Sparse canonical correlation analysis between an alcohol biomarker and self-reported alcohol consumption

Area of Science:

• Statistics
• Biostatistics
• Biomarker Research

Background:

• Estimating correlations between high-dimensional vectors with small sample sizes is challenging.
• Existing methods for canonical correlation analysis may lack sufficient theoretical grounding.
• Accurate correlation estimation is crucial for applications like linking alcohol biomarkers to self-reported behavior.

Purpose of the Study:

• To develop a statistically robust method for estimating canonical correlation between two high-dimensional random vectors with small sample sizes.
• To provide theoretical and empirical grounding for the proposed method.
• To adapt the method for analyzing the correlation between alcohol biomarkers and self-reports.

Main Methods:

• Developed a novel method to estimate canonical correlation for high-dimensional data and small sample sizes.
• Provided theoretical and empirical justification for the new method.
• Customized the method to select sparse, step-function linear combinations for practical application.

Main Results:

• The proposed method offers a theoretically and empirically grounded approach to canonical correlation analysis.
• The customized method effectively selects sparse, step-function linear combinations.
• The method is applicable to real-world problems, such as correlating alcohol biomarkers with self-reports.

Conclusions:

• The novel method provides a reliable way to estimate canonical correlations in high-dimensional, small-sample settings.
• This approach offers a theoretically sound alternative to existing methods.
• The technique has practical implications for biomarker research and behavioral studies.