The main objectives of this module are:

 

After completion of the course, students are able to

1. Define the role of an epidemiological researcher within the health sciences in relation to biostatisticians and health care workers.
2. Explain the importance and use of prediction modeling in decision making in screening for disease, diagnosis of disease, and treatment of disease.
3. Explain the relation between prediction research and health technology assessment. In particular the student is able to explain the need for a impact study and health economic evaluation.
4. Perform a critical appraisal of a medical scientific article detailing a study on a newly created prediction model, the validation  and  updating  of a prediction model, and  the use of a validated prediction model. In order to do so:

• Students can distinguish between research on prediction and etiology;
• Students can explain the pros and cons of various epidemiological study designs for use in prediction research;
• Students can define and explain various forms of bias specific to prediction research.

5. Design a research study to create and validate a new prediction model. To do so:

• Students can create a multivariable statistical prediction model;
• Students are able to evaluate the performance of a prediction model;
• Students  can  explain,  evaluate  and  ameliorate  the  consequences  of  modeling strategies.

6. Translate the results of a prediction model into a practically applicable format, meeting the needs of health care workers and the lay public.

The module 

Prediction models are used in public health and in clinical medicine to inform screening, diagnosis (secondary prevention) and treatment (tertiary prevention). Predictions can be used to personalize interventions. Moreover, prediction models are increasingly used to select patients for randomized clinical trials. A simple PubMed search on ‘prediction’ yields over 170,000 hits, and the number of studies dealing with prediction has exponentially increased over the years, highlighting the attention for prediction research in medical science. Unfortunately, the development and application of prediction models in medical science is often suboptimal due to small sample sizes, not dealing with missing data, poor understanding of the impact of variable selection strategies, and lack of internal and external validation of prediction models. Moreover, advances in statistical science have led to new insights and opportunities for more accurate and useful prediction models. However, new technologies for prediction are extremely difficult to apply without a solid understanding of the basic concepts underpinning prediction modeling. Finally, very few prediction models are actually used in clinical practice. Prediction model often do not address the health problem requiring intervention, be too complex to understand for lay persons and health workers, or do not offer benefit on health outcome.
During this course students will learn to address these common problems in prediction modeling. First they will learn to critically appraise studies that report newly created or validated prediction models. Next, they will learn to create and evaluate a prediction model using advanced statistical techniques. Finally, students will learn to translate their prediction model a format that is easy to understand for lay persons and health workers alike.
During this course students will learn to critically appraise reported studies on prediction models, to design a research study to create and validate a new prediction model and to translate the results of a prediction model into a practically applicable format.