Artificial intelligence models to stratify cardiovascular risk in incident hemodialysis patients

End stage renal disease condition increases the risk of cardiovascular disease. The mortality rates among hemodialysis patients are 20% higher than the general population, thus in recent years the preservation of the cardiovascular system has become a major point of focus for nephrology care in patients. Cardiovascular events jeopardize the life of a dialysis patient and must therefore be prevented. The aim of this study is to develop forecast models that can predict the cardiovascular outcome of incident hemodialysis (HD) patients. Data relating to the treatment methods and the physiological condition of patients was collected during the first 18 months of renal replacement therapy and then used to predict the insurgence of cardiovascular events within a 6-month time window. Information regarding 4246 incident hemodialysis patients was collected. A Lasso logistic regression model and a random forest model were developed and used for predictive comparison. Every forecast model was tested on 20% of the data and a 5-fold cross validation approach was used to validate the random forest model. Random forest showed higher performance with AUC of the ROC curve and sensitivity higher than 70% in both the temporal windows models, proving that random forests are able to exploit non-linear patterns retrieved in the feature space. Out of bag estimates of variable importance and regression coefficients were used to gain insight into the models implemented. We found out that malnutrition and an inflammatory condition strongly influence cardiovascular outcome in incident HD patients. Indeed the most important variables in the model were blood test variables such as the total protein content, percentage value of albumin, total protein content, creatinine and C reactive protein. Age of patients and weight loss in the first six months of renal replacement therapy were also highly involved in the prediction. A greater understanding of the mechanisms involved in the insurgence of cardiovascular events in dialysis patients can ensure physicians to intervene in the appropriate manner when a high-risk cardiovascular condition is identified.

► Prevention of cardiovascular system disruption is one of major focuses in nephrology.
► Forecast models of cardiovascular events in dialysis patients were built.
► Among the built models random forests showed the best predictive performance.
► Albumin, C-reactive protein and age of patients are strong predictors of events.
► Obtained results are better than what has already been published in literature.