Predicting the occurrence of vaccine-preventable diseases in Germany with Machine Learning approaches

Background:

Vaccination is one of the most effective methods of protecting people from infectious diseases and containing the spread of vaccine-preventable pathogens. Vaccination programmes significantly reduce the incidence of vaccine-preventable diseases and even eradicate some of them once a certain level of immunity, depending on the transmissibility of the pathogen, has been reached in the population. As an example, the elimination of measles has been a global goal for decades. In order to achieve elimination status, it must be proven that transmission chains of imported measles viruses are interrupted as fast as possible and do not last longer than 12 months. Although it is assumed that there has been no endemic transmission of measles in Germany for years, it is difficult to prove this due to the high population density and size, international importation and high contagiousness of measles. Although recommendations for standard vaccinations against whooping cough and varicella, for example, have been in place for many years, tens of thousands of cases of both pathogens continued to be observed each year before the pandemic and vaccine preventable diseases are resurging after the pandemic.

Aim/s:

Aim of the study is to implement, compare, and validate machine learning models to predict the spatial and temporal occurrence of vaccine preventable diseases as well as disease dynamics (example: measles) in Germany.

AI methods:

The aim of this methodological work is to develop machine learning models that can be used to analyse, understand and predict the occurrence of vaccine-preventable pathogens with regard to various parameters in order to develop suitable recommendations and measures and tailored. In particular, this project requires spatial ML models with prediction and regression tasks, including convolutional neural networks, Transformers and other attention-like approaches.

Keywords:

Vaccine-preventable diseases, Measles, Spatio-temporal analysis, Regression, Public Health Surveillance

to the top