Evaluation of the usefulness of the Monte Carlo method for estimating the optimization of hemodialysis

Abstract

Adequate hemodialysis minimizes complications and improves end-stage renal disease patient survival. Our study proposed a methodology for estimating hemodialysis parameters using two-compartment modelling together with Monte Carlo simulation of the probabilities of the model outputs. In addition, we modelled the removal of uremic toxins during hemodialysis, in comparison with the actual concentration of these toxins in the blood serum. Blood urea and phosphates were measured every 30 min throughout hemodialysis in 10 patients. Using a Monte Carlo simulation on the two-compartment model we estimated hemodialysis compatibility parameters for each patient individually. In patients with non-diabetic kidney disease, the actual urea and phosphate excretion dynamics were consistent with those predicted by the two-compartment model regardless of age, sex, non-diabetic comorbidities, duration of hemodialysis, residual diuresis, or type of vascular access. To measure compatibility, we used graph matching together with a quantitative measure given by a normalized coefficient of determination. In patients with end-stage diabetic kidney disease, the toxin elimination dynamics were significantly greater in the first 30 min of hemodialysis than in patients with non-diabetic kidney disease.