ľ¹Ï¸£ÀûÓ°ÊÓ

A general practitioner (GP) carefully balances risks in every individual patient to make medical decisions: estimating how symptoms will develop, whether treatment is needed, and what the chances are of serious complications or an underlying disease. Thus, a GP is constantly making medical predictions. 

Clinical prediction models based on medical research can provide evidence to support these decisions. This area of research is called 'prediction research', and the number of prediction models published in the medical literature is growing rapidly. However, only a few of these models have found their way into everyday clinical practice. This thesis explores how prediction research can improve clinical practice, with a focus on general practice.
There are specific prediction research challenges for general practice. For instance, most patients in primary care have a low risk of serious disease, because the overall prevalence of many diseases in the community is low. However, using prediction research, it is possible to identify specific patient populations at higher risk of a certain disease or health outcome. This makes prediction models particularly valuable in the general practice setting.
This thesis describes several case studies showing how prediction models can be used in general practice. These include models predicting thrombotic complications in patients with superficial vein thrombosis, hospitalisation risk in COVID-19 patients, and risk of stroke or transient ischaemic attack in out-of-hours primary care.
By combining clinical insights with statistical methods, this thesis aims to make prediction research more accessible and meaningful to GPs, clinical researchers, and other healthcare professionals. The ultimate aim is to guide towards clinically relevant and useful prediction models that enhance GP care.