Performance Metrics for the Comparative Analysis of Clinical Risk Prediction Models Employing Machine Learning
The DOI matches its Crossref record; 30 of the 30 references it cites were found in a registry or an index.
Abstract
Background: New methods such as machine learning techniques have been increasingly used to enhance the performance of risk predictions for clinical decision-making. However, commonly reported performance metrics may not be sufficient to capture the advantages of these newly proposed models for their adoption by health care professionals to improve care. Machine learning models often improve risk estimation for certain subpopulations that may be missed by these metrics. Methods and Results: This article addresses the limitations of commonly reported metrics for performance comparison and proposes additional metrics. Our discussions cover metrics related to overall performance, discrimination, calibration, resolution, reclassification, and model implementation. Models for predicting acute kidney injury after percutaneous coronary intervention are used to illustrate the use of these metrics. Conclusions: We demonstrate that commonly reported metrics may not have sufficient sensitivity to identify improvement of machine learning models and propose the use of a comprehensive list of performance metrics for reporting and comparing clinical risk prediction models.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Hosmer, D. W., Lemeshow, S., & Sturdivant, R. X. (2013) · Wiley Series in Probability and Statistics · doi:10.1002/9781118548387 · Cited by 13,861
Harrell (2015) · Springer Series in Statistics · doi:10.1007/978-3-319-19425-7 · Cited by 6,878
The relationship between Precision-Recall and ROC curves
Davis, J., & Goadrich, M. (2006) · Proceedings of the 23rd international conference on Machine learning - ICML '06 · doi:10.1145/1143844.1143874 · Cited by 6,559
Decision curve analysis: a novel method for evaluating prediction models
Vickers, A. J., & Elkin, E. B. (2006) · Medical Decision Making · doi:10.1177/0272989x06295361 · Cited by 6,341 · cited 1×
“41 Both of these metrics use weighting strategies to summarize 2 metrics to reflect the tradeoff between them.”
BackgroundSaito, T., & Rehmsmeier, M. (2015) · PLOS ONE · doi:10.1371/journal.pone.0118432 · Cited by 5,309
Assessing the Performance of Prediction Models: A Framework for Traditional and Novel Measures
Steyerberg, E. W., Vickers, A. J., Cook, N. R., Gerds, T., Gonen, M., Obuchowski, N., Pencina, M. J., & Kattan, M. W. (2010) · Epidemiology · doi:10.1097/ede.0b013e3181c30fb2 · Cited by 5,170
Moons, K. G. M., Altman, D. G., Reitsma, J. B., Ioannidis, J. P. A., Macaskill, P., Steyerberg, E. W., Vickers, A. J., Ransohoff, D. F., & Collins, G. S. (2015) · Annals of Internal Medicine · doi:10.7326/m14-0698 · Cited by 5,147 · cited 1×
“Numerous guidelines 13–16 have been proposed and should be followed for reporting model development and performance evaluation.”
BackgroundAUC: A misleading measure of the performance of predictive distribution models
Lobo, J. M., Jiménez‐Valverde, A., & Real, R. (2008) · Global Ecology and Biogeography · doi:10.1111/j.1466-8238.2007.00358.x · Cited by 3,499
Internal validation of predictive models
Steyerberg, E. W., Harrell, F. E., Borsboom, G. J. J. M., Eijkemans, M. J. C., Vergouwe, Y., & Habbema, J. D. F. (2001) · Journal of Clinical Epidemiology · doi:10.1016/s0895-4356(01)00341-9 · Cited by 2,746
Prognosis Research Strategy (PROGRESS) 3: Prognostic Model Research
Steyerberg, E. W., Moons, K. G. M., van der Windt, D. A., Hayden, J. A., Perel, P., Schroter, S., Riley, R. D., Hemingway, H., Altman, D. G., & for the PROGRESS Group (2013) · PLoS Medicine · doi:10.1371/journal.pmed.1001381 · Cited by 1,508
Prediction models need appropriate internal, internal-external, and external validation
Steyerberg, E. W., & Harrell, F. E. (2016) · Journal of Clinical Epidemiology · doi:10.1016/j.jclinepi.2015.04.005 · Cited by 1,304 · cited 1×
“61 Furthermore, models applied to new data sets may need updating when new data sets have different characteristics or outcome prevalence than the training data.”
