RIDDLE: Race and ethnicity Imputation from Disease history with Deep LEarning
Open accessAbstract
Anonymized electronic medical records are an increasingly popular source of research data. However, these datasets often lack race and ethnicity information. This creates problems for researchers modeling human disease, as race and ethnicity are powerful confounders for many health exposures and treatment outcomes; race and ethnicity are closely linked to population-specific genetic variation. We showed that deep neural networks generate more accurate estimates for missing racial and ethnic information than competing methods (e.g., logistic regression, random forest, support vector machines, and gradient-boosted decision trees). RIDDLE yielded significantly better classification performance across all metrics that were considered: accuracy, cross-entropy loss (error), precision, recall, and area under the curve for receiver operating characteristic plots (all p < 10-9). We made specific efforts to interpret the trained neural network models to identify, quantify, and visualize medical features which are predictive of race and ethnicity. We used these characterizations of informative features to perform a systematic comparison of differential disease patterns by race and ethnicity. The fact that clinical histories are informative for imputing race and ethnicity could reflect (1) a skewed distribution of blue- and white-collar professions across racial and ethnic groups, (2) uneven accessibility and subjective importance of prophylactic health, (3) possible variation in lifestyle, such as dietary habits, and (4) differences in background genetic variation which predispose to diseases.
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.
Integrity signals
Registry
Cites
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.
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?
Several studies looked at the use of imputation to increase the completeness of datasets and avoid casewise deletion.
“Several studies looked at the use of imputation, also referred to as indirect estimation, to increase the completeness of datasets and avoid casewise deletion.14,22,30,45,46,50,61,65,67–70”
From The quality of social determinants data in the electronic health record: a systematic review (2021) · cites this paper as [68]
The Introduction’s “Imputation” section (page unavailable) describes estimating missing race and ethnicity values and using them to generate an imputed dataset, supporting the dataset-completeness part of the claim. Our copy does not say that the study used imputation to avoid casewise deletion. (LLM, conf=0.82)
Experimental — accuracy not yet measuredTwo additional studies looked at imputing race/ethnicity data with anonymized clinical datasets.
“Finally, 2 additional studies looked at imputing race/ethnicity data with anonymized clinical datasets,14,68”
From The quality of social determinants data in the electronic health record: a systematic review (2021) · cites this paper as [68]
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
Adam: A Method for Stochastic Optimization
Kingma, D. P., & Ba, J. (2014) · arXiv · doi:10.48550/arxiv.1412.6980 · Cited by 98,946 · cited 1× · see citing works
“The MLP was trained iteratively using the Adam optimizer [23].”
MethodHao, X.-B., & Zhang, G.-G. (2017) · Encycl. Semantic Comput. Robotic Intell. · doi:10.1142/s2425038416300184 · Cited by 86,177 · cited 2×
“As the input data is transformed and propagated through this network, the data at each layer corresponds to a new representation of the sample [9].”
BackgroundMethodXGBoost: 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 · see citing works
Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification
He, K., Zhang, X., Ren, S., & Sun, J. (2015) · 2015 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2015.123 · Cited by 19,741 · see citing works
TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems
Abadi, M., Agarwal, A., Barham, P., Brevdo, E., Chen, Z., Citro, C., Corrado, G. S., Davis, A., Dean, J., Devin, M., et al. (2016) · arXiv · doi:10.48550/arxiv.1603.04467 · Cited by 9,408 · see citing works
Sterne, J. A. C., White, I. R., Carlin, J. B., Spratt, M., Royston, P., Kenward, M. G., Wood, A. M., & Carpenter, J. R. (2009) · BMJ · doi:10.1136/bmj.b2393 · Cited by 7,299 · cited 1× · see citing works
“of plausible values can be used to generate a single imputed dataset (e.g., by choosing plausible values with highest probability), or to generate multiple imputed datasets as in multiple imputation [4]. In our setting, the goal was to impute the distribution of mutually-exclusive racial and ethnic classes given a set…”
BackgroundLearning Important Features Through Propagating Activation Differences
Shrikumar, A., Greenside, P., & Kundaje, A. (2017) · arXiv · doi:10.48550/arxiv.1704.02685 · Cited by 2,761 · see citing works
Practical Recommendations for Gradient-Based Training of Deep Architectures
Bengio, Y. (2012) · Lecture Notes in Computer Science · doi:10.1007/978-3-642-35289-8_26 · Cited by 2,259 · see citing works
Neural Networks: Tricks of the Trade: Second Edition
(2012) · Lecture Notes in Computer Science · doi:10.1007/978-3-642-35289-8 · Cited by 2,116 · see citing works
The Effect of Race and Sex on Physicians' Recommendations for Cardiac Catheterization
Schulman, K. A., Berlin, J. A., Harless, W., Kerner, J. F., Sistrunk, S., Gersh, B. J., Dubé, R., Taleghani, C. K., Burke, J. E., Williams, S., et al. (1999) · New England Journal of Medicine · doi:10.1056/nejm199902253400806 · Cited by 1,934 · cited 1× · see citing works
“As an example, it has been reported that referrals for cardiac catheterization are rarer among African American patients than in White patients [2].”
BackgroundMining electronic health records: towards better research applications and clinical care
Jensen, P. B., Jensen, L. J., & Brunak, S. (2012) · Nature Reviews Genetics · doi:10.1038/nrg3208 · Cited by 1,680 · cited 1× · see citing works
“Electronic medical records (EMRs) are an increasingly popular source of biomedical research data [1].”
BackgroundThe importance of race and ethnic background in biomedical research and clinical practice.
Burchard, E., Ziv, E., Coyle, N., Gomez, S., Tang, H., Karter, A., Mountain, J., Pérez-Stable, E., Sheppard, D., & Risch, N. (2003) · New England Journal of Medicine · doi:10.1056/nejmsb025007 · Cited by 1,101 · see citing works
Not Just a Black Box: Learning Important Features Through Propagating Activation DifferencesPreprint
Shrikumar, A., Greenside, P., Shcherbina, A., & Kundaje, A. B. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1605.01713 · Cited by 704 · cited 3× · see citing works
“The DeepLIFT algorithm compares the activation of each node to a reference activation; the difference between the reference and observed activation is used to compute the contribution score of a neuron to a class [11].”
MethodInfluentialOlshansky, S. J., Antonucci, T., Berkman, L., Binstock, R. H., Boersch-Supan, A., Cacioppo, J. T., Carnes, B. A., Carstensen, L. L., Fried, L. P., Goldman, D. P., et al. (2012) · Health Affairs · doi:10.1377/hlthaff.2011.0746 · Cited by 693 · cited 1× · see citing works
“These are primarily diseases of older age, suggesting that lifespan varies across race and ethnicity due to socioeconomic and lifestyle reasons, as reported in literature [19, 20].”
BackgroundCrimmins, E., & Saito, Y. (2001) · Social Science & Medicine (1967) · doi:10.1016/s0277-9536(00)00273-2 · Cited by 579 · cited 1× · see citing works
“These are primarily diseases of older age, suggesting that lifespan varies across race and ethnicity due to socioeconomic and lifestyle reasons, as reported in literature [19, 20].”
Background"You Might Also Like:" Privacy Risks of Collaborative Filtering
Calandrino, J. A., Kilzer, A., Narayanan, A., Felten, E. W., & Shmatikov, V. (2011) · 2011 IEEE Symposium on Security and Privacy · doi:10.1109/sp.2011.40 · Cited by 311 · see citing works
Elliott, M., Fremont, A. M., Morrison, P., Pantoja, P. M., & Lurie, N. (2008) · Health Services Research · doi:10.1111/j.1475-6773.2008.00854.x · Cited by 216 · cited 1× · see citing works
“Bayesian approaches to race and ethnicity imputation using census data have been proposed [6] and have been used for race and ethnicity imputation in EMR datasets [7].”
MethodBarber, S., Hickson, D., Wang, X., Sims, M., Nelson, C. R., & Diez-Roux, A. (2016) · American Journal of Public Health · doi:10.2105/ajph.2016.303471 · Cited by 126 · cited 1× · see citing works
“This finding is also reflected in medical literature, where it has been reported that African American and Hispanic populations are at significantly higher risk for heart disease [11–13] and HIV [14–16] than their White peers.”
ResultCrepaz, N., Horn, A. K., Rama, S. M., Griffin, T., DeLuca, J. B., Mullins, M. M., & Aral, S. O. (2006) · Sexually Transmitted Diseases · doi:10.1097/01.olq.0000240342.12960.73 · Cited by 118 · cited 3× · see citing works
“This finding is also reflected in medical literature, where it has been reported that African Americans are at significantly higher risk for heart disease [12, 13] and HIV [14, 15] than their White peers.”
ResultCategorical missing data imputation for software cost estimation by multinomial logistic regression
Sentas, P., & Angelis, L. (2006) · Journal of Systems and Software · doi:10.1016/j.jss.2005.02.026 · Cited by 78 · cited 2×
“Traditionally, logistic regression has been used for categorical imputation tasks [8].”
MethodRacial Differences in Reported Lyme Disease Incidence
Fix, A. D. (2000) · American Journal of Epidemiology · doi:10.1093/aje/152.8.756 · Cited by 77 · cited 3× · see citing works
“It has been reported Lyme disease predominately occurs in Whites, and largely unreported for Hispanics or African Americans [17].”
BackgroundGillum, R., Mussolino, M., & Madans, J. (2000) · Journal of Clinical Epidemiology · doi:10.1016/s0895-4356(99)00208-5 · Cited by 75 · cited 1× · see citing works
“viduals than White individuals. This nding is also re ected in medical literature, where it has been reported that African Americans are at signicantly higher risk for heart disease [12, 13] and HIV [14, 15] than their White peers. The fact that these features are important for imputing race and ethnicity could re ect…”
BackgroundGrundmeier, R. W., Song, L., Ramos, M. J., Fiks, A. G., Elliott, M. N., Fremont, A., Pace, W., Wasserman, R. C., & Localio, R. (2015) · Health Services Research · doi:10.1111/1475-6773.12295 · Cited by 55 · cited 1× · see citing works
“Bayesian approaches to race and ethnicity imputation using census data have been proposed [6] and have been used for race and ethnicity imputation in EMR datasets [7].”
MethodKinsler, J. J., Lee, S.-J., Sayles, J., Newman, P., Diamant, A., & Cunningham, W. (2009) · Journal of health care for the poor and underserved · doi:10.1353/hpu.0.0204 · Cited by 50 · cited 1×
“This finding is also reflected in medical literature, where it has been reported that African American and Hispanic populations are at significantly higher risk for heart disease [11–13] and HIV [14–16] than their White peers.”
ResultEpidemiology and Management of Hypertension in the Hispanic Population
Guzman, N. (2012) · American Journal of Cardiovascular Drugs · doi:10.2165/11631520-000000000-00000 · Cited by 38 · see citing works
Gilbert, K. L., Elder, K. T., Lyons, S., Kaphingst, K. A., Blanchard, M. S., & Goodman, M. S. (2015) · Ethnicity & Disease · doi:10.18865/ed.25.3.295 · Cited by 19 · cited 1× · see citing works
“This finding is also reflected in medical literature, where it has been reported that African Americans are at significantly higher risk for heart disease [12, 13] and HIV [14, 15] than their White peers.”
ResultSrivastava (2014) · doi:10.5555/2627435.2670313 · cited 1×
“Dropout regularizes the neural network by randomly dropping neurons and their connections during training; this limits complex co-adaptations between neurons which may not generalize well outside of the training data [25].”
Background
Flagged first, then by citations
2 of the 29 references couldn’t be matched to a known paper.
