Using Artificial Intelligence to Predict the Equilibrated Postdialysis Blood Urea Concentration

Abstract

Total dialysis dose (Kt/V) is considered to be a major determinant of morbidity and mortality in hemodialyzed patients. The continuous growth of the blood urea concentration over the 30- to 60-min period following dialysis, a phenomenon known as urea rebound, is a critical factor in determining the true dose of hemodialysis. The misestimation of the equilibrated (true) postdialysis blood urea or equilibrated Kt/V results in an inadequate hemodialysis prescription, with predictably poor clinical outcomes for the patients. The estimation of the equilibrated postdialysis blood urea (eqU) is therefore crucial in order to estimate the equilibrated (true) Kt/V. In this work we propose a supervised neural network to predict the eqU at 60 min after the end of hemodialysis. The use of this model is new in this field and is shown to be better than the currently accepted methods (Smye for eqU and Daugirdas for eqKt/V). With this approach we achieve a mean difference error of 0.22 ± 7.71 mg/ml (mean % error: 1.88 ± 13.46) on the eqU prediction and a mean difference error for eqKt/V of –0.01 ± 0.15 (mean % error: –0.95 ± 14.73). The equilibrated Kt/V estimated with the eqU calculated using the Smye formula is not appropriate because it showed a great dispersion. The Daugirdas double-pool Kt/V estimation formula appeared to be accurate and in agreement with the results of the HEMO study.

Applying an Artificial Neural Network to Predict Total Body Water in Hemodialysis Patients

“…ANN has the advantage of recognizing relationships between inputs (data from cases) and outputs (known outcomes) that may not be obvious with conventional statistical methods [3]. Furthermore, ANNs can improve accuracy through learning algorithms and have been applied to evaluation of dialytic adequacy [4,5,6,7,8], diagnosis and prognosis of nephropathies [9,10,11,12,13], and issues of renal transplantation [14,15,16]. …”

Bedside Linear Regression Equations to Estimate Equilibrated Blood Urea

“…We showed that by means of linear models we were able to build bedside equations that can be easily implemented in any calculator or electronic spreadsheet such as Excel®. All the presented methods performed better than traditional methods (Smye et al, 1999) over the same data (Fernández et al, 2001) suggesting the appropriateness of the simple linear approaches. In addition, each hemodialysis centre can build its own predictor based on its own patient population by following the described process or implementing the accompanying source code (see appendix).…”

Partial Least Squares Regression: A Valuable Method for Modeling Molecular Behavior in Hemodialysis

“…In the case of urea, the most popular target biomarker, its equilibrium concentration is reached between 30 and 60 min after the end of the session. Most hemodialysis adequacy indices used in practice are based on the molecular concentration at the end of the session, 5,14,15,23,[27][28][29] although this could lead to inadequate hemodialysis (HD) dose estimation. By contrast, the use of an equilibrated concentration of the biomarker produces a better evaluation of treatment since it is independent of marker kinetic behavior.…”

“…By contrast, the use of an equilibrated concentration of the biomarker produces a better evaluation of treatment since it is independent of marker kinetic behavior. [14][15][16]27 In clinical practice, waiting for the achievement of the equilibrated urea concentration is usually impractical. Therefore, the availability of a model able to predict subject-specific equilibrated concentration will be very helpful.…”