Predicting anemia using NIR spectrum of spent dialysis fluid in hemodialysis patients.

Anemia is commonly present in hemodialysis (HD) patients and significantly affects their survival and quality of life. NIR spectroscopy and machine learning were used as a method to detect anemia in hemodialysis patients. The aim of this investigation has been to evaluate the near-infrared spectroscopy (NIRS) as a method for non-invasive on-line detection of anemia parameters from HD effluent by assessing the correlation between the spectrum of spent dialysate in the wavelength range of 700-1700 nm and the levels of hemoglobin (Hb), red blood cells (RBC), hematocrit (Hct), iron (Fe), total iron binding capacity (TIBC), ferritin (FER), mean corpuscular volume (MCV) and mean corpuscular hemoglobin concentration (MCHC) in patient blood. The obtained correlation coefficient (R) for RBC was 0.93, for Hb 0.92, for Fe 0.94, for TIBC 0.96, for FER 0.91, for Hct 0.94, for MCV 0.92, for MCHC 0.92 and for MCH 0.93. The observed high correlations between the NIR spectrum of the dialysate fluid and the levels of the studied variables support the use of NIRS as a promising method for on-line monitoring of anemia and iron saturation parameters in HD patients.