A highly efficient design strategy for regression with outcome pooling.

Area of Science:

• Biostatistics
• Bioinformatics
• Health Informatics

Background:

• Research involving biospecimens is often limited by the high cost of laboratory assays.
• Current cost-saving strategies like random selection or pooling can lead to substantial loss of statistical efficiency.
• Efficiently analyzing biospecimens is crucial for advancing biomedical research.

Purpose of the Study:

• To propose and evaluate a novel biospecimen pooling strategy using k-means clustering.
• To reduce laboratory costs in biospecimen research while preserving statistical efficiency.
• To compare the k-means pooling strategy with existing methods.

Main Methods:

• A novel pooling strategy based on the k-means clustering algorithm was developed.
• Simulations were performed using the BioCycle study as a model.
• The k-means pooling strategy was compared against random selection and random pooling techniques.
• Analyses were conducted under simple and multiple linear regression models.

Main Results:

• All methods produced unbiased estimates and appropriate confidence interval coverage.
• The k-means clustering pooling strategy yielded the most precise estimates.
• This method closely approximated results from analyzing the full dataset.
• Minimal precision was lost as the number of pools decreased.

Conclusions:

• K-means clustering pooling offers an effective method to reduce laboratory costs for biospecimen analysis.
• This strategy minimizes information loss, especially when laboratory tests are budget-limited.
• The approach is particularly beneficial in regression settings, maintaining high statistical efficiency.