Evaluating a new marker for risk prediction using the test tradeoff: an update

Area of Science:

• Biostatistics
• Medical Decision Making
• Risk Prediction Modeling

Background:

• Traditional risk prediction marker evaluation relies on statistical measures, which struggle to quantify the practical benefit of adding a new marker.
• Purely statistical evaluations lack a clear threshold for determining if a new marker's improvement justifies its cost or potential harm.

Purpose of the Study:

• To introduce and update the theory and estimation of the test tradeoff for evaluating additional risk prediction markers.
• To provide a medical decision-making framework for assessing the net benefit of incorporating new markers in risk prediction.
• To demonstrate the utility of the test tradeoff using a practical example.

Main Methods:

• Utilizes a medical decision-making approach centered on the risk threshold, where the utility of treatment and no treatment is equal.
• Defines the test tradeoff as the minimum number of new marker tests required to offset one true positive and increase net benefit.
• Applies sensitivity analysis by computing the test tradeoff across various risk thresholds.

Main Results:

• The test tradeoff offers an interpretable measure of net benefit for additional risk prediction markers.
• This approach allows for a direct comparison of the costs of testing against the benefits of improved risk prediction.
• The updated theory and estimation provide a robust method for marker evaluation.

Conclusions:

• The medical decision-making approach, specifically the test tradeoff, provides a more practical and interpretable method for evaluating new risk prediction markers than purely statistical measures.
• This framework aids clinicians and researchers in deciding whether the benefits of a new marker outweigh its costs.
• The test tradeoff is a valuable tool for optimizing risk prediction strategies.