File Download

There are no files associated with this item.

  Links for fulltext
     (May Require Subscription)
Supplementary

Article: A novel quantification of information for longitudinal data analyzed by mixed‐effects modeling

TitleA novel quantification of information for longitudinal data analyzed by mixed‐effects modeling
Authors
KeywordsFisher information
longitudinal data
nonlinear mixed‐effects model
relative information
Issue Date2020
PublisherJohn Wiley & Sons Ltd. The Journal's web site is located at http://www.interscience.wiley.com/jpages/1539-1604/
Citation
Pharmaceutical Statistics, 2020, Epub 2020-01-27 How to Cite?
AbstractNonlinear mixed‐effects (NLME) modeling is one of the most powerful tools for analyzing longitudinal data especially under the sparse sampling design. The determinant of the Fisher information matrix is a commonly used global metric of the information that can be provided by the data under a given model. However, in clinical studies, it is also important to measure how much information the data provide for a certain parameter of interest under the assumed model, for example, the clearance in population pharmacokinetic models. This paper proposes a new, easy‐to‐interpret information metric, the “relative information” (RI), which is designed for specific parameters of a model and takes a value between 0% and 100%. We establish the relationship between interindividual variability for a specific parameter and the variance of the associated parameter estimator, demonstrating that, under a “perfect” experiment (eg, infinite samples or/and minimum experimental error), the RI and the variance of the model parameter estimator converge, respectively, to 100% and the ratio of the interindividual variability for that parameter and the number of subjects. Extensive simulation experiments and analyses of three real datasets show that our proposed RI metric can accurately characterize the information for parameters of interest for NLME models. The new information metric can be readily used to facilitate study designs and model diagnosis.
Persistent Identifierhttp://hdl.handle.net/10722/281173
ISSN
2019 Impact Factor: 1.374
2015 SCImago Journal Rankings: 1.088

 

DC FieldValueLanguage
dc.contributor.authorYuan, M-
dc.contributor.authorLi, Y-
dc.contributor.authorYang, Y-
dc.contributor.authorXu, J-
dc.contributor.authorTao, F-
dc.contributor.authorZhao, L-
dc.contributor.authorZhou, H-
dc.contributor.authorPinheiro, J-
dc.contributor.authorXu, XS-
dc.date.accessioned2020-03-09T09:51:10Z-
dc.date.available2020-03-09T09:51:10Z-
dc.date.issued2020-
dc.identifier.citationPharmaceutical Statistics, 2020, Epub 2020-01-27-
dc.identifier.issn1539-1604-
dc.identifier.urihttp://hdl.handle.net/10722/281173-
dc.description.abstractNonlinear mixed‐effects (NLME) modeling is one of the most powerful tools for analyzing longitudinal data especially under the sparse sampling design. The determinant of the Fisher information matrix is a commonly used global metric of the information that can be provided by the data under a given model. However, in clinical studies, it is also important to measure how much information the data provide for a certain parameter of interest under the assumed model, for example, the clearance in population pharmacokinetic models. This paper proposes a new, easy‐to‐interpret information metric, the “relative information” (RI), which is designed for specific parameters of a model and takes a value between 0% and 100%. We establish the relationship between interindividual variability for a specific parameter and the variance of the associated parameter estimator, demonstrating that, under a “perfect” experiment (eg, infinite samples or/and minimum experimental error), the RI and the variance of the model parameter estimator converge, respectively, to 100% and the ratio of the interindividual variability for that parameter and the number of subjects. Extensive simulation experiments and analyses of three real datasets show that our proposed RI metric can accurately characterize the information for parameters of interest for NLME models. The new information metric can be readily used to facilitate study designs and model diagnosis.-
dc.languageeng-
dc.publisherJohn Wiley & Sons Ltd. The Journal's web site is located at http://www.interscience.wiley.com/jpages/1539-1604/-
dc.relation.ispartofPharmaceutical Statistics-
dc.rightsPreprint This is the pre-peer reviewed version of the following article: [FULL CITE], which has been published in final form at [Link to final article using the DOI]. This article may be used for non-commercial purposes in accordance with Wiley Terms and Conditions for Use of Self-Archived Versions. Postprint This is the peer reviewed version of the following article: [FULL CITE], which has been published in final form at [Link to final article using the DOI]. This article may be used for non-commercial purposes in accordance with Wiley Terms and Conditions for Use of Self-Archived Versions.-
dc.subjectFisher information-
dc.subjectlongitudinal data-
dc.subjectnonlinear mixed‐effects model-
dc.subjectrelative information-
dc.titleA novel quantification of information for longitudinal data analyzed by mixed‐effects modeling-
dc.typeArticle-
dc.identifier.emailXu, J: xujf@hku.hk-
dc.identifier.authorityXu, J=rp02086-
dc.description.naturelink_to_subscribed_fulltext-
dc.identifier.doi10.1002/pst.1996-
dc.identifier.pmid31989784-
dc.identifier.scopuseid_2-s2.0-85078729020-
dc.identifier.hkuros309362-
dc.identifier.volumeEpub 2020-01-27-
dc.publisher.placeUnited Kingdom-

Export via OAI-PMH Interface in XML Formats


OR


Export to Other Non-XML Formats