Prediction Model Integration
In the biomedical area an integrative predictive model is one that combines different biomarkers obtained by different means and apply them with a statistical model to select specific treatments for individual patients based on patient’s characteristics, disease status, and clinical outcome (George 2008).
In general terms, a predictive model aims to capture statistical interactions between biomarkers and treatments. It is commonly abstracted and designed from empirical data (i.e., clinical and omic data). Predictive models should be statistically validated to assess their goodness-of-fit to empirical data, and clinically validated to verify their applicability.
The complexity of biological processes frequently necessitates the use of complex statistical models with a large number of variables. Against this complexity, the amount of biomedical data available is often not sufficient to fully determine a statistical...