Pub Date : 2018-06-07DOI: 10.1080/24709360.2018.1477467
W. Henderson
ABSTRACT This article attempts to outline the most important aspects to consider when planning a randomized controlled clinical trial (RCT) and writing a proposal for the RCT. RCTs are generally formulated by a planning committee that should be comprised of members with expertise in the different important features of the trial. Important considerations include background, objectives/hypotheses, experimental design, patient population and recruitment/retention plan, stratification/randomization, experimental treatment, control or comparison treatment, blinding, primary and secondary outcomes, patient follow-up, data to be collected, capture of data and confidentiality, handling of adverse events, sample size/statistical power and feasibility, statistical analysis, ethical issues, and governance of the trial. Real world examples, mostly drawn from the US Department of Veterans Affairs Cooperative Studies Program, are used to illustrate the various important considerations.
{"title":"Clinical trials design and conduct","authors":"W. Henderson","doi":"10.1080/24709360.2018.1477467","DOIUrl":"https://doi.org/10.1080/24709360.2018.1477467","url":null,"abstract":"ABSTRACT This article attempts to outline the most important aspects to consider when planning a randomized controlled clinical trial (RCT) and writing a proposal for the RCT. RCTs are generally formulated by a planning committee that should be comprised of members with expertise in the different important features of the trial. Important considerations include background, objectives/hypotheses, experimental design, patient population and recruitment/retention plan, stratification/randomization, experimental treatment, control or comparison treatment, blinding, primary and secondary outcomes, patient follow-up, data to be collected, capture of data and confidentiality, handling of adverse events, sample size/statistical power and feasibility, statistical analysis, ethical issues, and governance of the trial. Real world examples, mostly drawn from the US Department of Veterans Affairs Cooperative Studies Program, are used to illustrate the various important considerations.","PeriodicalId":37240,"journal":{"name":"Biostatistics and Epidemiology","volume":"4 1","pages":"24 - 37"},"PeriodicalIF":0.0,"publicationDate":"2018-06-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/24709360.2018.1477467","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"42968453","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-06-07DOI: 10.1080/24709360.2018.1477468
M. Maciejewski
ABSTRACT Quasi-experiments are similar to randomized controlled trials in many respects, but there are many challenges in designing and conducting a quasi-experiment when internal validity threats are introduced from the absence of randomization. This paper outlines design, measurement and statistical issues that must be considered prior to the conduct of a quasi-experimental evaluation. We discuss challenges for the internal validity of quasi-experimental designs, inclusion/exclusion criteria, treatment and comparator cohort definitions, and the five types of explanatory variables that one must classify prior to analysis. We discuss data collection and confidentiality, statistical power and conclude with analytic issues that one must consider.
{"title":"Quasi-experimental design","authors":"M. Maciejewski","doi":"10.1080/24709360.2018.1477468","DOIUrl":"https://doi.org/10.1080/24709360.2018.1477468","url":null,"abstract":"ABSTRACT Quasi-experiments are similar to randomized controlled trials in many respects, but there are many challenges in designing and conducting a quasi-experiment when internal validity threats are introduced from the absence of randomization. This paper outlines design, measurement and statistical issues that must be considered prior to the conduct of a quasi-experimental evaluation. We discuss challenges for the internal validity of quasi-experimental designs, inclusion/exclusion criteria, treatment and comparator cohort definitions, and the five types of explanatory variables that one must classify prior to analysis. We discuss data collection and confidentiality, statistical power and conclude with analytic issues that one must consider.","PeriodicalId":37240,"journal":{"name":"Biostatistics and Epidemiology","volume":"4 1","pages":"38 - 47"},"PeriodicalIF":0.0,"publicationDate":"2018-06-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/24709360.2018.1477468","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"46679140","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-01-01DOI: 10.1080/24709360.2018.1469809
Zhicheng Du, Wangjian Zhang, Dingmei Zhang, Shicheng Yu, Y. Hao
ABSTRACT Hand, foot, and mouth disease (HFMD) has become a major public health issue in China, especially in Guangdong. The burden of severe cases deserves further attention. We hereby explored the epidemiological features of severe HFMD in Guangdong. Patients who were from rural areas (OR = 2.03, 95% CI: 1.86–2.21), males (OR = 1.17, 1.07–1.28), aged ≤3 years old (2.48, 1.68–3.68, and 1.63, 1.10–2.41, for ≤1 and 2–3 years, respectively), and/or infected with EV71 (6.69, 5.98–7.49) tended to progress to severe status. Cases from rural areas tended to have a longer interval from onset to diagnosis (p < .001; i.e. the proportions of each interval (≤1, ∼2, ∼3, ∼4, and >4 days) for rural and urban areas in 2009 were 14%, 13%, 14%, 8%, 51%, and 21%, 21%, 15%, 11%, 31%, respectively). The spatial pattern of the epidemics clarified by the flexible scan statistic showed that the clusters of severe cases were observed to be expanding from the Pearl River Delta Region to the Eastern Region and the Mountainous Region. Overall, the relative risk of the most likely clusters ranged from 5.548 to 15.558 (all p < .001). Our results were particularly practical and important for developing severe HFMD-targeted control programs in the context of disease surveillance. Abbreviations: CA16: Coxsackievirus A16; EV71: enterovirus 71; GDP: gross domestic product; HFMD: hand; foot and mouth disease.
