Abstract
Abstract Background and Aims Persistent, low-grade inflammation is a significant component of chronic kidney disease (CKD) that plays a pivotal role in its pathophysiology, progression, complications, and all-cause mortality. Previous studies have searched for simple, cost-effective, and universally available markers useful in the systemic inflammation assessment in CKD patients. In the adult population, an increase in neutrophil count, correlated with a reduction in lymphocyte count, indicates the rate of progression to dialysis and predicts mortality in hemodialysis and peritoneal dialysis patients. Due to modulating role of platelets within the inflammatory pathways, a mean platelet volume (MPV) and platelet-to-lymphocyte ratio have also been proposed as markers of inflammation. The systemic immune inflammation index (SII) is a newly defined ratio combining neutrophil, lymphocyte, and platelet counts. It is proposed as a prognostic indicator comprehensively reflecting patients' inflammatory and immune status. Recently, elevated SII has shown a predictive value for mortality risk among CKD adult patients. SII has not been evaluated in the pediatric CKD population so far. Thus, the study aimed to analyze complete blood cell count (CBC) driven parameters, including SII, in children with CKD and to assess their potential usefulness in the prediction of the need for chronic replacement therapy with the use of artificial intelligence tools. Method The study group consisted of 27 predialysis children with CKD stages 4-5 (stage 4 - 11 patients, stage 15 - 16 children) and 40 patients on chronic dialysis (HD - 21 children, APD - 19 patients). The patients’ age ranged from 5 to 18 years, children under 5 were excluded due to the different CBC profile regarding neutrophil – lymphocyte proportions. The evaluated CBC parameters were: hemoglobin, hematocrit, leukocyte, neutrophil, monocyte, lymphocyte, platelet counts and MPV. Kidney function, standard biochemical parameters, CRP and SII were also analyzed. This database was used for further analysis. The recursively selected subsets of input variables constituted the input of random forrest classifier (RFC). Each variant has been optimized, taking into account various features (accuracy, AUROC, precision, recall, MCC), then the best model was saved. The aim was to narrow the set of input parameters so that high predictive power is secured, whereas overfitting or overcomplicating could be avoided. Moreover, the GINI importance was measured in order to define the parameter with the largest share in the prediction. Results The best Random Forest Classifier contained neutrophil count, MPV, and SII as input variables, and achieved the following values: AUROC 0.9286, accuracy 93.75%, precision 0.9437, recall 0.9375 and MCC 0.87. The statistics for each class were as follows: precision 0.90, recall 1.00 and f1-score 0.95 for children with CKD 4-5 on conservative treatment; precision 1.0, recall 0.86, and f1-score 0.92 for patients on chronic dialysis. The values of mean GINI importance measured for 40 random splits of the base for MPV, neutrophil count and SII, were 0.28, 0.32 and 0.39, respectively. Conclusion RFC built up with the input variables of neutrophil count, MPV, and SII, was the best predictor of progression into pediatric end stage kidney disease requiring chronic dialysis. SII turned out the most important parameter in the created model, having the largest share in the prediction of need for renal replacement therapy.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.