Addressing Statistical Requirements and Practical Limitations in Development of Biomarker Panels and Prognostic Models

A key aspect of disease prevention and early detection for improved treatment, across a range of medical disciplines, is the development and implementation of prognostic models. Formulating such models, however, involves a range of practical challenges and associated statistical and epidemiological concerns. Although analytical approaches are generally well described for proper design, model development, and further analysis and validation of prognostic models, these methods and approaches are often poorly implemented. Overcoming these challenges must involve a combination of utilization of appropriate methodology (and statistical expertise) and better integration of resources across studies.