Validating pathophysiological models of aging using clinical electronic medical records

Area of Science:

• Bioinformatics
• Molecular Biology
• Pediatrics

Background:

• Bioinformatics analysis of clinical data offers insights into human physiological processes like adolescent development.
• Predictive aging models from cross-sectional data require validation in diverse populations, including clinical settings.

Purpose of the Study:

• To develop and validate regression models predicting adolescent chronological age.
• To assess the generalizability and clinical relevance of these predictive aging models.
• To identify sex-specific biomarkers for adolescent age prediction.

Main Methods:

• Regression models were constructed using 2001-2002 National Health and Nutrition Examination Survey (NHANES) data.
• Models were validated against independent 2003-2004 NHANES data and clinical data from a pediatric hospital.
• Sex-specific predictive biomarkers were identified.

Main Results:

• Distinct predictive models were developed for males and females.
• Alkaline phosphatase and creatinine were predictive biomarkers for both sexes.
• Hematocrit and mean cell volume were specific to males; total serum globulin was specific to females.
• Models demonstrated generalizability and clinical relevance.

Conclusions:

• Predictive aging models are generalizable and clinically relevant.
• Significant molecular and clinical differences exist between adolescent males and females impacting prediction accuracy.
• Integrating epidemiological and clinical data enhances model robustness for understanding physiological processes.