Overcoming the Limitations of Multiple Regression in Biological Age Calculation

Pysaruk AV

Published on: 2025-08-14

Abstract

This paper discusses the limitations of the multiple linear regression (MLR) method for calculating biological age (BA), including a systematic error that overestimates the age of young individuals and underestimates that of older individuals. Another significant limitation is the inflexibility of adding new biomarkers to the model without recollecting data, which is due to the interdependence (collinearity) of the indicators. To address these issues, two new approaches are presented. The first method involves correcting the MLR equation using residual regression, which significantly increases age prediction accuracy and eliminates the systematic error. The second, an alternative method, is not based on MLR and considers each biomarker independently. This approach allows for the use of incomplete datasets and the easy addition of new indicators, making it more flexible. However, using incomplete data does reduce the reliability of the assessment. Both proposed methods demonstrate high predictive power, making them suitable for clinical application in assessing age-related changes.