1  Epideliologic Research Goals

There ate two common goals for epidemiological research: prediction and causal inference.

1.1 Prediction goal

The primary objective of a prediction goal is to forecast the occurrence or risk of an outcome (Y) based on one or more risk factors (A, L). The focus of this goal is often on making accurate predictions using statistical and machine learning techniques to identify patterns and relationships in the data. Prediction goals help to identify populations at risk and inform targeted prevention strategies.

Total cholesterol level example

A predictive goal, on the other hand, would involve developing a model that can accurately predict total cholesterol levels based on various factors, including Rosuvastatin use and other relevant predictors (e.g., race, sex, and age).

flowchart LR
  A[Rosuvastatin] --> Y(Total cholesterol level)
  B[Race] --> Y
  C[Sex] --> Y
  D[Age] --> Y

1.2 Causal goal

The causal goal focuses on understanding the causal relationship between a risk factor (often a treatment, A) and a health outcome (Y). It aims to identify the factors that contribute to the development, progression, or prevention of a specific disease or health outcome. Control for confounding factors (L) is often a necessary step in understanding such a relationship, as these factors may obscure the true causal relationship between the treatment and the outcome. The focus of this goal is often on estimating the parameter ‘treatment effect’, which represents the causal effect of the treatment (A) on the outcome (Y). Causal goals guide the development of effective interventions and policies by understanding the mechanism by which the factors influence health outcomes.

Total cholesterol level example

Causal goal helps understanding the causal effect of Rosuvastatin on total cholesterol levels while accounting for confounding factors such as race, sex, and age.

flowchart LR
  A[Rosuvastatin] --> Y(Total cholesterol level)
  B[Race] --> Y
  C[Sex] --> Y
  D[Age] --> Y
  B --> A
  C --> A
  D --> A

Important

There can be a overlaps between causal and predictive goals in epidemiological research, but they serve distinct purposes and have different focuses.

  • Understanding the causal relationships between variables can help improve the accuracy of predictive models. For example, if you know that a certain factor causally affects a health outcome, incorporating that factor into the prediction model can lead to better predictions.
  • The process of building a predictive model may shed light on potential causal relationships between variables that could be further investigated in causal analyses.