README

This repository contains the code used for data analysis in the BONSAI
project (Financed by the Swedish funding agency Vinnova, 2018 - 2020).

The objective of the project was to explore methods for causal
inference for finding causal links between child cancer treatment and
late effects. The idea is to use a constraint based causal inference
method adapted for small data sets (most medical data sets are small
from a machine learning perspective), to distinguish whether a late
effect is caused by personal predisposition, the primary cancer
diagnosis, or one of the treatments for the cancer.

This repository contains two parts. One part contains the code for
arranging the data from several registers (person info, control person
info, diagnosis, treatments, in-patient care registry, out-patient
care registry, death registry, and drug prescription, into a common
file with all information about a patient or control person on one
line each. The other part contains the analysis and visualization code
based on that common file.

The actual data is not included here, since it contains sensitive
personal information and is therefore not publicly available.

All the code is "research project quality", not production quality,
which means that it is patchy, inconsistent, unclearn, not well
documented, and thus not easy to use by someone else. Neverthelsee, it
is included here for transparency of the methods used in the project.

For questions about the code and how to use reuse snippets of it in
other projects, contact Anders Holst: anders.holst@ri.se

For questions about availability of data, contact Helena Linge:
helena.linge@med.lu.se

For suggestion for future research collaboration around these issues,
you can contact any of the above.