Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Variable selection under multiple imputation using the bootstrap in a prognostic study.

Martijn W Heymans1, Stef van Buuren, Dirk L Knol

  • 1Vrije Universiteit, Institute for Health Sciences, Department of Methodology and Applied Biostatistics, Amsterdam, The Netherlands. mw.heymans@vumc.nl

BMC Medical Research Methodology
|July 17, 2007
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Systematic Review on the Content of Outcome Measurement Instruments on Scar Quality.

Plastic and reconstructive surgery. Global open·2019
Same author

The ability of four frailty screening instruments to predict mortality, hospitalization and dependency in (instrumental) activities of daily living.

European journal of ageing·2019
Same author

Observational pain assessment in older persons with dementia in four countries: Observer agreement of items and factor structure of the Pain Assessment in Impaired Cognition.

European journal of pain (London, England)·2019
Same author

Measurement equivalence of the Four-Dimensional Symptom Questionnaire (4DSQ) in adolescents and emerging adults.

PloS one·2019
Same author

Validity of the "Rate-a-Plate" Method to Estimate Energy and Protein Intake in Acutely Ill, Hospitalized Patients.

Nutrition in clinical practice : official publication of the American Society for Parenteral and Enteral Nutrition·2019
Same author

Interobserver reproducibility of tumor uptake quantification with <sup>89</sup>Zr-immuno-PET: a multicenter analysis.

European journal of nuclear medicine and molecular imaging·2019

Combining multiple imputation (MI) with bootstrapping effectively addresses missing data in prognostic studies. This approach enhances variable selection and improves the performance of prognostic models, accounting for both sampling and imputation variations.

Area of Science:

  • Biostatistics
  • Epidemiology
  • Health Data Science

Background:

  • Missing data poses significant challenges in prognostic research.
  • Multiple imputation (MI) is a statistical technique that accounts for imputation uncertainty.
  • Combining MI with bootstrapping offers a novel approach for prognostic variable selection.

Purpose of the Study:

  • To develop and test a methodology combining MI and bootstrapping for prognostic variable selection.
  • To assess prognostic variables for the chronicity of low back pain.
  • To evaluate the influence of sampling and imputation variation on prognostic models.

Main Methods:

  • A prospective cohort study merging data from three randomized controlled trials (RCTs).
  • Four methods were used: MI only, bootstrap only, and two combined MI-bootstrapping approaches.

Related Experiment Videos

  • Variable selection was based on inclusion frequency; model performance was assessed using discriminative and calibrative abilities.

    • Main Results:

    • Imputation variation had a greater impact on variable selection frequency than sampling variation.
    • Combined MI and bootstrapping yielded bootstrap-corrected c-index values of 0.70-0.71 and slope values of 0.64-0.86.
    • The combined methods demonstrated good performance across various variable selection levels.

    Conclusions:

    • Accounting for both imputation and sampling variation is recommended for handling missing data.
    • The combined MI-bootstrapping procedure yields high-performing multivariable prognostic models.
    • This methodology is valuable for datasets with missing values, particularly in prognostic studies.