• Home

  • What's new?

    • Latest research
    • Site updates log
  • Prognosis research

    • VIDEO introduction
    • What is prognosis?
    • What is prognosis research?
    • The PROGRESS framework
    • Quotes of note
  • Our book

    • Scope & aims
    • Table of contents
    • About the editors
  • Methods guidance

    • PROGRESS series
    • Prognostic factors
    • Prognostic (prediction) models
    • Predictors of treatment effect
    • Systematic Reviews & Meta-analysis
    • Reporting
    • Software, websites & apps
    • Textbooks
  • Videos

    • Introduction to risk prediction models
    • Introduction to prognosis research
    • Prediction model peformance measures
    • Sample size for prediction modelling
    • Penalisation, shrinkage & CV
    • Reporting & quality
    • Categorisation of continuous variables
    • COVID19 prediction models
    • Controversies in prediction modeling
    • Emerging topics in prediction modelling
    • Predictors of treatment effect
    • IPD meta-analysis
    • Systematic reviews
    • AI for prognosis & prediction
  • Courses & events

    • Training courses
    • Conferences
  • Contact

  • More

    Use tab to navigate through the menu items.
    ​

    Prof Gary Collins gave the talk "Penalisation and shrinkage methods do not guarantee a reliable prediction model" for the MEMTAB 2020 conference, and can be viewed below.

    ​

    The corresponding paper has now been published in the J Clinical Epidemiology (PDF).

    ​

    Penalisation, Shrinkage & cross-validation

    ​

    Prof Robert Tibshirani gave the talk "Cross-validation: what does it estimate and how well does it do it?", which has important implications for the use of CV in prediction model research, and links closely to the work presented by Gary Collins

    ​