{"title":"Epidemiological characteristics of severe cases of hand, foot, and mouth disease in Guangdong, China","authors":"Zhicheng Du, Wangjian Zhang, Dingmei Zhang, Shicheng Yu, Y. Hao","doi":"10.1080/24709360.2018.1469809","DOIUrl":"https://doi.org/10.1080/24709360.2018.1469809","url":null,"abstract":"ABSTRACT Hand, foot, and mouth disease (HFMD) has become a major public health issue in China, especially in Guangdong. The burden of severe cases deserves further attention. We hereby explored the epidemiological features of severe HFMD in Guangdong. Patients who were from rural areas (OR = 2.03, 95% CI: 1.86–2.21), males (OR = 1.17, 1.07–1.28), aged ≤3 years old (2.48, 1.68–3.68, and 1.63, 1.10–2.41, for ≤1 and 2–3 years, respectively), and/or infected with EV71 (6.69, 5.98–7.49) tended to progress to severe status. Cases from rural areas tended to have a longer interval from onset to diagnosis (p < .001; i.e. the proportions of each interval (≤1, ∼2, ∼3, ∼4, and >4 days) for rural and urban areas in 2009 were 14%, 13%, 14%, 8%, 51%, and 21%, 21%, 15%, 11%, 31%, respectively). The spatial pattern of the epidemics clarified by the flexible scan statistic showed that the clusters of severe cases were observed to be expanding from the Pearl River Delta Region to the Eastern Region and the Mountainous Region. Overall, the relative risk of the most likely clusters ranged from 5.548 to 15.558 (all p < .001). Our results were particularly practical and important for developing severe HFMD-targeted control programs in the context of disease surveillance. Abbreviations: CA16: Coxsackievirus A16; EV71: enterovirus 71; GDP: gross domestic product; HFMD: hand; foot and mouth disease.","PeriodicalId":37240,"journal":{"name":"Biostatistics and Epidemiology","volume":"2 1","pages":"114 - 99"},"PeriodicalIF":0.0,"publicationDate":"2018-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/24709360.2018.1469809","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"43540542","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-01-01DOI: 10.1080/24709360.2017.1406567
Binyan Jiang, Jialiang Li, J. Fine
ABSTRACT Instrumental variable (IV) methods are popular in non-experimental settings to estimate the causal effects of scientific interventions. These approaches allow for the consistent estimation of treatment effects even if major confounders are unavailable. There have been some extensions of IV methods to survival analysis recently. We specifically consider the two-step residual inclusion (2SRI) estimator proposed recently in the literature for the additive hazards regression model in this paper. Assuming linear structural equation models for the hazard function, we may attain a closed-form, two-stage estimator for the causal effect in the additive hazards model. The main contribution of this paper is to provide theoretical works for the 2SRI approach. The asymptotic properties of the estimators are rigorously established and the resulting inferences are shown to perform well in numerical studies.
