Individualized prediction is an important element of precision medicine. The studies in this thesis of Guiyou Yang centered around prediction of the adverse maternal outcomes of hypertensive disorders of pregnancy, mainly pre-eclampsia.
To the best knowledge, all the existing models predicting this outcome are static models as they were trained on baseline data with the aim of making one-time predictions at a clearly defined moment of the disease course. However, most clinical settings require monitoring of disease progression and repeated or even continuous prediction. Therefore, we sought to investigate whether static prediction models could be appropriately adapted for dynamic predictions in ongoing clinical assessments, as recommended by some clinical guidelines. We also evaluated the prognostic value of blood pressure changes, an important clinical parameter of pre-eclampsia. Furthermore, expecting that static models would prove to be unsuitable, we aimed to explore approaches leveraging repeated measurements to provide accurate dynamic prediction.
Finally, given that a vast number of prediction models have been developed, but only a limited number have undergone evaluation of their clinical impact, we also aimed to propose a method that provides preliminary evidence of clinical impact without conducting a full-scale randomized comparative study.
Guiyou Yang is part of Health in Context.