Research works printed but not linked: 1
- [17]Ojodu J, Hulihan MM, Pope SN, Grant AM, Centers for Disease C, Prevention. Incidence of sickle cell trait–United States, 2010. MMWR Morb Mortal Wkly Rep. 2014; 63(49): 1155– 8. 25503918 PMC4584538
Cited by
Sorted by citations
Opening the black box: interpretable machine learning for geneticists
Azodi, C. B., Tang, J., & Shiu, S.-H. (2020) · Trends in Genetics · doi:10.1016/j.tig.2020.03.005 · Cited by 473 · see citing works
Opening the Black Box: Interpretable Machine Learning for Geneticists.
Azodi, C. B., Tang, J.-L., & Shiu, S. (2020) · Trends in Genetics · doi:10.20944/preprints202002.0239.v1 · Cited by 411
“This approach (coined DeepLIFT [33]), has been used in multiple biological studies [36–38].”
Deep learning in bioinformatics: Introduction, application, and perspective in the big data era
Li, Y., Huang, C., Ding, L., Li, Z., Pan, Y., & Gao, X. (2019) · Methods · doi:10.1016/j.ymeth.2019.04.008 · Cited by 401 · see citing works
Self-citationDeep learning in bioinformatics: introduction, application, and perspective in big data era
Li, Y., Huang, C., Ding, L., Li, Z., Pan, Y.-J., & Gao, X. (2019) · bioRxiv · doi:10.1101/563601 · Cited by 344
“Meanwhile, deep learning has clearly demonstrated its power in promoting the bioinformatics field [104, 5, 18], including sequence analysis [198, 3, 25, 156, 87, 6, 80, 169, 157, 158, 175], structure prediction and reconstruction [167, 90, 38, 168, 180, 196, 170], biomolecular property and function prediction [85, 204…”
BackgroundSelf-citationThe quality of social determinants data in the electronic health record: a systematic review
Cook, L. A., Sachs, J., & Weiskopf, N. G. (2021) · Journal of the American Medical Informatics Association · doi:10.1093/jamia/ocab199 · Cited by 153 · see citing works
The accuracy of race & ethnicity data in US based healthcare databases: A systematic review.
Johnson, J., Moore, B., Hwang, E. K., Hickner, A., & Yeo, H. L. (2023) · American Journal of Surgery · doi:10.1016/j.amjsurg.2023.05.011 · Cited by 87 · see citing works
Li, Y., Kuwahara, H., Yang, P., Song, L., & Gao, X. (2019) · bioRxiv · doi:10.1101/532226 · Cited by 54 · see citing works
“…other hand, the recent success of graphical models and deep learning in bioinformatics (Zitnik et al. , 2018; Li et al. , 2017; Dai et al. , 2017; Kim et al. , 2018; Xia et al. , 2018) suggests the possibility to systematically incorporate multiple sources of information in the heterogeneous…”
BackgroundSelf-citationSmaili, F. Z., Tian, S., Roy, A., Alazmi, M., Arold, S. T., Mukherjee, S., Hefty, P. S., Chen, W., & Gao, X. (2021) · Genomics Proteomics & Bioinformatics · doi:10.1016/j.gpb.2021.02.001 · Cited by 23 · see citing works
Self-citationArtificial intelligence in respiratory care: perspectives on critical opportunities and challenges
Drummond, D., Adejumo, I., Hansen, K., Poberezhets, V., Slabaugh, G., & Hui, C. Y. (2024) · Breathe · doi:10.1183/20734735.0189-2023 · Cited by 17 · see citing works
Comparing injuries from e-scooters, e-bikes, and bicycles in the United States
Younes, H. (2025) · Journal of Cycling and Micromobility Research · doi:10.1016/j.jcmr.2025.100061 · Cited by 15 · see citing works
Multiple Imputation of Missing Race and Ethnicity in CDC COVID-19 Case-Level Surveillance Data
Zhang, G.-Y., Rose, C. E., Zhang, Y.-J., Li, R., Lee, F. C., Massetti, G. M., & Adams, L. (2022) · International Journal of Statistics in Medical Research · doi:10.6000/1929-6029.2022.11.01 · Cited by 14 · see citing works
“, [15] imputed missing race and ethnicity information using deep learning methods with around 15,000 features including demographic information and clinical events.”
