Computer guided analysis of hemodialyzed patients’ bioimpedance spectroscopy parameters
Keywords:
data retrieval, decision tree, hydrate status, extracellular compartment, intracellular compartmentAbstract
Introduction. Monitoring the hydration level in dialyzed patients is an important clinical aspect of treatment quality. Bioimpedance is one of the methods using electric properties of body tissues subjected to an alternating multi-frequency amplitude current in order to assess hydration states.
Aim. The aim of the study was to use the computer guided analysis of hemodialyzed patients’ bioimpedance spectroscopy parameters for desion trees algorithm.
Material and methods. The measurements were conduc-ted on two groups – 50 patients on chronic hemodialysis (the test group) 10 minutes before hemodialysis and 46 healthy volunteers (the control group). The studied parameters were: TBW (total body water), ECW (extracellular water) and ICW (intracellular water). For bioimpedance measurements a bio-impedance analyzer was used (Xitron Technologies, San Diego, CA, USA, model 4000B Bioimpedance spectroscopy device measuring at 50 frequencies between 5 kHz and 1 MHz) with electrodes (7.7 x 1.9 cm²). Bioimpedance was measured in a logarithmic spectrum of 10 frequencies starting from 5 to 500 kHz.
Results. The results of calculations on normal, discredited and normalized data were computed in the R language environment with algorithms from RWeka library to generate simple rules of identification of the aforementioned diseases. The executed experiments affirm a possibility to create good classifiers for detecting a proper patient with the help of J48 (decision tree) i QDA (quadratic discriminant analysis) but only after previous training. Thanks to an appropriate algorithm for testing classifiers, decision trees with the classification error below 5 % were obtained. In the future it will be possible to further diminish the classification error by measuring more variables using more complex algorithms for testing classifiers.
References
1. Perez-Garcia R, Lopez-Gomez JM, Jofre R. Haemodialysis dose, extracellular volume control and arterial hypertension. Nephrol Dial Transplant 2001;16:98-101.
2. Tattersall J. Bioimpedance analysis in dialysis: state of the art and what we can expect. Blood Purif. 2009;27(1):70-4.
3. Wizemann V, Wabel P, Chamney P, Załuska W, Moissl U, Rode C, Małecka-Massalska T, Marcelli D. The mortality risk of overhydration in haemodialysis patients. Nephrol Dial Transplant 2009;24(5):1574-9.
4. Yi-Chou Ch, Han-Hsiang Ch, Jui-Chi Y. Postdialysis extracellular volume is rational for evaluating dry weight in hemodialysis patients. Nephron 2002;90:109-110.
5. Schwenk A, Beisenherz A, Romer K. Phase angle from bioelectrical impedance analysis remains an independent predictive marker in HIV- infected patients in the era or highly active antiretroviral treatment. American Journal of Clinical Nutrition 2000;72:496-501.
6. Spinale FG, PhD, Reines HD, Cook MC. Noninvasive estimation of extravascular lung water using bioimpedance. Journal of Surgical Research 1989;47:535-540.
7. Kushner R. Bioelectrical Impedance Analysis: A Review of Principles and Applications. J Ani Coll of Nutr 1992;11(2):199-206.
8. Ripley B. D. Pattern Recognition and Neural Networks, Cambridge Univ. Press, (1996).
9. VenablesWN, RipleyBD, “Modern Applied Statistics with S”, Fourth edition, Springer, (2002).
