematic reviews found only low-quality evidence supporting most SEP-1 interventions. The Introduction and “Basis for potential harm with SEP-1” add that high-quality outcomes data are lacking for its mandated interventions. (LLM, conf=0.94)
- Could not verifycites [8] Rhee, 2018, Variability in determining sepsis time zero and bundle compliance rates for the centers for medicare and medicaid services SEP-1 measure, Infect. Control Hosp. Epidemiol. sourceExperimental — accuracy not yet measured
The Introduction describes SEP-1’s required bundles, and the Discussion addresses variability in determining compliance. Neither section assesses whether individual mandated elements have strong outcomes evidence; searches for that claim and for evidence of outcome benefits found no such passage in our copy. (LLM, conf=0.77)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“fixed-volume mandates embedded in SEP-1 may conflict with individualized, physiology-driven care, particularly for patients at elevated risk of fluid-related harm [22,23].”
§ Artificial Intelligence to Facilitate SEP-1 Measure Compliance and Fluid Management in Sepsis > 4. Ongoing Fluid Resuscitation Controversies in Septic Shock
What the sources say
- Partly supportedcites [22] Meyhoff, 2022, Restriction of Intravenous Fluid in ICU Patients with Septic Shock, N. Engl. J. Med. sourceExperimental — accuracy not yet measured
The paper’s opening passage says Surviving Sepsis Campaign guidelines suggest an initial fixed volume of 30 ml/kg despite low-certainty evidence, and it reports associations between higher fluid volumes and harm. It does not discuss SEP-1, characterize its requirements as mandates, or assess whether they conflict with individualized care. The Discussion reports no significant difference in mortality or serious adverse events between the trial’s restrictive- and standard-fluid groups. (LLM, conf=0.83)
- Partly supportedcites [23] Shapiro, 2023, Early Restrictive or Liberal Fluid Management for Sepsis-Induced Hypotension, N. Engl. J. Med. sourceExperimental — accuracy not yet measured
The Trial Procedures and protocol diagram (p. 502) allow clinicians to adjust fluids based on a patient's condition and to halt fluids when overload appears. Safety Outcomes reports fewer fluid-overload and pulmonary-edema events in the restrictive group. The paper does not discuss SEP-1 or establish that its fixed-volume mandates conflict with individualized care, including for patients at elevated risk of fluid-related harm. (LLM, conf=0.84)
hedgedstates a possible cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Systematic reviews and meta-analyses additionally found no evidence that fluid restriction or fluid de-resuscitation is superior to higher fluid volumes for mortality in septic shock, underscoring uncertainty about optimal cumulative fluid balance strategies [24,25,26].”
§ Artificial Intelligence to Facilitate SEP-1 Measure Compliance and Fluid Management in Sepsis > 4. Ongoing Fluid Resuscitation Controversies in Septic Shock
What the sources say
- Not checked — no full text of the sourcecites [24] Messmer, 2023, Active fluid de-resuscitation in critically ill patients with septic shock: A systematic review and meta-analysis, Eur. J. Intern. Med. source
- Partly supportedcites [25] Sivapalan, 2023, Lower vs Higher Fluid Volumes in Adult Patients with Sepsis: An Updated Systematic Review with Meta-Analysis and Trial Sequential Analysis, Chest sourceExperimental — accuracy not yet measured
The Results > Primary Outcomes (page not specified) report no significant mortality advantage for lower versus higher IV fluid volumes, and the Discussion > Strengths and Limitations says the septic-shock subgroup was consistent with the primary analysis. However, the review covers adults with sepsis as well as septic shock and does not establish a separate mortality finding for fluid de-resuscitation in septic shock. (LLM, conf=0.84)
- Partly supportedcites [26] Reynolds, 2023, Restrictive resuscitation in patients with sepsis and mortality: A systematic review and meta-analysis with trial sequential analysis, Pharmacotherapy sourceExperimental — accuracy not yet measured
The Results report no significant mortality reduction with fluid restriction versus usual care in patients with sepsis. The Discussion says the studies did not focus on de-resuscitation, so this paper does not establish the de-resuscitation part of the claim; the reported mortality result is also for sepsis patients, not specifically a septic-shock subgroup. (LLM, conf=0.87)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Every source the paper cites for this claim was checked and supports it.
In the paper
“Clinicians may therefore experience SEP-1 as a compliance exercise rather than a patient-centered framework [7].”
§ Artificial Intelligence to Facilitate SEP-1 Measure Compliance and Fluid Management in Sepsis > 3. Current State of SEP-1 Sepsis Measure Compliance and Implementation Barriers
What the sources say
- Supportedcites [7] Wang, 2020, Driving blind: Instituting SEP-1 without high quality outcomes data, J. Thorac. Dis. source
The Introduction (page not specified in our copy) reports that SEP-1’s reporting requirements may force clinicians to change their practice, consume substantial staff time, and leave insufficient room to adjust care to individual patients. That supports the citing sentence’s hedged characterization, although the paper does not use the phrase “compliance exercise.” (LLM, conf=0.85)
hedged- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
45 more claims · 24 with their cited sources checked
Create a free account to see all 107 claims and to verify the 1 reference still open.
