Time series analysis is a core tool for epidemiological and public health research, with applications spanning various research areas such as evaluating public health interventions and assessing health effects associated with environmental stressors and climate change. This training provides an introduction to quantitative time series data and analysis in health research for investigating the associations between exposure/predictor and response/outcome relationships of data collected at equally-spaced times. A common feature of the effect of exposure on an outcome (e.g. the effect of extreme temperature on mortality risk) is the lagged effect, i.e. a sustained effect over a period of time. The 3-dimensional exposure—lag—response relationships can be investigated together using distributed lag linear/non-linear models (DLMs/DLNMs), which will also be introduced in this training with a lecture and a demo showing how to conduct it in R/RStudio.
The learning outcomes are understanding the concepts of time series data and time series regression with distributed lag non-linear models and how to conduct the analysis in R/RStudio. Basic knowledge and skill of R/RStudio are preferable but not essential. The demonstration using R/RStudio will be explained step by step with a focus on understanding the model rather than coding.