Understanding the principles involved in designing, creating and testing clinical prediction rules
This accredited short course will cover the concepts of risk and prediction, statistical techniques used to develop clinical prediction rules, the importance of internal and external validation assessments, and critical appraisal of existing studies of clinical prediction rules. Practical sessions will give students the opportunity to implement common modelling approaches using statistical software (Stata).
The last date for receipt of complete applications is 5pm Friday 29th April 2022. Regrettably, late applications cannot be accepted.
By the end of the course students will:
- Be able to explain the concepts of risk and probability, and how they relate to clinical prediction
- Understand the common statistical models used in developing clinical prediction rules and be able to implement these methods using statistical software
- Gain awareness of more advanced statistical techniques used in developing clinical prediction rules
- Understand the importance of calibration and performing internal and external validation assessments of clinical prediction rules
- Be able to critically appraise publications that report the development or assessment of clinical prediction rules
- Have gained sufficient understanding to be able to design their own studies to either develop new clinical prediction rules or test existing rules
The key topics covered will be:
Modelling, prediction and development of clinical prediction rules
- An overview of the aims of using clinical prediction rules (CPRs)
- Summary of relevant study design issues, such as sample size considerations and the TRIPOD statement
- Using survival analysis and logistic regression for modelling and prediction
- Dealing with missing data
- Advanced modelling methods and updating CPRs
Using and evaluating clinical prediction rules
- Measures of discrimination and calibration
- Internal and external validation
- Critical appraisal and systematic review of CPRs
- Evaluating CPR impact and using CPRs in practice
Students attending this module should have at least some familiarity with the following topics:
- Distinction between continuous and categorical variables
- Interpretation of p-values and categorical variables
- Interpretation of confidence intervals
Students who have attended the module Introduction to Statistics for Health Care Research or received similar training elsewhere will meet these pre-requisites.