MethodArtificial neural networks as clinical decision support systems
Shafi, I., Ansari, S., Din, S., Jeon, G., & Paul, A. (2021) · Concurrency and Computation Practice and Experience · doi:10.1002/cpe.6342 · Cited by 16 · see citing works
Owosela, B., Steinberg, R. S., Leslie, S. L., Celi, L., Purkayastha, S., Shiradkar, R., Newsome, J. M., & Gichoya, J. (2023) · Int. J. Medical Informatics · doi:10.1016/j.ijmedinf.2023.105303 · Cited by 13 · see citing works
Missing Race and Ethnicity Data in Pediatric Studies
Vedelli, J. K. H., Azizi, Z., & Anand, K. J. S. (2023) · JAMA pediatrics · doi:10.1001/jamapediatrics.2023.4745 · Cited by 11 · see citing works
Cheng, L., Gallegos, I. O., Ouyang, D., Goldin, J. S., & Ho, D. E. (2023) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3593013.3594034 · Cited by 9 · see citing works
Jaba Tqemaladze, M. (2026) · Longevity Horizon · doi:10.65649/t7cfjv29 · Cited by 10
Patterson, A., Elbasir, A., Tian, B., & Auslander, N. (2023) · Cancers · doi:10.3390/cancers15071958 · Cited by 8 · see citing works
Quantifying the Uncertainty of Imputed Demographic Disparity Estimates: The Dual-BootstrapPreprint
Lu, B., Wan, J., Ouyang, D., Goldin, J. S., & Ho, D. E. (2024) · SSRN Electronic Journal · doi:10.2139/ssrn.4787403
“Recent work has also investigated the potential power of machine learning for this task (Cheng et al., 2023; Xue et al., 2019; Kim et al., 2018).”
BackgroundChen, Z., Edwards, A., Hicks, C., & Zhang, K. (2020) · Frontiers in Oncology · doi:10.3389/fonc.2020.00272 · Cited by 7 · see citing works
“(25, 35), National Cancer Institute (34), and Tsivian et al.”
BackgroundMethodInfluentialA Multi-Constraint Monte Carlo Simulation Approach to Downscaling Cancer Data
Liu, L.-B., Cowan, L., Wang, F.-H., & Onega, T. (2025) · Health and Place · doi:10.1016/j.healthplace.2024.103411 · Cited by 6
Reliable imputation of incomplete crash data for predicting driver injury severity.
Gao, X., Jiang, X.-K., Zhuang, D.-Y., Haworth, J., Wang, S.-H., Ilyankou, I., & Chen, H. (2025) · Accident Analysis and Prevention · doi:10.1016/j.aap.2025.108020 · Cited by 5 · see citing works
Self-citationLi, C.-H., Mosaddegh, A., Immergluck, L., Owusu, S., Chowdhury, S., & Baltrus, P. (2026) · Microorganisms · doi:10.3390/microorganisms14030662 · Cited by 1 · see citing works
“Moreover, deep learning approaches have also been used for demographic imputation, especially on large-scale or multimodal datasets [15].”
de la Vega, F. M., Pouliot, Y., & Rhead, B. (2024) · medRxiv · doi:10.1101/2024.04.17.24305991 · Cited by 1 · see citing works
“54 However, all methods suffer from suboptimal accuracy and significant no-call rates 55 and require as input personally identified information such as patient name and address, which makes them impractical in de-identified RWD settings.”