- It may state two separate claims as one. 46
- Probably not a substantial claim. 16
Citations
Cites
The third international consensus definitions for sepsis and septic shock (Sepsis-3).
Singer, M., Deutschman, C. S., Seymour, C. W., Shankar-Hari, M., Annane, D., Bauer, M., Bellomo, R., Bernard, G. R., Chiche, J.-D., Coopersmith, C. M., et al. (2016) · JAMA · doi:10.1001/jama.2016.0287 · Cited by 28,250 · cited 1×
“Sepsis is a life-threatening organ dysfunction caused by a dysregulated host response to infection, requiring early recognition and coordinated intervention [1,2].”
Early goal-directed therapy in the treatment of severe sepsis and septic shock
Rivers, E., Nguyen, B., Havstad, S., Ressler, J., Muzzin, A., Knoblich, B., Peterson, E., & Tomlanovich, M. (2001) · New England Journal of Medicine · doi:10.1056/nejmoa010307 · Cited by 10,743
Surviving Sepsis Campaign guidelines for management of severe sepsis and septic shock
Dellinger, R. P., Carlet, J. M., Masur, H., Gerlach, H., Calandra, T., Cohen, J., Gea-Banacloche, J., Keh, D., Marshall, J. C., Parker, M. M., et al. (2004) · Critical Care Medicine · doi:10.1097/01.ccm.0000117317.18092.e4 · Cited by 3,018 · cited 1×
“Early guidelines were strongly influenced by early goal-directed therapy which aimed to formalize time-sensitive resuscitation targets [4,5].”
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 · cited 3×
“Komorowski et al. developed and validated a reinforcement learning-based clinical decision support model (“AI Clinician”) to optimize fluid and vasopressor therapy in sepsis patients [62].”
Levy, M. M., Dellinger, R. P., Townsend, S. R., Linde-Zwirble, W. T., Marshall, J. C., Bion, J., Schorr, C., Artigas, A., Ramsay, G., Beale, R., et al. (2010) · Intensive Care Medicine · doi:10.1007/s00134-009-1738-3 · Cited by 868
An interpretable machine learning model for accurate prediction of sepsis in the ICU
Nemati, S., Holder, A., Razmi, F., Stanley, M. D., Clifford, G. D., & Buchman, T. G. (2018) · Critical Care Medicine · doi:10.1097/ccm.0000000000002936 · Cited by 832 · cited 1×
“A deep learning approach using recurrent neural networks predicted sepsis onset hours in advance within ICU cohorts, demonstrating that model-driven alerts could theoretically increase lead time for intervention [50].”
Restriction of Intravenous Fluid in ICU Patients with Septic Shock
Meyhoff, T. S., Hjortrup, P. B., Wetterslev, J., Sivapalan, P., Laake, J. H., Cronhjort, M., Jakob, S. M., Cecconi, M., Nalos, M., Ostermann, M., et al. (2022) · New England Journal of Medicine · doi:10.1056/nejmoa2202707 · Cited by 427 · cited 4×
“…versus liberal fluid strategies had patients enrolled after an approximately 30 mL/kg of crystalloid fluid administration, fixed-volume mandates embedded in SEP-1 may conflict with individualized, physiology-driven care, particularly for patients at elevated risk of fluid-related harm [22,23].”
InfluentialEarly Restrictive or Liberal Fluid Management for Sepsis-Induced Hypotension
The National Heart, L. (2023) · New England Journal of Medicine · doi:10.1056/nejmoa2212663 · Cited by 427 · cited 4×
“…versus liberal fluid strategies had patients enrolled after an approximately 30 mL/kg of crystalloid fluid administration, fixed-volume mandates embedded in SEP-1 may conflict with individualized, physiology-driven care, particularly for patients at elevated risk of fluid-related harm [22,23].”
InfluentialGoh, K. H., Wang, L., Yeow, A. Y. K., Poh, H., Li, K., Yeow, J. J. L., & Tan, G. Y. H. (2021) · Nature Communications · doi:10.1038/s41467-021-20910-4 · Cited by 415 · cited 1×
“The model achieved high predictive performance (AUROC up to 0.94) for both diagnosis and early prediction at 12 h before onset, outperforming physicians, and traditional scoring systems, increasing early detection by 32% while reducing false positives up to 17% [47].”
