Information about the project
The data science and AI division at Chalmers University of Technology is recruiting a PhD student for a project on Machine Learning for Causal Inference from Observational Data with Applications in Healthcare.
Many exciting applications of machine learning (ML) are found in medicine. Its potential use cases include diagnostics, treatment planning and understanding disease mechanisms. Notably, causality is a critical component in all of these problems: What is the underlying cause for symptoms? Will a change in treatment cause a beneficial effect? What genes increase the likelihood of developing a disease? These questions cannot always be answered through experimentation—as such experiments are often strictly regulated and sometimes infeasible—but must be addressed using observational data from the healthcare system, patient registries or biobanks.
In this project, we will study evaluation of sequential decision-making policies using observational data. This topic, also known as off-policy evaluation, is critical in many real-world applications, such as the continuous treatment of patients with chronic diseases. A key challenge in solving this problem is high dimensionality. The longer the sequences of decisions are, the more difficult it is to assess the impact of decisions made early in the sequence. Similarly, if the basis for decisions includes a large number of variables, finding two observations in contexts comparable across all variables quickly becomes infeasible. For this reason, it is of utmost importance to compress or represent these data in intelligent ways. In particular, the only data required for learning optimal policies, or evaluating them, are variables that were causal of the observed treatment and outcomes. This project aims to develop algorithms and theory for learning such representations, in a manner that preserves conditions for causal identification and ensures interpretability by domain experts.
The project is a collaboration between Chalmers University of Technology, Uppsala University, and AstraZeneca. The student will pursue a PhD in machine learning within computer science and engineering at Chalmers and will be funded though a collaboration grant from the WASP program.
This is a two-stage call. In the first phase, projects were applied for and evaluated. Approved projects then continued to this, the second phase, where each university open calls for PhD positions. Information about AI/MLX Collaboration Projects can be found in the initial call: https://wasp-sweden.org/positions/wasp-collaboration-projects-within-ai-mlx/