Two-stage procedures for selecting the best diagnostic biomarkers

Area of Science:

• Biostatistics
• Medical Diagnostics
• Statistical Modeling

Background:

• Accurate diagnostic biomarker selection is vital for clinical decision-making.
• The probability of correct biomarker selection is sensitive to nuisance parameter misspecification.
• Existing methods may lack robustness when underlying distributional assumptions are violated.

Purpose of the Study:

• To develop a reliable procedure for selecting the diagnostic biomarker with the highest classification rate.
• To address the impact of nuisance parameter misspecification on biomarker selection accuracy.
• To propose a method for calculating the necessary sample size to achieve a desired level of correct selection.

Main Methods:

• A two-stage procedure is introduced for sample size computation.
• The method accounts for nuisance parameters in the joint distribution of candidate biomarkers.
• Simulation studies were conducted to validate the proposed procedure.

Main Results:

• The proposed two-stage procedure effectively computes the required sample size.
• The method demonstrates robustness against misspecification of nuisance parameters.
• Simulation results confirm the procedure's ability to achieve desired correct selection rates.

Conclusions:

• A robust two-stage sample size determination method is presented for diagnostic biomarker selection.
• The procedure mitigates risks associated with nuisance parameter misspecification.
• This approach enhances the reliability of selecting high-performing diagnostic biomarkers.