Peter Kirwan

About

I’m a postdoctoral research fellow at the Barcelona Institute for Global Health and a visiting researcher at the MRC Biostatistics Unit in Cambridge. My current research focuses on the development of statistical microsimulation and forecasting models to project the health effects of changes in Official Development Assistance (ODA) in low- and middle-income countries. My other research interests include the development of statistical models to estimate infectious disease burden, specifically for HIV.

GitHub · ORCiD · Mastodon

Statistical methods for survival data

This series of blog posts provides an introduction to statistical methods for survival data, with applications for infectious disease modelling and epidemiology. Topics include: classical survival methods, competing risks analysis, and causal inference from observational data.

Part I: Key concepts in infectious disease research

An overview of key concepts, introducing some of the challenges in estimating the burden of infectious disease.

Part II: Survival analysis

18 min

Censoring

Truncation

An introduction to survival analysis, time-to-event data, and how we handle censored observations in epidemiological research.

Part III: Classical survival methods

6 min

Likelihood

Survival models

Exploring two fundamental approaches to analysing survival data: non-parametric Kaplan-Meier estimation and semi-parametric Cox proportional hazards models.

Part IV: Competing risks survival methods

6 min

Competing risks

Stratification

How to handle scenarios where multiple possible events can occur, and the challenges this presents for traditional survival methods.

Part V: Multi-state models

14 min

Transition intensity

Markov property

Some theory of multi-state models and their applications in epidemiology, including estimating transition intensities and the time spent in a state.

Part VI: Statistical inference for multi-state models

10 min

Multi-state models

Statistical inference

Details of two statistical inference methods for multi-state models and how these are well-suited to the investigation of intermittently-observed data.