CoChat
Sign Up
Cited by 448Machine Learning in HealthcareArtificial Intelligence in HealthcareHealthcare Technology and Patient Monitoring

Explainable artificial intelligence model to predict acute critical illness from electronic health records

Open access
Lauritsen, Simon Meyer, Kristensen, Mads, Olsen, Mathias Vassard, Larsen, Morten Skaarup, Lauritsen, Katrine Meyer, Jørgensen, Marianne Johansson, Lange, Jeppe, Thiesson, Bo·2020·Nature Communications
DOI 10.1038/s41467-020-17431-xPMC PMC7395744
Open this paper in CoChat — free accountOpen PDF

Abstract

Acute critical illness is often preceded by deterioration of routinely measured clinical parameters, e.g., blood pressure and heart rate. Early clinical prediction is typically based on manually calculated screening metrics that simply weigh these parameters, such as early warning scores (EWS). The predictive performance of EWSs yields a tradeoff between sensitivity and specificity that can lead to negative outcomes for the patient. Previous work on electronic health records (EHR) trained artificial intelligence (AI) systems offers promising results with high levels of predictive performance in relation to the early, real-time prediction of acute critical illness. However, without insight into the complex decisions by such system, clinical translation is hindered. Here, we present an explainable AI early warning score (xAI-EWS) system for early detection of acute critical illness. xAI-EWS potentiates clinical translation by accompanying a prediction with information on the EHR data explaining it.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41467-020-17431-x

Cites

45 of 45 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Mixed-method evaluations show that clinicians are more confident and more likely to follow AI-generated recommendations when supported by transparent explanations such as SHAP plots.

    “Mixed-method evaluations show that clinicians are more confident and more likely to follow AI-generated recommendations when supported by transparent explanations such as SHAP plots426,427.”

    From Artificial intelligence-driven multi-omics approaches in Alzheimer's disease: Progress, challenges, and future directions (2025) · cites this paper as [427]

    The Results (Explanations; page unavailable) say the model’s explanations can facilitate transparency and induce trust, and Methods (Evaluation; page unavailable) reports that specialists manually inspected their quality. The paper does not report a mixed-method evaluation showing increased clinician confidence or willingness to follow recommendations. Methods (Explanation module; page unavailable) also says the SHAP toolbox was not used to provide explanations. (LLM, conf=0.86)

    Experimental — accuracy not yet measured

  • An XAI-enhanced early warning system improved diagnostic accuracy for acute deterioration, and an interpretable decision support tool for remote strep throat diagnosis increased clinician performance.

    “For example, an XAI-enhanced early warning system improved diagnostic accuracy for acute deterioration, while an interpretable decision support tool for remote strep throat diagnosis increased clinician performance and highlighted the importance of intelligible feedback for sustained use427,428.”

    From Artificial intelligence-driven multi-omics approaches in Alzheimer's disease: Progress, challenges, and future directions (2025) · cites this paper as [427]

    The Results (Predictive performance; page not provided) report high cross-validated predictive performance for an explainable early warning system detecting sepsis, acute kidney injury, and acute lung injury. The paper does not establish that the system improved clinicians’ diagnostic accuracy in practice. The strep-throat tool is attributed to the other citation in the sentence, not this paper. (LLM, conf=0.85)

    Experimental — accuracy not yet measured

  • Multivariate regression-based models or AI-based early-warning systems have the potential to detect subtle trends in physiologic parameters over time to provide precision and reliability.

    “Multivariate regression-based models or AI-based early-warning systems have the potential to detect subtle trends in physiologic parameters over time to provide precision and reliability (38–41).”

    From Accessing Artificial Intelligence for Clinical Decision-Making (2021) · cites this paper as [39]

  • AI-based algorithms using time series data from the EHR are in development with strong results.

    “AI-based algorithms using time series data from the EHR are in development with strong results.”

    From Accessing Artificial Intelligence for Clinical Decision-Making (2021) · cites this paper as [39]

  • xAI-EWS has demonstrated feasibility using predictions for risk of acute injury, sepsis, and acute lung injury.

    “It has demonstrated feasibility using predictions for risk of acute injury, sepsis, and acute lung injury (39).”

    From Accessing Artificial Intelligence for Clinical Decision-Making (2021) · cites this paper as [39]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

