Algorithm on age partitioning for estimation of reference intervals using clinical laboratory database exemplified...

Establishing accurate reference intervals (RIs) is essential for interpreting clinical test results. Traditional RIs often assume uniformity across age groups, which may not reflect physiological changes during development. Prior research has shown that serum creatinine levels vary significantly with age, particularly in pediatric populations. However, the lack of age-specific RIs can lead to misinterpretation of test results. This gap motivated the development of age-partitioned RIs using large clinical datasets. No prior work had resolved how best to partition age groups for creatinine in children. Existing studies rely on limited sample sizes or predefined age categories. This paper introduces a novel approach using decision trees and statistical validation to identify age subgroups. The goal is to improve diagnostic accuracy by aligning RIs with developmental physiology.

The study aimed to develop and evaluate an algorithm for age-partitioned reference intervals (RIs) for serum creatinine in children. The researchers sought to address the limitations of generalized RIs by identifying age subgroups that reflect true physiological variation. They focused on serum creatinine because it is a commonly measured biomarker with known age-related changes. The motivation for this work stems from the need to enhance diagnostic precision in pediatric populations. By using a large clinical database, the study aimed to provide a reproducible method for age partitioning. The approach combines data visualization, decision tree analysis, and statistical validation. The study also aimed to compare its findings with existing pediatric RI studies. The ultimate goal was to provide a framework for improving RI establishment in clinical practice.

The researchers used a clinical laboratory database containing 164,710 outpatient creatinine test results from children under 18 years old. Data cleaning was performed to exclude outliers and invalid entries, resulting in 136,546 usable samples. Box plots were used to visualize the distribution of creatinine levels by gender and age. A decision tree algorithm was applied to identify optimal age subgroups. The Harris-Boyd and Lahti statistical methods were used to test for significant differences between subgroups. The algorithm was designed to automate age partitioning based on data patterns. The study compared its results with the Canadian Laboratory Initiative in Pediatric Reference Intervals (CLIPIRI) findings. The process combined exploratory data analysis with statistical validation to ensure reliability.

The algorithm identified eight age subgroups for serum creatinine in children aged 0 to 17 years. The decision tree method successfully partitioned the data into distinct age intervals. Statistical validation using Harris-Boyd and Lahti methods confirmed significant differences between subgroups. The results showed strong agreement with the CLIPIRI study's age partitions. Creatinine levels increased with age in both genders, as visualized through box plots. Data cleaning reduced the dataset to 136,546 valid samples, ensuring robust analysis. The method demonstrated reproducibility and consistency in age partitioning. The study provides a validated framework for establishing age-specific RIs in clinical settings.

The study developed an algorithm for age-partitioned reference intervals (RIs) for serum creatinine in children. The authors propose that this method improves diagnostic accuracy by aligning RIs with developmental physiology. The decision tree approach, combined with statistical validation, produced age subgroups with significant differences in creatinine levels. The results align with established pediatric RI studies like CLIPIRI. The algorithm offers a reproducible and data-driven method for age partitioning. The authors suggest that this approach can be adapted for other biomarkers. They emphasize the importance of age-specific RIs in clinical diagnostics. The study contributes a practical framework for improving reference interval establishment.