Indirect reference intervals using an R pipeline.

Area of Science:

• Clinical Chemistry
• Biostatistics
• Computational Biology

Background:

• Establishing accurate reference intervals (RIs) is crucial for distinguishing healthy from pathological analyte values.
• Robust statistical methodologies are essential for developing reliable indirect RIs.
• The R statistical programming language offers a flexible platform for creating such data pipelines.

Purpose of the Study:

• To develop and validate a data pipeline using R for the estimation of indirect reference intervals.
• To apply the pipeline to large-scale datasets for testosterone and aspartate aminotransferase.
• To assess the utility and limitations of the R-based pipeline for RI determination.

Main Methods:

• An R data pipeline was designed to process large datasets (NHANES).
• The pipeline included steps for data ingestion, partitioning, normalization, and outlier removal.
• Reference intervals were identified for testosterone (n=7,207) and aspartate aminotransferase (n=5,882).

Main Results:

• The pipeline generated reference interval estimates for AST and testosterone that closely approximated existing RIs.
• The accuracy of the estimated RIs was influenced by analyte pathology and the characteristics of the dataset used.
• The study highlights the need for careful consideration of these factors when applying the pipeline.

Conclusions:

• A robust statistical pipeline for indirect reference interval estimation can be successfully implemented using R.
• The developed R pipeline provides a valuable tool for clinical laboratories and researchers.
• Further refinement may be necessary depending on specific analytical and dataset considerations.