AI-Driven Predictions of Readmission and Mortality for Improved Discharge Decisions in Critical Care: A Retrospective Study
Open accessThe DOI matches its Crossref record; 26 of the 26 references it cites were found in a registry or an index; 7 of the 12 sources its claims cite were checked for support.
Abstract
Background/Objectives: The transition from the intensive care unit (ICU) to the hospital ward is a critical high-risk period for patients. Early ICU discharge reduces costs and frees up ICU resources but can lead to readmission or unexpected death if patients are discharged prematurely. Despite the availability of risk stratification tools such as the Stability and Workload Index for Transfer (SWIFT) score, predicting ICU readmission remains challenging and inconsistent. However, artificial intelligence (AI) and machine learning (ML) techniques have recently shown promise in improving clinical decision support systems, particularly in the ICU. This study aimed to identify the risk factors and assess the performance of AI models in predicting readmission or death within seven days of ICU discharge using the MIMIC-IV (between 2008 and 2019) and Kangwon National University Hospital (KNUH, between 1 January 2016 and 28 February 2023) databases. Methods: This retrospective cohort study utilized the MIMIC-IV database for model training and internal validation and the KNUH database for external validation. Various machine learning and deep learning models have been developed to predict ICU readmission or death within seven days of discharge. The performance of the primary model, GRU-D++, was compared to the SWIFT score. Statistical analysis focused on the area under the receiver operating characteristic curve (AUROC) data to evaluate model accuracy. Results: The GRU-D++ model outperformed the SWIFT score, achieving AUROC of 0.802 and 0.756 for internal and external validations, respectively. Both datasets demonstrated that the GRU-D++ model provided better predictive performance for ICU readmission or death within seven days than the traditional SWIFT score. Conclusions: Our findings suggest that the GRU-D++ deep learning model is a valuable tool for the early detection of patient deterioration after ICU discharge, potentially aiding the prevention of ICU readmission. This study highlights the potential of AI to improve clinical decision-making in intensive care settings.
Overview
CoChat Summary
## TL;DR
This retrospective study trained and externally validated a GRU-D++ deep learning model to predict ICU readmission or death within 7 days of discharge, using MIMIC-IV for training/internal validation and KNUH for external validation. GRU-D++ outperformed the SWIFT score and achieved AUROC 0.802 internally and 0.756 externally.
## Key findings
- The study included 50,261 ICU stays from MIMIC-IV and 4,852 ICU stays from KNUH; the outcome was readmission to the ICU or death within 7 days of ICU discharge.
- GRU-D++ achieved the best predictive performance among compared models, with AUROC 0.802 in internal validation and AUROC 0.756 in external validation.
- Compared with SWIFT, GRU-D++ performed better in both datasets: 0.82 vs 0.69 AUROC in MIMIC and 0.76 vs 0.68 AUROC in KNUH, with p < 0.001 by DeLong test.
- SHAP analysis identified BUN, GCS, and WBC as the most influential features for predicting post-discharge readmission/mortality.
## Methods
- Retrospective cohort study using MIMIC-IV v2.2 for model development/internal validation and KNUH data for external validation; included adult ICU stays with ICU length of stay > 24 h and excluded in-ICU deaths and transfers to other specialized units/OR.
- Started with 48 clinical variables, averaged daily, then selected 26 variables with <50% missingness; numerical variables were normalized to zero mean and unit variance.
- Compared GRU-D++ against nine other ML/DL models and the SWIFT score; internal validation used 10-fold cross-validation, with performance assessed mainly by AUROC (and also AUPR).
## Limitations
- The authors state that MIMIC-IV is disproportionately White (81.8%) and note that applying MIMIC-only findings to Asian populations may be limited; they addressed this with KNUH fine-tuning and external validation.
- They state that important system-level factors (e.g., ICU bed occupancy, nighttime/weekend discharge, step-down unit availability, rapid response teams) were not available in the retrospective datasets, so predictions may reflect institutional practice as well as physiology.
