Automatic Construction of Dynamic and Setting-Specific Prediction Rules in Radiology
Start date: 01-03-2013
Duration: 48 months
Status: In progress
The rapid advancement in medical imaging technology has dramatically increased the availability of diagnostic imaging procedures. Clinical prediction rules that estimate disease probabilities can be used to determine the appropriate imaging strategy. However, these prediction rules are usually not setting specific, i.e., they are not tailored to the patient population of a specific hospital, and they are static, i.e., they are developed at a certain point in time, causing them to become outdated rather quickly.
In this project, we develop a system for the automatic construction and dynamic adaptation of clinical prediction rules. We focus on two clinical problems: the use of computed tomography in case of minimal head injury, and the use of computed tomography angiography in case of suspected coronary artery disease.
The project is funded by the Dutch Technology Foundation STW, and is done in collaboration with the Departments of Radiology and Epidemiology of the Erasmus MC.