BackgroundRisk prediction models: II. External validation, model updating, and impact assessment
Moons, K. G. M., Kengne, A. P., Grobbee, D. E., Royston, P., Vergouwe, Y., Altman, D. G., & Woodward, M. (2012) · Heart · doi:10.1136/heartjnl-2011-301247 · Cited by 1,195
Assessing the Generalizability of Prognostic Information
Justice, A. C., Covinsky, K. E., & Berlin, J. A. (1999) · Annals of Internal Medicine · doi:10.7326/0003-4819-130-6-199903160-00016 · Cited by 1,160
Luo, W., Phung, D., Tran, T., Gupta, S., Rana, S., Karmakar, C., Shilton, A., Yearwood, J., Dimitrova, N., Ho, T. B., et al. (2016) · Journal of Medical Internet Research · doi:10.2196/jmir.5870 · Cited by 988
A critical look at methods for handling missing covariates in epidemiologic regression analyses.
Greenland, S., & Finkle, W. D. (1995) · American Journal of Epidemiology · doi:10.1093/oxfordjournals.aje.a117592 · Cited by 891 · cited 1×
“52,53 The amount of missingness in clinical data might be higher for racial or ethnic minorities, which may result in the model performing worse for these subgroups.”
BackgroundPrognosis Research Strategy (PROGRESS) 2: Prognostic Factor Research
Riley, R. D., Hayden, J. A., Steyerberg, E. W., Moons, K. G. M., Abrams, K., Kyzas, P. A., Malats, N., Briggs, A., Schroter, S., Altman, D. G., et al. (2013) · PLoS Medicine · doi:10.1371/journal.pmed.1001380 · Cited by 823
External validation is necessary in prediction research:
Bleeker, S. E., Moll, H. A., Steyerberg, E. W., Donders, A. R. T., Derksen-Lubsen, G., Grobbee, D. E., & Moons, K. G. M. (2003) · Journal of Clinical Epidemiology · doi:10.1016/s0895-4356(03)00207-5 · Cited by 710 · cited 1×
“59,60 Internal validation examines the performance of the model for a population with similar characteristics to those in the training data, while external validation examines whether the model could be extended for use in a slightly different population.”
Reporting of artificial intelligence prediction models
Collins, G. S., & Moons, K. G. M. (2019) · The Lancet · doi:10.1016/s0140-6736(19)30037-6 · Cited by 701
A targeted real-time early warning score (TREWScore) for septic shock.
Henry, K. E., Hager, D. N., Pronovost, P. J., & Saria, S. (2015) · Science Translational Medicine · doi:10.1126/scitranslmed.aab3719 · Cited by 598
Framingham risk score and prediction of lifetime risk for coronary heart disease
Lloyd-Jones, D. M., Wilson, P. W. F., Larson, M. G., Beiser, A., Leip, E. P., D'Agostino, R. B., & Levy, D. (2004) · The American Journal of Cardiology · doi:10.1016/j.amjcard.2004.03.023 · Cited by 566
Interpreting Incremental Value of Markers Added to Risk Prediction Models
Pencina, M. J., D'Agostino, R. B., Pencina, K. M., Janssens, A. C. J. W., & Greenland, P. (2012) · American Journal of Epidemiology · doi:10.1093/aje/kws207 · Cited by 506
Analysis of machine learning techniques for heart failure readmissions
Mortazavi, B. J., Downing, N. S., Bucholz, E. M., Dharmarajan, K., Manhapra, A., Li, S.-X., Negahban, S. N., & Krumholz, H. M. (2016) · Circulation: Cardiovascular Quality and Outcomes · doi:10.1161/circoutcomes.116.003039 · Cited by 401 · cited 1×
“4 This advantage may produce better risk estimates for certain subpopulations, 5–7 but the commonly reported metrics may miss the subtle improvements.”
BackgroundSelf-citationAustin, P. C., & Steyerberg, E. W. (2014) · Statistics in Medicine · doi:10.1002/sim.5941 · Cited by 362 · cited 1×
“44 The calibration slope has several limitations.”
