Fewer biomarkers can be used to predict systemic lupus erythematosus (SLE) related kidney injury. This paper presents an apriori algorithm of association rules to mine the predictive biomarkers for SLE-related kidney injury of negative proteinuria. An apriori algorithm of association rules was employed to identify biomarkers, and logistic regression analysis and spearman correlation analysis were used to evaluate the correlation between triglycerides and SLE-related kidney injury of negative proteinuria. Triglycerides were mined out by the apriori algorithm of association rules. The level of triglycerides was significantly higher, and it was an independent risk factor for SLE-related kidney injury. In the high-triglycerides group, the number of patients with SLE-related kidney injury, SLEDAI-2K, urine P-CAST, the level of blood urea nitrogen, serum creatinine, and proteinuria were increased. Triglycerides level was positively correlated with proteinuria and P-CAST and negatively correlated with albumin and IgG. The area under the ROC curve of triglycerides and triglycerides combined proteinuria was 0.72 and 0.82, respectively. Significantly, 50% of SLE-related kidney injuries of negative proteinuria could be identified by high triglycerides levels. High triglycerides level was found at the time of onset of kidney injury, and it was opposite to glomerular filtration rate. Triglycerides may be a potential marker for predicting SLE-related kidney injury, especially in SLE-related kidney injury of negative proteinuria. Triglycerides combined proteinuria could predict SLE-related kidney injury effectively.
Keywords: apriori algorithm; association rule mining; kidney injury; systemic lupus erythematosus; triglycerides.