{"title":"On two-step residual inclusion estimator for instrument variable additive hazards model","authors":"Binyan Jiang, Jialiang Li, J. Fine","doi":"10.1080/24709360.2017.1406567","DOIUrl":"https://doi.org/10.1080/24709360.2017.1406567","url":null,"abstract":"ABSTRACT Instrumental variable (IV) methods are popular in non-experimental settings to estimate the causal effects of scientific interventions. These approaches allow for the consistent estimation of treatment effects even if major confounders are unavailable. There have been some extensions of IV methods to survival analysis recently. We specifically consider the two-step residual inclusion (2SRI) estimator proposed recently in the literature for the additive hazards regression model in this paper. Assuming linear structural equation models for the hazard function, we may attain a closed-form, two-stage estimator for the causal effect in the additive hazards model. The main contribution of this paper is to provide theoretical works for the 2SRI approach. The asymptotic properties of the estimators are rigorously established and the resulting inferences are shown to perform well in numerical studies.","PeriodicalId":37240,"journal":{"name":"Biostatistics and Epidemiology","volume":"2 1","pages":"47 - 60"},"PeriodicalIF":0.0,"publicationDate":"2018-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/24709360.2017.1406567","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"48126753","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-01-01DOI: 10.1080/24709360.2018.1529347
R. Sakurai, S. Hattori
ABSTRACT Interval-censored data are common in medical research. Fully parametric models provide simple and efficient inference for the estimation of survival functions using interval-censored observations. Inference based on a parametric regression model requires the complete specification of the probability density function, and therefore, correctly specifying the model is crucial, while the regression diagnostic is a very important step. However, regression diagnostic methods for use with the interval-censored data have not been completely developed. Here, we developed a model-checking procedure based on the cumulative martingale residuals for the interval-censored observations. We employed the conditional expectation of residuals for the diagnostics, because the data showing the exact failure time cannot be obtained for the interval-censoring analyses, and developed the formal resampling-based supremum-type test and graphical model-checking techniques. A simulation study demonstrated an excellent performance of the proposed method during the detection of a misspecified functional form of covariates in the finite sample. Furthermore, we used this method for the analysis of the medical checkup data obtained in Japan.
{"title":"Goodness-of-fit test for the parametric proportional hazard regression model with interval-censored data","authors":"R. Sakurai, S. Hattori","doi":"10.1080/24709360.2018.1529347","DOIUrl":"https://doi.org/10.1080/24709360.2018.1529347","url":null,"abstract":"ABSTRACT Interval-censored data are common in medical research. Fully parametric models provide simple and efficient inference for the estimation of survival functions using interval-censored observations. Inference based on a parametric regression model requires the complete specification of the probability density function, and therefore, correctly specifying the model is crucial, while the regression diagnostic is a very important step. However, regression diagnostic methods for use with the interval-censored data have not been completely developed. Here, we developed a model-checking procedure based on the cumulative martingale residuals for the interval-censored observations. We employed the conditional expectation of residuals for the diagnostics, because the data showing the exact failure time cannot be obtained for the interval-censoring analyses, and developed the formal resampling-based supremum-type test and graphical model-checking techniques. A simulation study demonstrated an excellent performance of the proposed method during the detection of a misspecified functional form of covariates in the finite sample. Furthermore, we used this method for the analysis of the medical checkup data obtained in Japan.","PeriodicalId":37240,"journal":{"name":"Biostatistics and Epidemiology","volume":"2 1","pages":"115 - 131"},"PeriodicalIF":0.0,"publicationDate":"2018-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/24709360.2018.1529347","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"48434276","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-01-01DOI: 10.1080/24709360.2018.1529346
Duo Jiang, Miaoyan Wang
ABSTRACT The advent of large-scale genetic studies has helped bring a new era of biomedical research on dissecting the genetic architecture of complex human disease. Genome-wide association studies (GWASs) and next-generation sequencing studies are two popular tools for identifying genetic variants that are associated with complex traits. This article overviews some of the most important statistical tools for analyzing data from these two types of studies, with an emphasis on single-SNP tests for common variants and region-based tests for rare variants. We compare various statistical methods for common and rare variants in humans, and describe some critical considerations to guide the choice of an analysis method. Also discussed are the related topics of sample ascertainment, missing heritability, and multiple testing correction, as well as some remaining analytical challenges presented by complex trait association mapping using genomic data obtained via high-throughput technologies.
{"title":"Recent developments in statistical methods for GWAS and high-throughput sequencing association studies of complex traits","authors":"Duo Jiang, Miaoyan Wang","doi":"10.1080/24709360.2018.1529346","DOIUrl":"https://doi.org/10.1080/24709360.2018.1529346","url":null,"abstract":"ABSTRACT The advent of large-scale genetic studies has helped bring a new era of biomedical research on dissecting the genetic architecture of complex human disease. Genome-wide association studies (GWASs) and next-generation sequencing studies are two popular tools for identifying genetic variants that are associated with complex traits. This article overviews some of the most important statistical tools for analyzing data from these two types of studies, with an emphasis on single-SNP tests for common variants and region-based tests for rare variants. We compare various statistical methods for common and rare variants in humans, and describe some critical considerations to guide the choice of an analysis method. Also discussed are the related topics of sample ascertainment, missing heritability, and multiple testing correction, as well as some remaining analytical challenges presented by complex trait association mapping using genomic data obtained via high-throughput technologies.","PeriodicalId":37240,"journal":{"name":"Biostatistics and Epidemiology","volume":"2 1","pages":"132 - 159"},"PeriodicalIF":0.0,"publicationDate":"2018-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/24709360.2018.1529346","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"45235221","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-01-01Epub Date: 2017-11-30DOI: 10.1080/24709360.2017.1407866
Jarcy Zee, Sharon X Xie
We propose a method for estimating and comparing proportions of study participants who reached an event of interest during a randomized controlled trial. Standard methods for estimating this proportion include the intent-to-treat method, which counts the number who reached the event of interest divided by the total number of participants, and the completers-only method, which counts the number who reached the event only among those who completed the entire study. When participants drop out of the study early, however, these methods will either be biased or inefficient. We propose to use the Kaplan-Meier method from survival analysis to estimate the proportion of interest in this non-survival setting. We show through extensive simulation studies that the Kaplan-Meier method has less bias and is more efficient than the standard methods. We demonstrate the performance of all methods for estimating proportions in one sample and for comparing proportions across two samples. Finally, we apply the proposed method to a data set for estimating and comparing proportions of patients who achieved treatment response during a Parkinson's disease trial for the treatment of impulse control disorders.
