Approximation of Glomerular Filtration Rate after 1 Year Using Annual Medical Examination Data

Area of Science:

• Nephrology
• Medical Informatics
• Biostatistics

Background:

• Glomerular filtration rate (GFR) is a key indicator of kidney function.
• Predicting future GFR is crucial for managing chronic kidney disease.
• Annual medical data offers a potential resource for GFR prediction models.

Purpose of the Study:

• To develop a predictive formula for GFR after one year.
• To utilize routine medical examination data for GFR forecasting.
• To identify key clinical parameters influencing GFR changes over time.

Main Methods:

• A cohort of 41,337 individuals provided annual medical data.
• Random forest machine learning identified significant clinical parameters.
• Multiple linear regression analysis constructed the predictive formula.
• Pearson correlation assessed the formula's accuracy against estimated GFR (eGFR).

Main Results:

• The derived formula is: approximate GFR = -0.054 × age + 0.162 × hemoglobin - 0.085 × uric acid + 0.849 × eGFR + 11.5.
• The formula demonstrated a strong correlation (r = 0.884, p < 0.001) with actual 1-year eGFR.
• Age, hemoglobin, uric acid, and baseline eGFR were key predictors.

Conclusions:

• The developed approximation formula offers a practical method for predicting GFR one year in advance.
• This tool may enhance proactive kidney function management in general populations.
• The formula's reliance on commonly available data makes it widely applicable.