Job Description

UMC is offering Master's thesis opportunities for spring 2027. We are looking for creative and motivated students with a keen interest in data science or software development. We currently have one proposed thesis project in information technology.

Please note that the thesis proposal has not been finalised. It will be discussed further with the student to tailor it to their interests and background.

Automatic Extraction of data from Summary of Product Characteristics documents

All medicines on the market are accompanied by a Summary of Product Characteristics (SPC), a mandatory document which describes how the medicine works and how to use it safely. The SPC contains information such as pharmaceutical form, strength, therapeutic indications, dosage, and frequency and nature of adverse events (side effects).

In pharmacovigilance, we investigate "combinations": pairs of medicines and adverse events. We look for statistical signals indicating that a medicine may cause a particular adverse event. For this work, it would be useful to have access to structured information from SPCs, so we can identify combinations that are already known, and avoid reinventing the wheel. However, the currently available data sources store SPCs as PDFs, which means the information is unstructured and cannot be used automatically.

The proposed project aims to explore automatic extraction of data such as adverse events and their frequencies from SPCs. This includes investigating tools for data extraction, such as large language models, or Azure's "Text analytics for health" and evaluating how well they perform using metrics such as recall and precision. The exact methods will be selected together with the student, depending on their background and the final scope of the project.