Quantiles based personalized treatment selection for multivariate outcomes and multiple treatments

Area of Science:

• Biostatistics
• Clinical Trial Design
• Personalized Medicine

Background:

• Individualized treatment selection is challenging with multiple correlated outcomes.
• Existing methods may not adequately handle complex response patterns or incorporate patient preferences.

Purpose of the Study:

• To propose a flexible method for individualized treatment selection in scenarios with correlated multiple responses.
• To develop a rank aggregation technique for combining correlated rank lists.
• To enable the incorporation of patient and clinician preferences into treatment decisions.

Main Methods:

• Utilizes ranks of quantiles of outcome variables, conditional on patient-specific scores derived from covariates.
• Employs a rank aggregation technique to combine potentially correlated lists of ranks.
• Applicable to any number of treatments, outcome variables, and quantiles, across a broad set of models.

Main Results:

• A simulation study demonstrated the method's performance in finite samples.
• Illustrative examples using diabetes and HIV-1 clinical trial data showcased the procedure's applicability.
• The method effectively handles correlated multiple responses for individualized treatment selection.

Conclusions:

• The proposed method offers a flexible and robust approach to individualized treatment selection.
• It accommodates complex data structures and allows for personalized decision-making.
• Demonstrated applicability in real-world clinical trial data for diabetes and HIV-1.