{"title":"The Kaplan-Meier Method for Estimating and Comparing Proportions in a Randomized Controlled Trial with Dropouts.","authors":"Jarcy Zee, Sharon X Xie","doi":"10.1080/24709360.2017.1407866","DOIUrl":"https://doi.org/10.1080/24709360.2017.1407866","url":null,"abstract":"<p><p>We propose a method for estimating and comparing proportions of study participants who reached an event of interest during a randomized controlled trial. Standard methods for estimating this proportion include the intent-to-treat method, which counts the number who reached the event of interest divided by the total number of participants, and the completers-only method, which counts the number who reached the event only among those who completed the entire study. When participants drop out of the study early, however, these methods will either be biased or inefficient. We propose to use the Kaplan-Meier method from survival analysis to estimate the proportion of interest in this non-survival setting. We show through extensive simulation studies that the Kaplan-Meier method has less bias and is more efficient than the standard methods. We demonstrate the performance of all methods for estimating proportions in one sample and for comparing proportions across two samples. Finally, we apply the proposed method to a data set for estimating and comparing proportions of patients who achieved treatment response during a Parkinson's disease trial for the treatment of impulse control disorders.</p>","PeriodicalId":37240,"journal":{"name":"Biostatistics and Epidemiology","volume":"2 1","pages":"23-33"},"PeriodicalIF":0.0,"publicationDate":"2018-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/24709360.2017.1407866","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"36507405","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-01-01DOI: 10.1080/24709360.2017.1400714
Sifan Liu, L. Tian, Steve Lee, Min‐ge Xie
ABSTRACT Meta-analysis with fixed-effects and random-effects models provides a general framework for quantitatively summarizing multiple comparative studies. However, a majority of the conventional methods rely on large-sample approximations to justify their inference, which may be invalid and lead to erroneous conclusions, especially when the number of studies is not large, or sample sizes of the individual studies are small. In this article, we propose a set of ‘exact’ confidence intervals for the overall effect, where the coverage probabilities of the intervals can always be achieved. We start with conventional parametric fixed-effects and random-effects models, and then extend the exact methods beyond the commonly postulated Gaussian assumptions. Efficient numerical algorithms for implementing the proposed methods are developed. We also conduct simulation studies to compare the performance of our proposal to existing methods, indicating our proposed procedures are better in terms of coverage level and robustness. The new proposals are then illustrated with the data from meta-analyses for estimating the efficacy of statins and BCG vaccination.
{"title":"Exact inference on meta-analysis with generalized fixed-effects and random-effects models","authors":"Sifan Liu, L. Tian, Steve Lee, Min‐ge Xie","doi":"10.1080/24709360.2017.1400714","DOIUrl":"https://doi.org/10.1080/24709360.2017.1400714","url":null,"abstract":"ABSTRACT Meta-analysis with fixed-effects and random-effects models provides a general framework for quantitatively summarizing multiple comparative studies. However, a majority of the conventional methods rely on large-sample approximations to justify their inference, which may be invalid and lead to erroneous conclusions, especially when the number of studies is not large, or sample sizes of the individual studies are small. In this article, we propose a set of ‘exact’ confidence intervals for the overall effect, where the coverage probabilities of the intervals can always be achieved. We start with conventional parametric fixed-effects and random-effects models, and then extend the exact methods beyond the commonly postulated Gaussian assumptions. Efficient numerical algorithms for implementing the proposed methods are developed. We also conduct simulation studies to compare the performance of our proposal to existing methods, indicating our proposed procedures are better in terms of coverage level and robustness. The new proposals are then illustrated with the data from meta-analyses for estimating the efficacy of statins and BCG vaccination.","PeriodicalId":37240,"journal":{"name":"Biostatistics and Epidemiology","volume":"2 1","pages":"1 - 22"},"PeriodicalIF":0.0,"publicationDate":"2018-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/24709360.2017.1400714","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"46269654","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-01-01DOI: 10.1080/24709360.2017.1406040
V. Kulothungan, M. Subbiah, R. Ramakrishnan, R. Raman
ABSTRACT The realm of medical statistics or epidemiology encourages the repeated application of few variants of generalized linear model. This work has identified a situation in understanding the risk factor modelling for diabetic retinopathy, major source for blindness in adults and associated with Type II diabetes. Main objective of this study is to retain the ordinal nature of the response variable, one of the main concerns in ordinal regression procedures; and to emphasize the need for applying stereotype regression for bio medical data. Analysis plan envisaged in this study has shown the relevance and scope to extend the use of ordinal regression models.
