uu.seUppsala University Publications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Model averaging using a cross-validation weighting approach
Uppsala University, Disciplinary Domain of Medicine and Pharmacy, Faculty of Pharmacy, Department of Pharmaceutical Biosciences.
2019 (English)Independent thesis Advanced level (degree of Master (One Year)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

Deciding upon an appropriate dose for a phase III trial is crucial for any drug development project. To help with this decision one can apply model-based estimations. A model can describe the relationship between variables, such as dose and effect. A way to apply this is by fitting several models to the available data and select the most promising one, model selection. However, using this approach will neglect the bias occurring from the model selection step and therefore provide too narrow uncertainty intervals around the estimate. A way to avoid this is the use of model averaging. Instead of selecting a single model, model averaging considers the estimation of all models that sufficiently describe the data. This is done by weighting each model depending on predefined criteria, e.g. how well the model fits the data. The aim of this project was to evaluate model averaging techniques using weights depending on their ability to predict data. To quantify the predictive ability, each model was cross-validated. A cross-validation is done by dividing the dataset into K equal parts. K-1 parts are then used for parameter estimation and the remaining part is used for validation. This is then repeated for all parts and the result from each part is summed up. This project was based on the work of Buatois et al. and the disease progression model of wet age-related macular degeneration was used together with four effect models. Four different methods were compared in this project. Model selection and model averaging depending on either how well the model fit the data or based on cross-validation. These methods were tested on three different dosing scenarios. The methods performed similarly on two of the scenarios but deviated from each other in one scenario. In this scenario the model selection using cross-validation outperformed the rest of the methods. This report shows that cross-validation could be useful in modelling both as a model selection approach or used as a weighting tool for model averaging. However, more research is needed to confirm this.

Place, publisher, year, edition, pages
2019. , p. 23
Keywords [en]
Pharmacometrics, Model Averaging, Cross-Validation, NONMEM
National Category
Pharmaceutical Sciences
Identifiers
URN: urn:nbn:se:uu:diva-377261OAI: oai:DiVA.org:uu-377261DiVA, id: diva2:1289287
Subject / course
Pharmacokinetics
Educational program
Master of Science Programme in Pharmacy
Supervisors
Examiners
Available from: 2019-02-20 Created: 2019-02-16 Last updated: 2019-02-20Bibliographically approved

Open Access in DiVA

No full text in DiVA

By organisation
Department of Pharmaceutical Biosciences
Pharmaceutical Sciences

Search outside of DiVA

GoogleGoogle Scholar

urn-nbn

Altmetric score

urn-nbn
Total: 118 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf