Survival and longitudinal data Analysis
- Concepts of time, order and random censoring (right and left, likelihood function in these cases,), Introduction to R or Stata.
- Survival Function-Actuarial estimator, Kaplan-Meier (K-M) estimator, Graphical Display for Survival,
- Median survival time and confidence interval for median survival, Hazard Ratio, Relation between Hazard Ratio, Relative Risk and Odds ratio, Relative survival estimation.
- Nonparametric methods for comparing survival functions/distributions: log rank test, Confidence Interval for Hazard Ratio, Stratified Log-rank test, Peto’s test, Gehan test, Mantel-Haenzel test.
- Life Time distributions: Parametric (exponential, gamma, Weibull, log-logistic), Linear failure rate, Parametric inference (point estimation, confidence intervals, scores, Likelihood ratio test. Accelerated failure time model, Cox-Snell residuals.
- Identification of prognostic factors related to survival time: Cox’s proportional hazards regression model with one and several covariates, rank test for the regression coefficients, adequacy assessment of the Proportional Hazards (PH) model, Time dependent extension of the Cox model, Tests with non-proportional hazards, parametric and nonparametric inference for this model.
- Advanced topics in Survival analysis: Correlated survival time models (such as shared frailty models, …), multivariate survival analysis models (such as copula models,).
- Introduction to longitudinal data: data structure and notation, exploring longitudinal data, conceptual framework for continuous response, models for correlation structure; exploring correlation structure, discrete response
- Handling and describing longitudinal data: maximum likelihood estimation under normality, restricted maximum likelihood, large sample inference, implications of missing data.
- Population-averaged vs. subject-specific modelling, overview of general and generalised linear models for longitudinal data, population-Averaged Linear Models for continuous response, model specification, linear random and mixed effects models
ongitudinal study designs, models for two measures, (random effects) growth curve models, marginal models, dynamic (autoregressive) models, latent class models, and models for multivariate outcomes, fixed, random and mixed effects linear models; generalised linear models; diagnostics and model checking; and missing data and non-response issues.
Multilevel analysis of longitudinal data, handling attrition of subjects in longitudinal studies during multilevel analysis
- Reporting and interpreting results from longitudinal data analysis.