InfluentialAdams, R., Henry, K. E., Sridharan, A., Soleimani, H., Zhan, A., Rawat, N., Johnson, L., Hager, D. N., Cosgrove, S. E., Markowski, A., et al. (2022) · Nature Medicine · doi:10.1038/s41591-022-01894-0 · Cited by 365 · cited 1×
“A few studies have shown that AI sepsis prediction models were associated with decreased patient mortality; however, they were observational cohorts or before-and-after designs [55–57].”
A Systematic Review of the Barriers to the Implementation of Artificial Intelligence in Healthcare
Ahmed, M. I., Spooner, B., Isherwood, J., Lane, M., Orrock, E., & Dennison, A. (2023) · Cureus · doi:10.7759/cureus.46454 · Cited by 364
Sendak, M. P., Ratliff, W., Sarro, D., Alderton, E., Futoma, J., Gao, M., Nichols, M., Revoir, M., Yashar, F., Miller, C., et al. (2020) · JMIR Medical Informatics · doi:10.2196/15182 · Cited by 287 · cited 9×
“Implementation science frameworks can help specify who receives alerts, what actions are expected, and what resources are required to respond, thereby reducing alert fatigue and ensuring accountability [10,77].”
Influentialvan de Sande, D., van Genderen, M. E., Huiskens, J., Gommers, D., & van Bommel, J. (2021) · Intensive Care Medicine · doi:10.1007/s00134-021-06446-7 · Cited by 279 · cited 1×
“Systematic reviews show that most ICU AI studies remain retrospective, with limited external validation and scarce evidence on patient-centered outcomes after real-world deployment [35].”
Jabbour, S., Fouhey, D., Shepard, S., Valley, T. S., Kazerooni, E. A., Banovic, N., Wiens, J., & Sjoding, M. W. (2023) · JAMA · doi:10.1001/jama.2023.22295 · Cited by 236
Automation Bias and Assistive AI: Risk of Harm From AI-Driven Clinical Decision Support
Khera, R., Simon, M. A., & Ross, J. S. (2023) · JAMA · doi:10.1001/jama.2023.22557 · Cited by 222 · cited 1×
“In a randomized vignette study, clinicians exposed to a systematically biased model had lower diagnostic accuracy, and explainability features did not fully mitigate this effect [81,82].”
Fluid Therapy for Critically Ill Adults With Sepsis: A Review
Zampieri, F. G., Bagshaw, S. M., & Semler, M. W. (2023) · JAMA · doi:10.1001/jama.2023.7560 · Cited by 184 · cited 2×
“Contemporary recommendations emphasize early hemodynamic assessment, and ongoing reassessment to avoid fluid overload after stabilization [27].”
Nair, M., Svedberg, P., Larsson, I., & Nygren, J. M. (2024) · PLOS ONE · doi:10.1371/journal.pone.0305949 · Cited by 179 · cited 5×
“Finally, post-deployment evaluation should be ongoing, including monitoring model drift, recalibrating thresholds, auditing equity, and ensuring that model updates are governed in ways that preserve trust and clinical accountability [34,78].”
InfluentialSepsis: key insights, future directions, and immediate goals. A review and expert opinion
Martin-Loeches, I., Singer, M., & Leone, M. (2024) · Intensive Care Medicine · doi:10.1007/s00134-024-07694-z · Cited by 178 · cited 2×
“Clinical signs are nonspecific, and commonly used screening tools (e.g., SIRS, qSOFA, SOFA) have limitations for early identification, especially outside the intensive care unit (ICU) and in patients with chronic organ dysfunction [9].”
Impact of a deep learning sepsis prediction model on quality of care and survival
Boussina, A., Shashikumar, S. P., Malhotra, A., Owens, R. L., El-Kareh, R., Longhurst, C. A., Quintero, K., Donahue, A., Chan, T. C., Nemati, S., et al. (2024) · npj Digital Medicine · doi:10.1038/s41746-023-00986-6 · Cited by 170
Integrating a Machine Learning System Into Clinical Workflows: Qualitative Study
Sandhu, S., Lin, A. L., Brajer, N., Sperling, J., Ratliff, W., Bedoya, A. D., Balu, S., O'Brien, C., & Sendak, M. P. (2020) · Journal of Medical Internet Research · doi:10.2196/22421 · Cited by 168
Prescott, H. C., Antonelli, M. M., Alhazzani, W. A., Møller, M. H., Alshamsi, F. E., Azevedo, L. C. P., Belley‐Côté, E. P., Waele, J. D., Derde, L. P. G., Dionne, J. C., et al. (2026) · Critical Care Medicine · doi:10.1097/ccm.0000000000007089 · Cited by 168
Use of artificial intelligence in critical care: opportunities and obstacles
Pinsky, M. R., Bedoya, A., Bihorac, A., Celi, L., Churpek, M., Economou-Zavlanos, N. J., Elbers, P., Saria, S., Liu, V., Lyons, P. G., et al. (2024) · Critical Care · doi:10.1186/s13054-024-04860-z · Cited by 160
Human Factors Challenges for the Safe Use of Artificial Intelligence in Patient Care
Sujan, M., Furniss, D., Grundy, K., Grundy, H., Nelson, D., Elliott, M., White, S., Habli, I., & Reynolds, N. (2019) · BMJ Health & Care Informatics · doi:10.1136/bmjhci-2019-100081 · Cited by 144 · cited 2×
“Recent guidance and human-factors literature emphasize that safe clinician–AI interaction requires human oversight, calibrated trust, awareness of model limitations, and deliberate verification of AI outputs against the patient’s examination, physiology, and clinical trajectory [79,80].”