- They state that the composite endpoint combines ICU readmission and post-ICU death, which may represent different clinical trajectories; the text also makes evident a limitation of class imbalance (about 7% failure rate) and retrospective design with need for prospective multi-center validation.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures






Verification
Claims the paper makes
5 Valid · 4 Not valid · 3 Not checked · 71 Own results and interpretations
At least one cited source doesn’t support this claim or points to a different paper, and none confirmed it — see the verdicts below.
In the paper
“As demonstrated in the study by Rosa et al. [11], the SWIFT score often exhibits low discriminative power, yielding an AUROC of only 0.58 in prospective cohorts.”
§ AI-Driven Predictions of Readmission and Mortality for Improved Discharge Decisions in Critical Care: A Retrospective Study > 1. Introduction
What the sources say
- Disputedcites Rosa et al. [11] Rosa, R.G., Roehrig, C., Oliveira, R.P.D., Maccari, J.G., Antônio, A.C.P., Castro, P.D.S., Neto, F.L.D., Balzano, P.D.C., and Teixeira, C. (2015). Comparison of Unplanned Intensive Care Unit Readmission Scores: A Prospective Cohort Study. PLoS ONE, 10. sourceExperimental — accuracy not yet measured
The Results (Table 7) report a SWIFT AUC of 0.65 in Rosa et al.’s prospective cohort, not 0.58. The Discussion attributes the 0.58 AUC to a separate retrospective study by Kastrup et al.; Rosa et al. describe their own score’s discrimination as moderate. (LLM, conf=0.90)
Checked by a reasoning model — may be wrong.
boosted- 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.
- Disputed
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Nevertheless, although the SWIFT criterion is the only validated risk score for ICU discharge, it performs poorly and inconsistently upon implementation [10,11].”
§ AI-Driven Predictions of Readmission and Mortality for Improved Discharge Decisions in Critical Care: A Retrospective Study > 1. Introduction
What the sources say
- Not checked yetcites [10] Shi, 2022, Predicting Unplanned 7-Day Intensive Care Unit Readmissions with Machine Learning Models for Improved Discharge Risk Assessment, AMIA Summits Transl. Sci. Proc.
- Partly supportedcites [11] Rosa, R.G., Roehrig, C., Oliveira, R.P.D., Maccari, J.G., Antônio, A.C.P., Castro, P.D.S., Neto, F.L.D., Balzano, P.D.C., and Teixeira, C. (2015). Comparison of Unplanned Intensive Care Unit Readmission Scores: A Prospective Cohort Study. PLoS ONE, 10. sourceExperimental — accuracy not yet measured
The Introduction describes SWIFT as a score developed to predict readmission or death after ICU discharge, and the Discussion reports studies of its predictive performance. Neither section says it is the only validated risk score for ICU discharge; the paper also evaluates other scores without establishing their validation status. (LLM, conf=0.85)
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.
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“A primary limitation of GRU-D, however, is the assumption that variables at the initial time step are fully observed.”
§ AI-Driven Predictions of Readmission and Mortality for Improved Discharge Decisions in Critical Care: A Retrospective Study > 2. Materials and Methods > 2.5. Deep Learning Model GRU-D++ for Predictive Systems
What the sources say
- Could not verifycites [22] Che, 2018, Recurrent Neural Networks for Multivariate Time Series with Missing Values, Sci. Rep. sourceExperimental — accuracy not yet measured
The Methods > Notations section assumes the first observation occurs at time-stamp 0, but does not say that every variable is observed then. Methods > GRU-D describes how missing variables are handled without stating an initial-step requirement. The Discussions > Limitations section does not identify that requirement as a limitation; searches for initial-step and first-observation assumptions found no passage establishing the claim. No page numbers are available in our copy. (LLM, conf=0.73)
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.
