PhD position - methods for clinical impact assessment of (AI) prediction algorithms | Utrecht | UMC Utrecht (2024)

Prediction models are typically developed to predict the presence of a particular condition or outcomenow (diagnosis and management) or in the future (prognosis and prevention). The predictive performanceof such models is assessed by comparing predicted probabilities from the model to observed outcomes inempirical datasets through calibration and discrimination (supervised learning). But good calibration anddiscrimination do not guarantee positive impact on decision making, behavioural changes, or healthoutcomes. Typically this impact is assessed through large randomised studies, but these are expensive,poorly generalizable, and require substantial time investment. Methods for assessing impact of predictionmodels at the development stage, during which empirical information on effectiveness and impact ofthese models is not yet available, are lacking or seldomly used and need to be expanded on. Time for achange!

• You will be working interdisciplinary between the fields of epidemiology and health economic evaluation. Your work will focus on updating existing methods, such as decision curve analysis, and on the development of novel methods to assess the impact of prediction models on decision making and healthoutcomes for which randomised studies are or will not be performed.
• You will be performing evidencesynthesis of existing literature, assessing performance of novel study techniques in simulation studies,and performing analyses on real world empirical datasets.
• You will participate in a large multidisciplinary team, attend weekly meetings, provide presentations ofyour research both during these meetings and (international) conferences, and contribute to education byteaching epidemiology to (medicine) students.

Applicants for this position should have a relevant MSc, for example, in (clinical) epidemiology, health economics, biomedical or health sciences, biostatistics, (applied) data sciences, or artificial intelligence.

You demonstrate a keen interest in the methodology of medical research, have expertise in working with prediction modelling techniques, have experience with programming (e.g., in R), and can work in teams and communicating about your ideas with peers.

Fluency in written and spoken English is required, and fluency in Dutch is an advantage.

• A salary between €3017 and €3824 gross per month (salary scale OIO), based on full-time employment (36 hours).
• Year-end bonus of 8.3% and holiday allowance of 8%.
• Pension insurance with ABP. We take care of approximately 70% of the monthly contribution.
• Access to a variety of online courses through GoodHabitz.