Clinical applications of artificial intelligence in sepsis: A narrative review
Schinkel, M., Paranjape, K., Nannan Panday, R. S., Skyttberg, N., & Nanayakkara, P. W. B. (2019) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2019.103488 · Cited by 130 · cited 1×
“Sepsis AI tools are sensitive to data labeling choices, confounded by treatment indications, and changes in practice over time that can degrade accuracy and transportability [71,74].”
Li, F., Wang, S., Gao, Z., Qing, M., Pan, S. L., Liu, Y., & Hu, C. (2025) · Frontiers in Medicine · doi:10.3389/fmed.2024.1510792 · Cited by 119 · cited 2×
“Combined bedside ultrasound features, laboratory, and clinician documentation have also been examined to predict tolerance of additional fluids and risk of fluid overload [14,42].”
Fluid resuscitation in sepsis: the great 30 mL per kg hoax
Marik, P. E., Byrne, L., & van Haren, F. (2020) · Journal of Thoracic Disease · doi:10.21037/jtd.2019.12.84 · Cited by 116 · cited 1×
“Critics argue that septic shock is not uniformly a hypovolemic state and that indiscriminate fluid loading may exacerbate vasoplegia, myocardial dysfunction, endothelial injury, and tissue edema, thereby impairing oxygen delivery despite increased intravascular volume [20].”
Lower vs Higher Fluid Volumes During Initial Management of Sepsis
Meyhoff, T. S., Møller, M. H., Hjortrup, P. B., Cronhjort, M., Perner, A., & Wetterslev, J. (2020) · Chest · doi:10.1016/j.chest.2019.11.050 · Cited by 107
Dellinger, R. P., Rhodes, A., Evans, L., Alhazzani, W. A., Beale, R. J., Jaeschke, R., Machado, F. R., Masur, H., Osborn, T. M., Parker, M. M., et al. (2023) · Critical Care Medicine · doi:10.1097/ccm.0000000000005804 · Cited by 105
Knop, M., Weber, S., Mueller, M., & Niehaves, B. (2022) · JMIR Human Factors · doi:10.2196/28639 · Cited by 103 · cited 2×
“Recent guidance and human-factors literature emphasize that safe clinician–AI interaction requires human oversight, calibrated trust, awareness of model limitations, and deliberate verification of AI outputs against the patient’s examination, physiology, and clinical trajectory [79,80].”
Moazemi, S., Vahdati, S., Li, J., Kalkhoff, S., Castano, L. J. V., Dewitz, B., Bibo, R., Sabouniaghdam, P., Tootooni, M. S., Bundschuh, R. A., et al. (2023) · Frontiers in Medicine · doi:10.3389/fmed.2023.1109411 · Cited by 97 · cited 1×
“Results showed generally strong predictive performance (AU-ROC ~0.79–0.96), but highlighted major limitations including lack of external validation ( ≈ 75% of studies), poor generalizability, and limited interpretability [32].”
InfluentialNational Performance on the Medicare SEP-1 Sepsis Quality Measure
Barbash, I. J., Davis, B., & Kahn, J. M. (2019) · Critical Care Medicine · doi:10.1097/ccm.0000000000003613 · Cited by 86 · cited 3×
“However, individual elements can vary in feasibility obscuring true performance [15,16].”
Artificial Intelligence for Clinical Decision Support in Sepsis
Wu, M., Du, X., Gu, R., & Wei, J. (2021) · Frontiers in Medicine · doi:10.3389/fmed.2021.665464 · Cited by 84
Artificial Intelligence in Intensive Care Medicine: Bibliometric Analysis
Tang, R., Zhang, S., Ding, C., Zhu, M., & Gao, Y. (2022) · Journal of Medical Internet Research · doi:10.2196/42185 · Cited by 79 · cited 2×
“AI is being explored not only for early sepsis prediction but also to support implementation—routing the right information to the right team at the right time, reducing cognitive load, and promoting consistent, evidence-based care while preserving clinician oversight [10,11].”