- Could not verify
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“the dataset has a disproportionate representation, with 81.8% White, 10.5% Black, 4.3% Hispanic, and 3.3% Asian populations [29].”
§ AI-Driven Predictions of Readmission and Mortality for Improved Discharge Decisions in Critical Care: A Retrospective Study > 5. Limitation
What the sources say
- Partly supportedcites [29] Mohammed, 2023, Racial Disparities in Invasive ICU Treatments Among Septic Patients: High Resolution Electronic Health Records Analysis from MIMIC-IV, Yale J. Biol. Med. sourceExperimental — accuracy not yet measured
The Results (Baseline Study Cohort) give exactly these percentages, but for the final cohort of 23,914 ICU admissions selected for the sepsis study, not for the entire MIMIC-IV dataset. The Methods (Exposure and Outcome) instead describe MIMIC-IV as approximately 70% White. (LLM, conf=0.90)
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
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Additionally, we conducted a fine-tuning experiment to demonstrate that the pre-trained GRU-D++ model (i.e., GRU-D++ trained on MIMIC-IV) can be adapted to a new hospital setting, even with a limited amount of training data.”
§ AI-Driven Predictions of Readmission and Mortality for Improved Discharge Decisions in Critical Care: A Retrospective Study > 3. Results > 3.2. Predictive Performances of Machine Learning Models We developed patient deterioration prediction systems using GRU-D++ and compared them with various artificial intelligence (AI) models. To evaluate these models, we employed two metrics: the area under the receiver operating characteristic curve (AUROC) and the area under the precision-recall curve (AUPR). AUROC is a standard classification metric, whereas the AUPR is particularly suitable for imbalanced data. Table 2 presents the predictive performances of the ML models on the MIMIC-IV and KNUH datasets. For internal validation, we applied ten-fold cross-validation using nine folds for training and one fold for testing while reserving 10% of the training data for model selection. The GRU-D++ model demonstrated the highest performance in internal validation, with an AUROC of 0.802, and achieved an external validation AUROC of 0.756. Additionally, we conducted a fine-tuning experiment to demonstrate that the pre-trained GRU-D++ model (i.e., GRU-D++ trained on MIMIC-IV) can be adapted to a new hospital setting, even with a limited amount of training data. In this experiment, we used 40% of the KNUH data for testing, 10% for model selection, and varying percentages (10%, 20%, 30%, 40%, and 50%) for training. Table 2 presents the results, showing that the predictive performance of the pre-trained GRU-D++ improved as the size of the additional training data increased. In this study, AUROC was employed as the primary metric to ensure standardized benchmarking against existing literature, such as the original SWIFT score, and to evaluate the overall discriminative power of the GRU-D++ architecture across all possible decision thresholds. However, we acknowledge that AUPR is a more sensitive indicator in the context of class imbalance, where the ‘failure’ group represents a small fraction (approximately 7%) of the total population. The observed difference between AUROC and AUPR values reflects the inherent complexity of predicting rare clinical events with high precision. By reporting both metrics, we provide a comprehensive evaluation that balances general model discrimination with the practical challenges of clinical utility in real-world, imbalanced ICU settings.
boostedstates a 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.
40 more claims
Create a free account to see all 83 claims and to verify the 3 references still open.
- It may state two separate claims as one. 25
- Probably not a substantial claim. 19
Citations
Cites
Hochreiter, S., & Schmidhuber, J. (1997) · Neural Computation · doi:10.1162/neco.1997.9.8.1735 · Cited by 102,318 · cited 1×
“Furthermore, recurrent neural networks (RNNs), GRU [20], and long short-term memory models (LSTMs) [21] are DL models designed for time-series analysis and can process real-time data.”
XGBoost: A Scalable Tree Boosting System
Chen, T., & Guestrin, C. (2016) · Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining · doi:10.1145/2939672.2939785 · Cited by 54,569 · cited 1×
“Notably, methods such as logistic regression, decision trees, random forests, and extreme gradient boosting [19] have been widely used in previous studies.”