Flagged first, then by citations

Research works printed but not linked: 20

  1. [5]Moor, M., Horn, M., Rieck, B., Roqueiro, D. & Borgwardt, K. Temporal convolutional networks and dynamic time warping can drastically improve the early prediction of Sepsis. In Proceedings of the 4th Machine Learning for Healthcare Conference. PMLR 106 (2019).
  2. [10]Futoma, J. et al. An improved multi-output gaussian process RNN with real-time validation for early Sepsis detection. In Proceedings of Machine Learning for Healthcare 2017. JMLR, 68 (2017).
  3. [11]Futoma, J., Hariharan, S. & Heller, K. Learning to detect sepsis with a multitask gaussian process RNN Classifier. In Proceedings of the 34th International Conference on Machine Learning. ICML vol. 70, 1174–1182 (2017).
  4. [12]Vellido, A., Ribas, V., Morales, C., Sanmartín, A. R. & Ruiz-Rodríguez, J. C. Machine learning for critical care: an overview and a Sepsis case study. In Bioinformatics and Biomedical Engineering 2017 (eds Rojas, I. & Ortuño, F.) (IWBBIO) vol. 10208, 15–30 (2017).
  5. [28]Lea, C., Vidal, R., Reiter, A. & Hager, G. D. Temporal convolutional networks: a unified approach to action segmentation. In Computer Vision – ECCV 2016 Workshops. 47–54 (Springer International Publishing, 2016).
  6. [29]Aksan, E. & Hilliges, O. STCN: Stochastic temporal convolutional networks. In International Conference on Learning Representations 2019, CoRR (2019).
  7. [36]Choi, E. et al. RETAIN: An interpretable predictive model for healthcare using reverse time attention mechanism. In 30th Annual Conference on Neural Information Processing Systems (NIPS 2016). Advances in Neural Information Processing Systems 3512–3520 (2016).
  8. [39]Seymour CW Assessment of clinical criteria for sepsis: for the third international consensus definitions for sepsis and septic shock (Sepsis-3) JAMA 2016 315 762 774 10.1001/jama.2016.0288 26903335 PMC5433435 · DOI · Paper page
  9. [43]Nair, V. & Hinton, G. E. Rectified Linear Units Improve Restricted Boltzmann Machines. In ICML'10: Proceedings of the 27th International Conference on International Conference on Machine Learning, 807–814 (Omnipress, 2010).
  10. [49]Montavon, G., Binder, A., Lapuschkin, S., Samek, W. & Müller, K. R. In Explainable AI: Interpreting, Explaining and Visualizing Deep Learning (eds Samek, W., Montavon, G., Vedaldi, A., Hansen, L. & Müller, K. R.), Vol. 11700 (Springer, 2019).
  11. [54]Zeiler, M. D. & Fergus, R. Visualizing and understanding convolutional networks. In Computer Vision—ECCV. 818–833 (Springer, Cham, 2014).
  12. [55]Springenberg, J. T., Dosovitskiy, A., Brox, T. & Riedmiller, M. Striving for Simplicity: The all convolutional net. In ICLR (workshop track) (2014).
  13. [56]Zintgraf, L. M., Cohen, T. S., Adel, T. & Welling, M. Visualizing deep neural network decisions: prediction difference analysis. In 5th International Conference on Learning Representations (2017).
  14. [57]Ribeiro, M. T., Singh, S. & Guestrin, C. ‘Why Should I Trust You?’: Explaining the predictions of any classifier. In ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD) (2016).
  15. [58]Shrikumar, A., Greenside, P. & Kundaje, A. Not just a black box: learning important features through propagating activation differences. In Proceedings of the 34th International Conference on Machine Learning Vol. 70, 3145–3153 (2017).
  16. [59]Shrikumar, A., Greenside, P. & Kundaje, A. Learning important features through propagating activation differences. In Proceedings of the 34th International Conference on Machine Learning Vol. 70, 3145–3153 (2017).
  17. [60]Sundararajan, M., Taly, A. & Yan, Q. Axiomatic attribution for deep networks. In Proceedings of the 34th International Conference on Machine Learning Vol. 70, 3319–3328 (2017).
  18. [62]Ancona, M., Ceolini, E., Öztireli, C. & Gross, M. Towards better understanding of gradient-based attribution methods for Deep Neural Networks. In Sixth International Conference on Learning Representations (2018).
  19. [64]Ancona, M., Öztireli, C. & Gross, M. Explaining deep neural networks with a polynomial time algorithm for shapley values approximation. In Thirty-sixth International Conference on Machine Learning. ICML (2019).
  20. [68]Riis, A. H. CROSS-TRACKS—a population-based danish cross-sectorial cohort. Preprint at https://www.medrxiv.org/content/10.1101/2020.05.13.20100263v1 (2019). · DOI

Cited by

Sorted by citations

Authors
Lauritsen, S. M., Kristensen, M., Olsen, M. V., Larsen, M. S., Lauritsen, K. M., Jørgensen, M. J., Lange, J., & Thiesson, B.
Affiliations
  • Aarhus University (DK)
  • Aalborg University (DK)
  • Regional Hospital Horsens (DK)
Corresponding author
  • Lauritsen, Simon Meyer
  • Lauritsen, Katrine Meyer
ORCID
Venue
Nature Communications
Publisher
Springer Science and Business Media LLC
Published
2020-07-31
Volume
11
Issue
1
Pages
3852-3852
ISSN
2041-1723 (ISSN-L)
DOI
10.1038/s41467-020-17431-x
PMC
PMC7395744
OpenAlex
W3046761449
Cited by
448
References
45
Citations by year
2026: 54 · 2025: 75 · 2024: 88 · 2023: 78 · 2022: 84 · 2021: 60 · 2020: 9
FWCI
22.70
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Communications (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • arXiv (Cornell University) (submittedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • VBN Forskningsportal (Aalborg Universitet) (submittedVersion, cc-by)
  • arXiv (Cornell University)
Funders
Innovationsfonden (8053-00076B)
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 › Surgery › Healthcare Technology and Patient Monitoring
Keywords
acute critical illness, early warning scores, explainable AI, electronic health records, early detection, clinical translation, clinical deterioration prediction
MeSH terms
Acute Disease, Artificial Intelligence, Blood Pressure, Heart Rate, Humans, Prognosis, ROC Curve, Critical Illness, Sepsis, Area Under Curve, Early Diagnosis, Acute Lung Injury, Electronic Health Records, Acute Kidney Injury

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