Artificial Intelligence in Sepsis Management: An Overview for Clinicians
Bignami, E. G., Berdini, M., Panizzi, M., Domenichetti, T., Bezzi, F., Allai, S., Damiano, T., & Bellini, V. (2025) · Journal of Clinical Medicine · doi:10.3390/jcm14010286 · Cited by 79 · cited 1×
“We refer the reader to other resources on the use of AI in prediction, prognostication, and overall management of sepsis [12–14].”
Artificial intelligence in critical illness and its impact on patient care: a comprehensive review
Saqib, M., Iftikhar, M., Neha, F., Karishma, F., & Mumtaz, H. (2023) · Frontiers in Medicine · doi:10.3389/fmed.2023.1176192 · Cited by 77
Hong, N., Liu, C., Gao, J., Han, L., Chang, F., Gong, M., & Su, L. (2022) · JMIR Medical Informatics · doi:10.2196/28781 · Cited by 76 · cited 1×
“Across ICU applications, predictive analytics dominate, particularly for the early detection of deterioration and outcome prognostication [31].”
Gallifant, J., Zhang, J., del Pilar Arias Lopez, M., Zhu, T., Camporota, L., Celi, L. A., & Formenti, F. (2022) · British Journal of Anaesthesia · doi:10.1016/j.bja.2021.09.025 · Cited by 72 · cited 1×
“Reporting and reproducibility challenges were also observed, including incomplete transparency in model development, inadequate handling of missing data, and insufficient attention to calibration and post-deployment performance monitoring [36].”
van der Vegt, A. H., Scott, I. A., Dermawan, K., Schnetler, R. J., Kalke, V. R., & Lane, P. J. (2023) · Journal of the American Medical Informatics Association · doi:10.1093/jamia/ocad075 · Cited by 72 · cited 6×
“At the same time, studies must explicitly monitor adverse effects, such as unnecessary antibiotic use, excessive or inadequate fluid administration, increased ICU utilization, and resource strain [10,46].”
InfluentialFDA-Authorized AI/ML Tool for Sepsis Prediction: Development and Validation
Bhargava, A., López-Espina, C., Schmalz, L., Khan, S., Watson, G. L., Urdiales, D., Updike, L., Kurtzman, N., Dagan, A., Doodlesack, A., et al. (2024) · NEJM AI · doi:10.1056/aioa2400867 · Cited by 72
Townsend, S. R., Phillips, G. S., Duseja, R., Tefera, L., Cruikshank, D., Dickerson, R., Nguyen, H. B., Schorr, C. A., Levy, M. M., Dellinger, R. P., et al. (2022) · CHEST · doi:10.1016/j.chest.2021.07.2167 · Cited by 69 · cited 4×
“Each element is tied to a strict time window [7,19].”
InfluentialSelf-citationVenkatesh, A. K., Slesinger, T., Whittle, J., Osborn, T., Aaronson, E., Rothenberg, C., Tarrant, N., Goyal, P., Yealy, D. M., & Schuur, J. D. (2018) · Annals of Emergency Medicine · doi:10.1016/j.annemergmed.2017.06.032 · Cited by 58 · cited 2×
“However, individual elements can vary in feasibility obscuring true performance [15,16].”
Clinical Sepsis Phenotypes in Critically Ill Patients
Παπαθανάκος, Γ., Andrianopoulos, I., Xenikakis, M., Papathanasiou, A. I., Koulenti, D., Blot, S. I., & Koulouras, V. P. (2023) · Microorganisms · doi:10.3390/microorganisms11092165 · Cited by 57 · cited 3×
“Phenotyping may identify subgroups with differing fluid tolerance and responsiveness, enabling “right patient, right fluid” strategies that reconcile quality measures with individualized care [72].”
Bataille, B., de Selle, J., Moussot, P.-E., Marty, P., Silva, S., & Cocquet, P. (2021) · British Journal of Anaesthesia · doi:10.1016/j.bja.2020.11.039 · Cited by 56 · cited 1×
“In a prospective observational study, Bataille et al. aimed to evaluate whether machine learning models could improve prediction of fluid responsiveness in patients with severe sepsis or septic shock in a single-center ICU [69].”
O'Reilly, D., McGrath, J., & Martin-Loeches, I. (2024) · Journal of Intensive Medicine · doi:10.1016/j.jointm.2023.10.001 · Cited by 55
Artificial intelligence in critical care nursing: A scoping review
Park, Y., Chang, S. J., & Kim, E. (2025) · Australian Critical Care · doi:10.1016/j.aucc.2025.101225 · Cited by 54
The application of artificial intelligence in the management of sepsis
Yang, J., Hao, S., Huang, J., Chen, T., Liu, R., Zhang, P., Feng, M., He, Y., Xiao, W., Hong, Y., et al. (2023) · Medical Review · doi:10.1515/mr-2023-0039 · Cited by 53 · cited 3×
“We refer the reader to other resources on the use of AI in prediction, prognostication, and overall management of sepsis [12–14].”