DeLong, E. R., DeLong, D. M., & Clarke-Pearson, D. L. (1988) · Biometrics · doi:10.2307/2531595 · Cited by 22,649 · cited 1×
“We performed the DeLong test [23] to compare the statistical significance of AUROCs between SWIFT and GRU-D++.”
ASSESSMENT OF COMA AND IMPAIRED CONSCIOUSNESS
Teasdale, G., & Jennett, B. (1974) · The Lancet · doi:10.1016/s0140-6736(74)91639-0 · Cited by 13,215 · cited 1×
“In the ICU, several tools, such as the Acute Physiology and Chronic Health Evaluation (APACHE) score [6] and the Glasgow Coma Scale (GCS) [7], have been used to evaluate patients’ conditions.”
On the Properties of Neural Machine Translation: Encoder–Decoder Approaches
Cho, K., van Merrienboer, B., Bahdanau, D., & Bengio, Y. (2014) · Proceedings of SSST-8, Eighth Workshop on Syntax, Semantics and Structure in Statistical Translation · doi:10.3115/v1/w14-4012 · Cited by 7,099 · cited 1×
“Furthermore, recurrent neural networks (RNNs), GRU [20], and long short-term memory models (LSTMs) [21] are DL models designed for time-series analysis and can process real-time data.”
MIMIC-IV, a freely accessible electronic health record dataset
Johnson, A. E. W., Bulgarelli, L., Shen, L., Gayles, A., Shammout, A., Horng, S., Pollard, T. J., Hao, S., Moody, B., Gow, B., et al. (2023) · Scientific Data · doi:10.1038/s41597-022-01899-x · Cited by 3,401 · cited 1×
“One notable resource contributing to this progress is the Medical Information Market for Intensive Care IV (MIMIC-IV) database [15], which offers a rich array of patient data in the ICU setting.”
Recurrent Neural Networks for Multivariate Time Series with Missing Values
Che, Z., Purushotham, S., Cho, K., Sontag, D., & Liu, Y. (2018) · Scientific Reports · doi:10.1038/s41598-018-24271-9 · Cited by 2,370 · cited 1×
“GRU-D [22] was proposed to address this issue as a GRU-based model that directly handles missing values.”
Clinical review: Scoring systems in the critically ill
Vincent, J.-L., & Moreno, R. (2010) · Critical Care · doi:10.1186/cc8204 · Cited by 634 · cited 1×
“However, these scores can be subjectively interpreted because of the nature of some variables [8].”
Predicting Intensive Care Unit Readmission with Machine Learning Using Electronic Health Record Data
Rojas, J. C., Carey, K. A., Edelson, D. P., Venable, L. R., Howell, M. D., & Churpek, M. M. (2018) · Annals of the American Thoracic Society · doi:10.1513/annalsats.201710-787oc · Cited by 177 · cited 1×
“Rojas et al. [25] described a machine-learning approach to predict ICU readmission that was significantly more accurate than previously published algorithms, as demonstrated in both internal and external validations for the MIMIC-III cohort.”
Gajic, O., Malinchoc, M., Comfere, T. B., Harris, M. R., Achouiti, A., Yilmaz, M., Schultz, M. J., Hubmayr, R. D., Afessa, B., & Farmer, J. C. (2008) · Critical Care Medicine · doi:10.1097/ccm.0b013e318164e3b0 · Cited by 173 · cited 1×
“Developed by Gajic et al. [9] in 2008, the SWIFT score predicts unplanned readmission.”
Predicting outcome in critical care: the current status of the APACHE prognostic scoring system
Wong, D. T., & Knaus, W. A. (1991) · Canadian Journal of Anaesthesia · doi:10.1007/bf03007629 · Cited by 161 · cited 1×
“In the ICU, several tools, such as the Acute Physiology and Chronic Health Evaluation (APACHE) score [6] and the Glasgow Coma Scale (GCS) [7], have been used to evaluate patients’ conditions.”
