Background:

Smoking is the single most important health risk in industrialised countries and the leading cause of premature mortality. The costs to the healthcare system for the treatment of diseases and health problems caused by smoking are estimated to be 30.3 billion euros in Germany.

Different tobacco control interventions have been implemented in Germany, such as increased taxation, partial smoking bans, increased sales age, media campaigns, and partial advertising bans. For the evaluation of such preventive measures Public Health Impact Analysis (PHIA) can be used. This is a methodological approach to assess the impact of public policies but also societal changes on health.

In Germany, various surveys on smoking behaviour are available, but there is a lack of research that systematically integrates long-term trend analysis with causal inference methods to assess the effectiveness of tobacco control policies.

Aim/s:

The aim of the PhD project is to describe trends in smoking behaviour in Germany over the last 30 years and their link to tobacco control policies, in order to assess their effectiveness and examine vulnerable groups, using principles of causal inference and trustworthy AI to accurately analyse and interpret the findings.

AI methods:

You will combine data from different data sources (in-house and external) and build a model describing time trends and regional differences in smoking behaviour. You will use visualization tools to present the results of the model. You will relate the results of the model to the implementation of tobacco control interventions in different federal states in order to critically assess the effectivity of the interventions, using machine-learning methods and causal reasoning.

Keywords:

Data synthesis, causal machine learning, changepoint analysis, tobacco control, public health impact analysis