Enhanced Identification of Hispanic Ethnicity Using Clinical Data
Ochoa‐Allemant, P., Tate, J. P., Williams, E. C., Gordon, K. S., Marconi, V. C., Bensley, K. M., Rentsch, C. T., Wang, K., Taddei, T. H., Justice, A. C., et al. (2023) · Medical Care · doi:10.1097/mlr.0000000000001824 · Cited by 3 · see citing works
Towards Structured Prediction in Bioinformatics with Deep LearningPreprint
Li, Y. (2020) · arXiv (Cornell University) · doi:10.25781/kaust-2w40w
“On the other hand, the recent success of graphical models and deep learning in bioinformatics [269, 12, 45, 270, 8] suggests the possibility to systematically incorporate multiple sources of information in the heterogeneous network and learn the highly nonlinear relationship between diseases and genes.”
BackgroundKrishnadas, P., Hackstein, U., Bošnjaković, A., & Aston, P. J. (2026) · IEEE International Conference on Fuzzy Systems (FUZZ-IEEE) · doi:10.1109/fuzz69877.2026.11626495
The Impact of Imputation Timing on Model Performance Estimation
Mitchell, B., & Shrestha, S. (2025) · International Conference on Artificial Intelligence Testing · doi:10.1109/aitest66680.2025.00032
“Researchers have explored more sophisticated imputation methods, such as multiple imputation (MI) [34], singular value decomposition (SVD), K-nearest neighbor (KNN) [42], random forests [40], and deep learning techniques [12], [23].”
Yang, D. X., Hui, Y., Wang, B., Avesta, A., Zuber, C., Park, H., & Aneja, S. (2022) · medRxiv · doi:10.1101/2022.06.12.22276306
“41 Nevertheless, our findings reinforce that while ML techniques hold substantial promise in imputing missing clinical data within oncology, data and model specific performances need to be considered.”
- Authors
- Kim, J.-S., Gao, X., & Rzhetsky, A.
- Affiliations
- Princeton University (US)
- King Abdullah University of Science and Technology (SA)
- University of Chicago (US)
- Corresponding author
- Rzhetsky, Andrey
- Venue
- PLOS Computational Biology
- Publisher
- Public Library of Science (PLoS)
- Published
- 2018-04-26
- Volume
- 14
- Issue
- 4
- Pages
- e1006106
- ISSN
- 1553-734X (ISSN-L)
- 1553-7358
- PMC
- PMC5940243
- OpenAlex
- W2725551975
- Cited by
- 31
- References
- 29
- Citations by year
- 2026: 3 · 2025: 4 · 2024: 4 · 2023: 6 · 2022: 4 · 2021: 3 · 2020: 3 · 2019: 3 · 2018: 1
- FWCI
- 1.41
- Type
- journal article
- Language
- en
- Update notices
- new_version (2018-05-08)
- Open access
- gold
- doi.org
- Licence
- Locations
- PLoS Computational Biology (publishedVersion, cc-by)
- Europe PMC (PubMed Central) (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- Knowledge@UChicago (University of Chicago) (submittedVersion, cc-by-nc-nd)
- RePEc: Research Papers in Economics (submittedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Figshare (submittedVersion, cc-by)
- King Abdullah University of Science and Technology Repository (King Abdullah University of Science and Technology) (submittedVersion, other-oa)
- arXiv (Cornell University) (cc-by)
- University of Chicago (cc-by)
- Funders
- National Institute of Mental Health (1P50MH094267)
- National Heart, Lung, and Blood Institute (R01HL122712)
- Defense Advanced Research Projects Agency (W911NF1410333)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning in Healthcare
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Genetics › Race, Genetics, and Society
- Health Sciences › Health Professions › Health Information Management › Artificial Intelligence in Healthcare
- Keywords
- electronic health records, deep neural network, model interpretability, genetic ancestry
- MeSH terms
- Supervised Machine Learning, Ethnicity, Genetics, Population, Humans, Genetic Variation, Epidemiologic Factors, Neural Networks, Computer, Computational Biology, Genetic Predisposition to Disease, Racial Groups, Electronic Health Records
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
