Diagnosis of lupus nephritis via urine samples by NMR Spectroscopy using genetic algorithm (GA) - back-propagation artificial neural network (BP-ANN)

Lupus is a chronic inflammatory and complex disease associated with many clinical symptoms involving several organs, including joints, skin, heart, nervous system, lungs, blood vessels, and kidneys. Symptoms of lupus can be observed in all parts of the body, especially in the skin and joints. At progressed stages, it affects the kidneys and causes inflammations. This disease has been reported in males and females of all races and ages from children to adults. More than ۹۰% of these patients are women; therefore, it seems that estrogen metabolism and its relation with the immune system may play a crucial role in the increased rate of this disease in women [۱]. One of the most common and widely used spectroscopic techniques for metabolomics is nuclear magnetic resonance (NMR) spectroscopy. Therefore, in this study, NMR spectroscopic method in combination with chemometric tools has been proposed as a rapid and accurate classification and to discriminate urine samples of the lupus with from those of healthy samples [۲].The dataset of ۵۷ samples was split into two subsets as calibration and validation sets through randomly. ۴۶ samples were used to construct the calibration model and ۱۱ samples were used for the test set. Also, the genetic algorithm (GA) as the variable reduction was employed to develop the diagnostic model. Urine samples were collected from Imam Khomeini Hospital, Tehran, Iran. All patients were well informed about our investigation aim and voluntarily shared in search tests.The results are summarized in Table۱. Hence, the proposed method is an accurate, reliable, rapid, non-destructive, non-invasive diagnostic and suitable for the diagnosis of lupus nephritis using NMR spectroscopic method along chemometric method based on their biomarkers.