Readmissions and Death after ICU Discharge: Development and Validation of Two Predictive Models
Badawi, O., & Breslow, M. J. (2012) · PLoS ONE · doi:10.1371/journal.pone.0048758 · Cited by 116 · cited 1×
“However, it can also increase the likelihood of ICU readmission and unanticipated post-discharge death if patients are discharged before they become stable [2].”
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×
“Artificial intelligence (AI) models continue to evolve into clinical decision support systems [13], and these technologies have aided clinicians in various labor-intensive tasks such as rapid diagnosis, prediction of patient outcomes, risk stratification, optimizing resource allocation, and…”
Hosein, F. S., Bobrovitz, N., Berthelot, S., Zygun, D., Ghali, W. A., & Stelfox, H. T. (2013) · Critical Care · doi:10.1186/cc12747 · Cited by 96 · cited 1×
“Risk stratification for patients discharged from the ICU is a complex process with many potential challenges [12].”
Predicting ICU readmission using grouped physiological and medication trends
Xue, Y., Klabjan, D., & Luo, Y. (2019) · Artificial Intelligence in Medicine · doi:10.1016/j.artmed.2018.08.004 · Cited by 73 · cited 1×
“Similarly, Xue et al. [27] built graphs of the temporal patterns of variables and then applied frequent subgraph mining.”
Plotnikoff, K. M., Krewulak, K. D., Hernández, L., Spence, K., Foster, N., Longmore, S., Straus, S. E., Niven, D. J., Parsons Leigh, J., Stelfox, H. T., et al. (2021) · Critical Care · doi:10.1186/s13054-021-03857-2 · Cited by 69 · cited 1×
“Discharge of patients from the intensive care unit (ICU) to the hospital ward is one of the most challenging and high-risk transitions of care during which critically ill patients change their physicians and nurses [1].”
Admission and discharge of critically ill patients
Capuzzo, M., Moreno, R. P., & Alvisi, R. (2010) · Current Opinion in Critical Care · doi:10.1097/mcc.0b013e32833cb874 · Cited by 61
Comparison of Unplanned Intensive Care Unit Readmission Scores: A Prospective Cohort Study
Rosa, R. G., Roehrig, C., Oliveira, R. P. D., Maccari, J. G., Antônio, A. C. P., Castro, P. D. S., Neto, F. L. D., Balzano, P. D. C., & Teixeira, C. (2015) · PLOS ONE · doi:10.1371/journal.pone.0143127 · Cited by 56 · cited 2×
“Nevertheless, although the SWIFT criterion is the only validated risk score for ICU discharge, it performs poorly and inconsistently upon implementation [10,11].”
InfluentialEarly prediction of ICU readmissions using classification algorithms
Loreto, M., Lisboa, T., & Moreira, V. P. (2020) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2020.103636 · Cited by 53 · cited 1×
“Furthermore, Loreto et al. [28] studied the problem of predicting ICU readmissions by using a binary classification task.”
Artificial Intelligence in the Intensive Care Unit: Current Evidence on an Inevitable Future Tool
Suresh, V., Singh, K. K., Vaish, E., Gurjar, M., Anubuvanan, A. M., Khulbe, Y., & Muzaffar, S. N. (2024) · Cureus · doi:10.7759/cureus.59797 · Cited by 48
Prediction of ICU Readmissions Using Data at Patient Discharge
Pakbin, A., Rafi, P., Hurley, N., Schulz, W., Harlan Krumholz, M., & Bobak Mortazavi, J. (2018) · 2018 40th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) · doi:10.1109/embc.2018.8513181 · Cited by 42 · cited 1×
“In another study, Pakbin et al. [26] created a model for predicting the risk of ICU readmission at various time points using data from EHRs in the MIMIC-III database.”