{"title":"Identifying associated risk factors for severity of diabetic retinopathy from ordinal logistic regression models","authors":"V. Kulothungan, M. Subbiah, R. Ramakrishnan, R. Raman","doi":"10.1080/24709360.2017.1406040","DOIUrl":"https://doi.org/10.1080/24709360.2017.1406040","url":null,"abstract":"ABSTRACT The realm of medical statistics or epidemiology encourages the repeated application of few variants of generalized linear model. This work has identified a situation in understanding the risk factor modelling for diabetic retinopathy, major source for blindness in adults and associated with Type II diabetes. Main objective of this study is to retain the ordinal nature of the response variable, one of the main concerns in ordinal regression procedures; and to emphasize the need for applying stereotype regression for bio medical data. Analysis plan envisaged in this study has shown the relevance and scope to extend the use of ordinal regression models.","PeriodicalId":37240,"journal":{"name":"Biostatistics and Epidemiology","volume":"78 1","pages":"34 - 46"},"PeriodicalIF":0.0,"publicationDate":"2018-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/24709360.2017.1406040","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"41272256","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-01-01DOI: 10.1080/24709360.2018.1435608
Kehao Zhu, Ying Huang, Xiao‐Hua Zhou
ABSTRACT There is a growing interest in the development of statistical methods for personalized medicine or precision medicine, especially for deriving optimal individualized treatment rules (ITRs). An ITR recommends a patient to a treatment based on the patient's characteristics. The common parametric methods for deriving an optimal ITR, which model the clinical endpoint as a function of the patient's characteristics, can have suboptimal performance when the conditional mean model is misspecified. Recent methodology development has cast the problem of deriving optimal ITR under a weighted classification framework. Under this weighted classification framework, we develop a weighted random forests (W-RF) algorithm that derives an optimal ITR nonparametrically. In addition, with the W-RF algorithm, we propose the variable importance measures for quantifying relative relevance of the patient's characteristics to treatment selection, and the out-of-bag estimator for the population average outcome under the estimated optimal ITR. Our proposed methods are evaluated through intensive simulation studies. We illustrate the application of our methods using data from Clinical Antipsychotic Trials of Intervention Effectiveness Alzheimer's Disease Study.
{"title":"Tree-based ensemble methods for individualized treatment rules","authors":"Kehao Zhu, Ying Huang, Xiao‐Hua Zhou","doi":"10.1080/24709360.2018.1435608","DOIUrl":"https://doi.org/10.1080/24709360.2018.1435608","url":null,"abstract":"ABSTRACT There is a growing interest in the development of statistical methods for personalized medicine or precision medicine, especially for deriving optimal individualized treatment rules (ITRs). An ITR recommends a patient to a treatment based on the patient's characteristics. The common parametric methods for deriving an optimal ITR, which model the clinical endpoint as a function of the patient's characteristics, can have suboptimal performance when the conditional mean model is misspecified. Recent methodology development has cast the problem of deriving optimal ITR under a weighted classification framework. Under this weighted classification framework, we develop a weighted random forests (W-RF) algorithm that derives an optimal ITR nonparametrically. In addition, with the W-RF algorithm, we propose the variable importance measures for quantifying relative relevance of the patient's characteristics to treatment selection, and the out-of-bag estimator for the population average outcome under the estimated optimal ITR. Our proposed methods are evaluated through intensive simulation studies. We illustrate the application of our methods using data from Clinical Antipsychotic Trials of Intervention Effectiveness Alzheimer's Disease Study.","PeriodicalId":37240,"journal":{"name":"Biostatistics and Epidemiology","volume":"2 1","pages":"61 - 83"},"PeriodicalIF":0.0,"publicationDate":"2018-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1080/24709360.2018.1435608","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"46111890","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：