Recommendations for Reporting Machine Learning Analyses in Clinical Research
Stevens, L. M., Mortazavi, B. J., Deo, R. C., Curtis, L., & Kao, D. P. (2020) · Circulation: Cardiovascular Quality and Outcomes · doi:10.1161/circoutcomes.120.006556 · Cited by 268 · cited 1×
“17,18 Given the need for clinicians to be informed consumers of studies comparing model performance, we sought to address the strengths and limitations of established metrics and propose additional metrics, with particular attention to their ability to differentiate performance for various risk…”
BackgroundSelf-citationPepe, M. S., Fan, J., Feng, Z., Gerds, T., & Hilden, J. (2015) · Statistics in Biosciences · doi:10.1007/s12561-014-9118-0 · Cited by 191 · cited 2×
“11,48 Moreover, these summary statistics cannot assess the difference of the models in predicting risks for an individual patient.”
BackgroundA review of statistical updating methods for clinical prediction models
Su, T.-L., Jaki, T., Hickey, G. L., Buchan, I., & Sperrin, M. (2018) · Statistical Methods in Medical Research · doi:10.1177/0962280215626466 · Cited by 147
Validation of probabilistic predictions
Miller, M. E., Langefeld, C. D., Tierney, W. M., Hui, S. L., & McDonald, C. J. (1993) · Medical Decision Making · doi:10.1177/0272989x9301300107 · Cited by 129 · cited 1×
“43 One can also derive nonparametric fit between the outcomes and the predictions, capturing potential nonlinearity in the relationship.”
BackgroundBayati, M., Braverman, M., Gillam, M., Mack, K. M., Ruiz, G., Smith, M. S., & Horvitz, E. (2014) · PLoS ONE · doi:10.1371/journal.pone.0109264 · Cited by 122
Labgold, K., Hamid, S., Shah, S., Gandhi, N. R., Chamberlain, A., Khan, F., Khan, S., Smith, S., Williams, S., Lash, T. L., et al. (2021) · Epidemiology · doi:10.1097/ede.0000000000001314 · Cited by 89
Verification of forecasts expressed in terms of probability
BRIER, G. W. (1950) · Monthly Weather Review · doi:10.1175/1520-0493(1950)078%3c0001:vofeit%3e2.0.co;2 · cited 1×
“22 This score is a proper scoring rule for predicting binary outcomes, meaning that a better model always dominates in Brier score.”
Background
Cited by
Metabolomic profiles predict individual multidisease outcomes
Buergel, T., Steinfeldt, J., Ruyoga, G., Pietzner, M., Bizzarri, D., Vojinovic, D., Upmeier zu Belzen, J., Loock, L., Kittner, P., Christmann, L., et al. (2022) · Nature Medicine · doi:10.1038/s41591-022-01980-3 · Cited by 406
Muse, E. D., & Topol, E. J. (2024) · Cell Metabolism · doi:10.1016/j.cmet.2024.02.002 · Cited by 50
Jain, S. S., Mortazavi, B. J., Chan You, S., Yao, X., Lam, C. S. P., Elias, P., Poterucha, T. J., Avram, R., Tison, G. H., Pirruccello, J., et al. (2025) · JACC · doi:10.1016/j.jacc.2025.07.040 · Cited by 13
Self-citationThe Application of Artificial Intelligence in Spine Surgery: A Scoping Review
Shi, L., Wang, H., & Shea, G. K.-H. (2025) · JAAOS: Global Research and Reviews · doi:10.5435/jaaosglobal-d-24-00405 · Cited by 9
Metadata
- Authors
- Huang, C., Li, S.-X., Caraballo, C., Masoudi, F. A., Rumsfeld, J. S., Spertus, J. A., Normand, S.-L. T., Mortazavi, B. J., & Krumholz, H. M.
- Venue
- Circulation: Cardiovascular Quality and Outcomes
- Published
- 2021-10
- Volume
- 14
- Issue
- 10
- Pages
- e007526-e007526
- ISSN
- 1941-7705 (ISSN-L)
- 1941-7713
- OpenAlex
- W3203359277
- Cited by
- 83
- References
- 30
- FWCI
- 9.58
- Type
- journal article
- Language
- en
- Open access
- closed
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
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.