Algor-ethics: charting the ethical path for AI in critical care
Montomoli, J., Bitondo, M. M., Cascella, M., Rezoagli, E., Romeo, L., Bellini, V., Semeraro, F. F., Gamberini, E., Frontoni, E., Agnoletti, V., et al. (2024) · Journal of Clinical Monitoring and Computing · doi:10.1007/s10877-024-01157-y · Cited by 53
Celi, L. A., Hinske Christian, L., Alterovitz, G., & Szolovits, P. (2008) · Critical Care · doi:10.1186/cc7140 · Cited by 52 · cited 1×
“Celi et al. performed a proof-of-concept study aimed to evaluate the feasibility of using AI to predict individualized fluid requirements in ICU patients by analyzing data from the MIMIC II database at a single academic center, including 3014 patients on vasopressors [61].”
Artificial intelligence to advance acute and intensive care medicine
Biesheuvel, L. A., Dongelmans, D. A., & Elbers, P. (2024) · Current Opinion in Critical Care · doi:10.1097/mcc.0000000000001150 · Cited by 52
Rhee, C., Brown, S. R., Jones, T. M., O’Brien, C., Pande, A., Hamad, Y., Bulger, A. L., Tobin, K. A., Massaro, A. F., Anderson, D. J., et al. (2018) · Infection Control & Hospital Epidemiology · doi:10.1017/ice.2018.134 · Cited by 51 · cited 5×
“…targeted workflow redesign— standardized triage, early order sets, and escalation pathways—while acknowledging that strict SEP-1 abstraction may not fully capture clinical judgment in complex cases, particularly when fluid dosing or antibiotic timing decisions need to be individualized [8,19].”
InfluentialJoshi, M., Mecklai, K., Rozenblum, R., & Samal, L. (2022) · JAMIA Open · doi:10.1093/jamiaopen/ooac022 · Cited by 46 · cited 3×
“Implementation barriers for sepsis AI cluster around trust and interpretability, work-flow integration, technical readiness and generalizability, governance and safety, leadership alignment, and the resources required for sustainment (Table 4) [10,33,34,46,48,77,78].”
Lower vs Higher Fluid Volumes in Adult Patients With Sepsis
Sivapalan, P., Ellekjaer, K. L., Jessen, M. K., Meyhoff, T. S., Cronhjort, M., Hjortrup, P. B., Wetterslev, J., Granholm, A., Møller, M. H., & Perner, A. (2023) · CHEST · doi:10.1016/j.chest.2023.04.036 · Cited by 45
Evolving Management Practices for Early Sepsis-induced Hypoperfusion: A Narrative Review
Munroe, E. S., Hyzy, R. C., Semler, M. W., Shankar-Hari, M., Young, P. J., Zampieri, F. G., & Prescott, H. C. (2023) · American Journal of Respiratory and Critical Care Medicine · doi:10.1164/rccm.202209-1831ci · Cited by 40 · cited 1×
“Practical bedside frameworks highlight dynamic assessment of fluid responsiveness and tolerance; e.g., passive leg raise with stroke volume response, echocardiographic surrogates of preload responsiveness, and careful monitoring of venous congestion and oxygenation [28].”
Kijpaisalratana, N., Saoraya, J., Nhuboonkaew, P., Vongkulbhisan, K., & Musikatavorn, K. (2024) · Internal and Emergency Medicine · doi:10.1007/s11739-024-03535-5 · Cited by 40
Messmer, A. S., Dill, T., Müller, M., & Pfortmueller, C. A. (2023) · European Journal of Internal Medicine · doi:10.1016/j.ejim.2023.01.009 · Cited by 35 · cited 1×
“Systematic reviews and meta-analyses additionally found no evidence that fluid restriction or fluid de-resuscitation is superior to higher fluid volumes for mortality in septic shock, underscoring uncertainty about optimal cumulative fluid balance strategies [24–26].”
Driving blind: Instituting SEP-1 without high quality outcomes data
Wang, J., Strich, J. R., Applefeld, W. N., Sun, J., Cui, X., Natanson, C., & Eichacker, P. Q. (2020) · Journal of Thoracic Disease · doi:10.21037/jtd.2019.12.100 · Cited by 34 · cited 6×
“While some studies associate compliance with improved outcomes, measurement variability and limited evidence for certain mandated elements remain central concerns [7,8,15,19].”
