{"title":"贝叶斯模型平均:改进匹配病例对照研究的变量选择。","authors":"Yi Mu, I. See, J. Edwards","doi":"10.2427/13048","DOIUrl":null,"url":null,"abstract":"Background\nThe problem of variable selection for risk factor modeling is an ongoing challenge in statistical practice. Classical methods that select one subset of exploratory risk factors dominate the medical research field. However, this approach has been criticized for not taking into account the uncertainty of the model selection process itself. This limitation can be addressed by a Bayesian model averaging approach: instead of focusing on a single model and a few factors, Bayesian model averaging considers all the models with non-negligible probabilities to make inference.\n\n\nMethods\nThis paper reports on a simulation study designed to emulate a matched case-control study and compares classical versus Bayesian model averaging selection methods. We used Matthews's correlation coefficient to measure the quality of binary classifications. Both classical and Bayesian model averaging were also applied and compared for the analysis of a matched case-control study of patients with methicillin-resistant Staphylococcus aureus infections after hospital discharge 2011-2013.\n\n\nResults\nBayesian model averaging outperformed the classical approach with much lower false positive rates and higher Matthew's correlation scores. Bayesian model averaging also produced more reliable and robust effect estimates.\n\n\nConclusion\nBayesian model averaging is a conceptually simple, unified approach that produces robust results. It can be used to replace controversial P-values for case-control study in medical research.","PeriodicalId":45811,"journal":{"name":"Epidemiology Biostatistics and Public Health","volume":"15 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2022-02-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"6","resultStr":"{\"title\":\"Bayesian model averaging: improved variable selection for matched case-control studies.\",\"authors\":\"Yi Mu, I. See, J. Edwards\",\"doi\":\"10.2427/13048\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Background\\nThe problem of variable selection for risk factor modeling is an ongoing challenge in statistical practice. Classical methods that select one subset of exploratory risk factors dominate the medical research field. However, this approach has been criticized for not taking into account the uncertainty of the model selection process itself. This limitation can be addressed by a Bayesian model averaging approach: instead of focusing on a single model and a few factors, Bayesian model averaging considers all the models with non-negligible probabilities to make inference.\\n\\n\\nMethods\\nThis paper reports on a simulation study designed to emulate a matched case-control study and compares classical versus Bayesian model averaging selection methods. We used Matthews's correlation coefficient to measure the quality of binary classifications. Both classical and Bayesian model averaging were also applied and compared for the analysis of a matched case-control study of patients with methicillin-resistant Staphylococcus aureus infections after hospital discharge 2011-2013.\\n\\n\\nResults\\nBayesian model averaging outperformed the classical approach with much lower false positive rates and higher Matthew's correlation scores. Bayesian model averaging also produced more reliable and robust effect estimates.\\n\\n\\nConclusion\\nBayesian model averaging is a conceptually simple, unified approach that produces robust results. It can be used to replace controversial P-values for case-control study in medical research.\",\"PeriodicalId\":45811,\"journal\":{\"name\":\"Epidemiology Biostatistics and Public Health\",\"volume\":\"15 1\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2022-02-07\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"6\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Epidemiology Biostatistics and Public Health\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.2427/13048\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q3\",\"JCRName\":\"Nursing\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Epidemiology Biostatistics and Public Health","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.2427/13048","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"Nursing","Score":null,"Total":0}
Bayesian model averaging: improved variable selection for matched case-control studies.
Background
The problem of variable selection for risk factor modeling is an ongoing challenge in statistical practice. Classical methods that select one subset of exploratory risk factors dominate the medical research field. However, this approach has been criticized for not taking into account the uncertainty of the model selection process itself. This limitation can be addressed by a Bayesian model averaging approach: instead of focusing on a single model and a few factors, Bayesian model averaging considers all the models with non-negligible probabilities to make inference.
Methods
This paper reports on a simulation study designed to emulate a matched case-control study and compares classical versus Bayesian model averaging selection methods. We used Matthews's correlation coefficient to measure the quality of binary classifications. Both classical and Bayesian model averaging were also applied and compared for the analysis of a matched case-control study of patients with methicillin-resistant Staphylococcus aureus infections after hospital discharge 2011-2013.
Results
Bayesian model averaging outperformed the classical approach with much lower false positive rates and higher Matthew's correlation scores. Bayesian model averaging also produced more reliable and robust effect estimates.
Conclusion
Bayesian model averaging is a conceptually simple, unified approach that produces robust results. It can be used to replace controversial P-values for case-control study in medical research.
期刊介绍:
Epidemiology, Biostatistics, and Public Health (EBPH) is a multidisciplinary journal that has two broad aims: -To support the international public health community with publications on health service research, health care management, health policy, and health economics. -To strengthen the evidences on effective preventive interventions. -To advance public health methods, including biostatistics and epidemiology. EBPH welcomes submissions on all public health issues (including topics like eHealth, big data, personalized prevention, epidemiology and risk factors of chronic and infectious diseases); on basic and applied research in epidemiology; and in biostatistics methodology. Primary studies, systematic reviews, and meta-analyses are all welcome, as are research protocols for observational and experimental studies. EBPH aims to be a cross-discipline, international forum for scientific integration and evidence-based policymaking, combining the methodological aspects of epidemiology, biostatistics, and public health research with their practical applications.