Liu, W., Tao, G., Zhang, Y., Xiao, W., Zhang, J., Liu, Y., Lu, Z., Hua, T., & Yang, M. (2022) · Frontiers in Medicine · doi:10.3389/fmed.2021.814566 · Cited by 25 · cited 1×
“We initially collected 48 clinical variables based on previous studies [10,17].”
Mohammed, S., Matos, J., Doutreligne, M., Celi, L. A., & Struja, T. (2023) · The Yale Journal of Biology and Medicine · doi:10.59249/wdji8829 · Cited by 13
Kim, M., Kim, T.-H., Kim, D., Lee, D., Kim, D., Heo, J., Kang, S., Ha, T., Kim, J., Moon, D. H., et al. (2023) · Journal of Clinical Medicine · doi:10.3390/jcm13010036 · Cited by 13 · cited 1×
“The GRU-D++ model demonstrated superior performance in the early prediction of pressure ulcers [18].”
Self-citationLu, R., Kacha, S., Phothikun, N., et al. (2025) · The American Journal of Surgery · doi:10.1016/j.amjsurg.2025.116460 · Cited by 6 · cited 1×
“Recent studies report a mean ICU readmission rate of 7%, with a range between 4% and 14% [4].”
The Epidemiology of ICU Readmissions Across Ten Health Systems
Amagai, S., Chaudhari, V., Chhikara, K., Ingraham, N. E., Hochberg, C. H., Barker, A., Mao, C., Ortiz, A. C., Weissman, G. E., Schmid, B. E., et al. (2025) · Critical Care Explorations · doi:10.1097/cce.0000000000001341 · Cited by 4
Research works printed but not linked: 3
- Shi, 2022, Predicting Unplanned 7-Day Intensive Care Unit Readmissions with Machine Learning Models for Improved Discharge Risk Assessment, AMIA Summits Transl. Sci. Proc., 2022, 446
- Royalty, J.P. (2021). Machine Learning Time-to-Event Mortality Prediction in MIMIC-IV Critical Care Database. [Doctoral Dissertation, Texas A&M University].
- Lundberg, S.M., and Lee, S.-I. (2017). A Unified Approach to Interpreting Model Predictions. Advances in Neural Information Processing Systems, Curran Associates, Inc.
Metadata
- Authors
- Heo, Y., Kim, M., Han, S.-S., Kim, T.-H., Heo, J., Kim, D., Kim, W. J., Lee, S.-J., Kwon, O. B., Kim, Y., Choi, H.-S., & Moon, D. H.
- Affiliations
- Kangwon National University (KR)
- Department of Research & Development, Ziovision Co., Ltd., Chuncheon 24341, Republic of Korea
- Institute of Medical Science, School of Medicine, Kangwon National University, Chuncheon 24341, Republic of Korea
- Department of Pulmonology, Chinjujeil Hospital, Jinju 52709, Republic of Korea
- Seoul National University of Science and Technology (KR)
- Corresponding author
- Choi, Hyun-Soo
- Moon, Da Hye
- ORCID
- Venue
- Diagnostics
- Publisher
- MDPI AG
- Published
- 2026-03-16
- Volume
- 16
- Issue
- 6
- Pages
- 874
- ISSN
- 2075-4418 (ISSN-L)
- PMC
- PMC13025911
- OpenAlex
- W7137124956
- Cited by
- 0
- References
- 23
- FWCI
- 0.00
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- Diagnostics (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- Korea Health Technology R&D Project through the Korea Health Industry Development Institute
- Ministry of Health & Welfare, Republic of Korea (RS-2021-KH114109)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning in Healthcare
- Health Sciences › Health Professions › Health Information Management › Artificial Intelligence in Healthcare
- Health Sciences › Medicine › Critical Care and Intensive Care Medicine › Sepsis Diagnosis and Treatment
- Keywords
- ICU readmission prediction, post-ICU mortality, deep learning, clinical decision support, risk stratification, MIMIC-IV
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.