InfluentialA machine learning approach for predicting urine output after fluid administration
Lin, P.-C., Huang, H.-C., Komorowski, M., Lin, W.-K., Chang, C.-M., Chen, K.-T., Li, Y., & Lin, M. (2019) · Computer Methods and Programs in Biomedicine · doi:10.1016/j.cmpb.2019.05.009 · Cited by 30
Aaronson, E. L., Filbin, M. R., Brown, D. F. M., Tobin, K., & Mort, E. A. (2017) · The Journal of Emergency Medicine · doi:10.1016/j.jemermed.2016.08.009 · Cited by 26 · cited 1×
“From the clinician perspective, strict requirements may incentivize unnecessary broad-spectrum antibiotics or fixed-volume fluid administration in patients with tenuous cardiopulmonary reserve, shifting attention from physiology-driven resuscitation to documentation-driven completion [7,18].”
Reynolds, P. M., Stefanos, S., & MacLaren, R. (2023) · Pharmacotherapy: The Journal of Human Pharmacology and Drug Therapy · doi:10.1002/phar.2764 · Cited by 21 · cited 1×
“Systematic reviews and meta-analyses additionally found no evidence that fluid restriction or fluid de-resuscitation is superior to higher fluid volumes for mortality in septic shock, underscoring uncertainty about optimal cumulative fluid balance strategies [24–26].”
Valan, B., Prakash, A., Ratliff, W., Gao, M., Muthya, S., Thomas, A., Eaton, J. L., Gardner, M., Nichols, M., Revoir, M., et al. (2025) · npj Digital Medicine · doi:10.1038/s41746-025-01664-5 · Cited by 21
Enhancing sepsis management through machine learning techniques: A review
Ocampo-Quintero, N., Vidal-Cortés, P., del Río Carbajo, L., Fdez-Riverola, F., Reboiro-Jato, M., & Glez-Peña, D. (2022) · Medicina Intensiva · doi:10.1016/j.medin.2020.04.003 · Cited by 20 · cited 1×
“Implementation therefore must include guardrails, accountability of alert response, and balancing outcomes [10,53].”
Cui, X., Chang, Y., Yang, C., Cong, Z., Wang, B., & Leng, Y. (2022) · Journal of Personalized Medicine · doi:10.3390/jpm13010050 · Cited by 19 · cited 1×
“AI research on critical care has expanded rapidly, with literature searches demonstrating accelerated growth in publications and applications over the last decade [11,30].”
Su, L., Li, Y., Liu, S., Zhang, S., Zhou, X. Q., Weng, L., Su, M., Du, B., Zhu, W., & Long, Y. (2022) · Frontiers in Medicine · doi:10.3389/fmed.2022.766447 · Cited by 18
Mollura, M., Drudi, C., Lehman, L.-W. H., & Barbieri, R. (2022) · Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) · doi:10.1109/embc48229.2022.9871055 · Cited by 15 · cited 1×
“Accordingly, prospective evaluation should prioritize safety, calibration to local practice, and physiologic plausibility [62,63,75].”
Role of Artificial Intelligence in Critical Care Medicine: A Literature Review
ABDELBAKY, A., Elmasry, W. G., Awad, A. H., & Khan, S. (2025) · Cureus · doi:10.7759/cureus.90149 · Cited by 14
Optimal Sepsis Patient Treatment using Human-in-the-loop Artificial Intelligence
Gupta, A., Lash, M. T., & Nachimuthu, S. K. (2021) · Expert Systems with Applications · doi:10.1016/j.eswa.2020.114476 · Cited by 13 · cited 2×
“Accordingly, prospective evaluation should prioritize safety, calibration to local practice, and physiologic plausibility [62,63,75].”
Warstadt, N. M., Caldwell, J. R., Tang, N., Mandola, S., Jamin, C., & Dahn, C. (2022) · BMJ Open Quality · doi:10.1136/bmjoq-2021-001624 · Cited by 12 · cited 1×
“For example, approaches that combine gradient boosting with feature attribution or rule extraction aim to increase interpretability so that clinicians can understand what is driving a high-risk alert [51,52].”
Liang, W., & Jia, J. (2025) · Computer Methods and Programs in Biomedicine · doi:10.1016/j.cmpb.2025.108754 · Cited by 12
Harnessing AI in critical care: opportunities, challenges and key steps for success
Kalimouttou, A., Stevens, R. D., & Pirracchio, R. (2026) · Thorax · doi:10.1136/thorax-2024-222125 · Cited by 12
Fixler, A., Oliaro, B., Frieden, M., Girardo, C., Winterbottom, F. A., Fort, L. B., & Hill, J. (2023) · Ochsner Journal · doi:10.31486/toj.22.0098 · Cited by 11 · cited 1×
“For example, approaches that combine gradient boosting with feature attribution or rule extraction aim to increase interpretability so that clinicians can understand what is driving a high-risk alert [51,52].”
