Determination of the chemical composition of urinary stones in vivo by the profile of nutrient consumption


DOI: 10.29188/2222-8543-2020-13-4-50-56

M.Yu. Prosyannikov, S.A. Golovanov, O.V. Konstantinova, D.A. Voytko, N.V. Anokhin, A.V. Sivkov, O.I. Apolikhin
№4 2020

Introduction. Сurrently, existing methods for determining the chemical composition of the stone in vivo do not have the necessary accuracy. In this regard, the development of methods for high-precision determination of the chemical nature of urinary stones using modern technologies is relevant for modern urology.

Materials and methods. 72 patients with urolithiasis who were treated at the Institute of urology and interventional radiology. N.A. Lopatkin – a branch of the Federal state budgetary institution «NMIC of radiology» of the Ministry of health of Russia, along with standard methods of examination, performed the determination of the chemical composition of urinary stones using infrared spectroscopy. Urinary stones were classified according to the predominant mineral component into six main types (calcium-oxalate, uric acid, calcium phosphate, magnesium-ammonium-phosphate, urate–ammonium and mixed), as well as determining the nutrition stereotype for 25 nutrients using an electronic questionnaire. The classification model was built using the tools of a modern set of Data mining methods-IBM SPSS Modeler 18.0 (IBM Corporation, USA).

Results. the data set includes the results of the survey of pattern of food and chemical composition analysis of urinary stones on the basis of which a model was produced, allowing high-precision to predict in vivo chemical type of urinary stones. The constructed machine-learning model (C5.0 algorithm) has high predictive accuracy (98.6-100%), specificity (98.2-100%), and sensitivity (100% for the main types of stones and 75.0% for mixed ones).

Conclusions. the Developed method for determining the chemical composition of urinary calculus in vivo based on the indicators of the patients nutrition stereotype has a high specificity, sensitivity and accuracy, which allows using this prognostic model in clinical practice.

Read in the source