New research project on Individual multi-state health modeling using machine learning and LLMs
Led jointly by Professors Mathias Lindholm and Filip Lindskog of Stockholm University, this project will run from 2025 to 2027
SCOR is backing a new health research project in conjunction with Stockholm University, titled: Individual multi-state health modeling using machine learning and LLMs.
The new project focuses on health state modeling, and in particular how state-of-the-art machine learning techniques such as LLMs can improve current practice. This is important for insurers, healthcare providers, and anyone having to better plan for future healthcare needs.
The research will look at individual-level health-state modeling using machine learning techniques, understanding and deconstructing LLM-based multi-state approaches, and generating pseudo multi-state health data.
It will address a number of different issues, such as how to handle widely varying amounts of individual-level health-state data within state-of-the-art machine learning techniques, the dynamics captured and potentially missed by LLMs when summarizing individual life event histories, and how to generate shareable individual-level synthetic pseudo-health data from sensitive information.