Implementation of an Automated Sepsis Screening Tool in a Community Hospital Setting
Cooper, P. B., Hughes, B. J., Verghese, G. M., Just, J. S., & Markham, A. J. (2021) · Journal of Nursing Care Quality · doi:10.1097/ncq.0000000000000501 · Cited by 10
Learning to Treat Hypotensive Episodes in Sepsis Patients Using a Counterfactual Reasoning FrameworkPreprint
Jeter, R., Lehman, L.-W. H., Josef, C. S., Shashikumar, S. P., & Nemati, S. (2021) · medRxiv · doi:10.1101/2021.03.03.21252863 · Cited by 10
Fluids or vasopressors for the initial resuscitation of septic shock
Macdonald, S., Peake, S. L., Corfield, A. R., & Delaney, A. (2022) · Frontiers in Medicine · doi:10.3389/fmed.2022.1069782 · Cited by 6 · cited 1×
“In a clinical trial completed in Africa, aggressive fluid strategies for septic patients were associated with worsening respiratory status and mortality, underscoring that one-size-fits-all approach should not be standard practice [21].”
Personalized Fluid Management in Patients with Sepsis and AKI: A Casual Machine Learning ApproachPreprint
Oh, W., Takkavatakarn, K., Al-taie, Z., Kittrell, H. D., Shawwa, K., Gómez, H., Sawant, A. S., Tandon, P., Kumar, G., Sterling, M., et al. (2024) · medRxiv · doi:10.1101/2024.08.06.24311556 · Cited by 5
Gaieski, D. F., Carr, B., Toolan, M., Ciotti, K., Kidane, A., Flaada, D., Christina, J., & Aggarwal, R. (2022) · Critical Care Explorations · doi:10.1097/cce.0000000000000767 · Cited by 4
Organizational Factors Associated with Sepsis Bundle Compliance: A Nationwide Study
Peach, B. C., & Ng, B. P. (2020) · Health Services Research · doi:10.1111/1475-6773.13444 · Cited by 3 · cited 1×
“A nationwide study linked lower odds of SEP-1 compliance to larger hospital size and teaching status, suggesting that case-mix complexity, competing priorities, and operational load may interact with performance [17].”
Catling, F. J., Nagendran, M., Festor, P., Bien, Z., Harris, S., Faisal, A. A., Gordon, A. C., & Komorowski, M. (2023) · doi:10.1101/2023.08.26.23294666 · Cited by 3 · cited 2×
“RL and counterfactual evaluations rely on assumptions embedded in off-policy estimators and may not generalize across settings or evolving practice patterns [62,71].”
Estiri, E., & Mirinejad, H. (2024) · 2024 46th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) · doi:10.1109/embc53108.2024.10781780 · Cited by 3 · cited 1×
“An automated fluid resuscitation framework using a variational autoencoder (VAE)-based nonlinear state-space model combined with model predictive control (MPC) was developed by Estiri et al. to optimize hemodynamic responses [73].”
Grooms, E., Biesack, K., Abban, B., & Kramer, J. (2025) · Journal for Healthcare Quality · doi:10.1097/jhq.0000000000000493 · Cited by 1 · cited 1×
“A few studies have shown that AI sepsis prediction models were associated with decreased patient mortality; however, they were observational cohorts or before-and-after designs [55–57].”
Wells, A. J., Sapp, A., Walker, D. K., & Baker, K. A. (2025) · Critical Care Explorations · doi:10.1097/cce.0000000000001215 · Cited by 1
Research works printed but not linked: 1
- Kamaleswaran, 2020, Predicting Volume Responsiveness Among Sepsis Patients Using Clinical Data and Continuous Physiological Waveforms, AMIA Annu. Symp. Proc., 2020, 619
Metadata
- Authors
- Nguyen, H. B., Krishtopaytis, E., Lopez, E., Farnoudi, N., Van, T., Kharalampova, V., & Coz Yataco, A.
- Affiliations
- Loma Linda University (US)
- Division of Critical Care, Respiratory Institute, Cleveland Clinic, Cleveland, OH 44195, USA
- Corresponding author
- Nguyen, H. Bryant
- Venue
- Journal of Clinical Medicine
- Publisher
- MDPI AG
- Published
- 2026-05-01
- Volume
- 15
- Issue
- 9
- Pages
- 3477
- ISSN
- 2077-0383 (ISSN-L)
- PMC
- PMC13163497
- OpenAlex
- W7160093997
- Cited by
- 0
- References
- 76
- FWCI
- 0.00
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Journal of Clinical Medicine (publishedVersion, cc-by)
- PubMed (publishedVersion)
- PubMed Central (publishedVersion, cc-by)
- Topics
- Health Sciences › Medicine › Critical Care and Intensive Care Medicine › Sepsis Diagnosis and Treatment
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning in Healthcare
- Keywords
- sepsis, artificial intelligence, fluid resuscitation, fluid management, early warning systems, natural language processing, telemedicine, reinforcement learning, causal machine learning, timing of antibiotic administration, model drift
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.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
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
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.