A common reason for stroke is atherosclerosis, that develops during several years without symptoms. Furthermore, atherosclerosis involves many organs and subprocesses affecting the development, and there are many risk factors, making the disease progressions highly complex and individual. Therefore, I want to use mathematical models to capture this complexity, in a way that it can be adjusted to individuals, to be able to predict a stroke before it happens.
The kinds of models I use are of two kinds: 1) mechanistic, that describe the physiology behind a certain biological system (such as blood fat transport) and that can be trained on a specific person’s data to then predict how biomarkers (such as blood fats) evolve over time, and 2) machine learning models, that can estimate the risk of having a stroke in the future. These two kinds of models can then be combined into so called hybrid models, that can estimate your risk of having a stroke given certain scenarios, such as a medication or diet.