External validation of the 2-year mortality prediction tool in hemodialysis patients developed using a Bayesian network.

Background: In recent years, a number of predictive models have appeared to predict the risk of medium-term mortality in hemodialysis patients, but only one, limited to patients aged over 70 years, has undergone sufficiently powerful external validation. Recently, using a national learning database and an innovative approach based on Bayesian networks and 14 carefully selected predictors, we have developed a clinical prediction tool to predict all-cause mortality at 2 years in all incident hemodialysis patients. In order to generalize the results of this tool and propose its use in routine clinical practice, we carried out an external validation using an independent external validation database. Methods: A regional, multicenter, observational, retrospective cohort study was conducted to externally validate the tool for predicting 2-year all-cause mortality in incident and prevalent hemodialysis patients. This study recruited a total of 142 incident and 697 prevalent adult hemodialysis patients followed up in one of the eight Association pour l'Utilisation du Rein Artificiel dans la région Lyonnaise (AURAL) Alsace dialysis centers. Results: In incident patients, the 2-year all-cause mortality prediction tool had an area under the receiver curve (AUC-ROC) of 0.73, an accuracy of 65%, a sensitivity of 71% and a specificity of 63%. In prevalent patients, the performance for the external validation were similar in terms of AUC-ROC, accuracy and specificity, but was lower in term of sensitivity. Conclusion: The tool for predicting all-cause mortality at 2 years, developed using a Bayesian network and 14 routinely available explanatory variables, obtained satisfactory external validation in incident patients, but sensitivity was insufficient in prevalent patients.

Factors associated with kalemia in renal disease.

BACKGROUND: International recommendations promote a strict potassium diet in order to avoid hyperkalemia in chronic kidney disease (CKD) patients. However, the efficiency of such a dietary recommendation has never been demonstrated. The objectives of this study were to define the relationship between kalemia, dietary potassium intake estimated by kaliuresis and renal function, and to define the factors associated with kalemia in patients using artificial intelligence. METHODS: To this extent, data from patients followed in a nephrology unit, included in the UniverSel study and whose kalemia (measured on the day of urine collection; n = 367) were analyzed. RESULTS: The patients included had a wide range of estimated glomerular filtration rate (eGFR), but few had stage 5 CKD. Kalemia was negatively and linearly correlated to eGFR (P < .001) but was not correlated to kaliuresis (P = .55). Kaliuresis was not correlated to eGFR (P = .08). Factors associated with kalemia were analyzed using a Bayesian network. The five variables most associated with kalemia were, in descending order, eGFR, original nephropathy, age, diabetes and plasma bicarbonate level. CONCLUSION: The results of this study do not support a strict dietary potassium control to regulate kalemia in stage 1-4 CKD patients.

Prediction of all-cause mortality for chronic kidney disease patients using four models of machine learning.

BACKGROUND: The prediction tools developed from general population data to predict all-cause mortality are not adapted to chronic kidney disease (CKD) patients, because this population displays a higher mortality risk. This study aimed to create a clinical prediction tool with good predictive performance to predict the 2-year all-cause mortality of stage 4 or stage 5 CKD patients. METHODS: The performance of four different models (deep learning, random forest, Bayesian network, logistic regression) to create four prediction tools was compared using a 10-fold cross validation. The model that offered the best performance for predicting mortality in the Photo-Graphe 3 cohort was selected and then optimized using synthetic data and a selected number of explanatory variables. The performance of the optimized prediction tool to correctly predict the 2-year mortality of the patients included in the Photo-Graphe 3 database were then assessed. RESULTS: Prediction tools developed using the Bayesian network and logistic regression tended to have the best performances. Although not significantly different from logistic regression, the prediction tool developed using the Bayesian network was chosen because of its advantages and then optimized. The optimized prediction tool that was developed using synthetic data and the seven variables with the best predictive value (age, erythropoietin-stimulating agent, cardiovascular history, smoking status, 25-hydroxy vitamin D, parathyroid hormone and ferritin levels) had satisfactory internal performance. CONCLUSIONS: A Bayesian network was used to create a seven-variable prediction tool to predict the 2-year all-cause mortality in patients with stage 4-5 CKD. Prior to external validation, the proposed prediction tool can be used at: https://dev.hed.cc/?a=jpfauvel&n=2022-05%20Modele%20Bayesien%2020000%20Mortalite%207%20variables%20Naif%20Zou%20online(1).neta for research purposes.

Prediction Tool to Estimate Potassium Diet in Chronic Kidney Disease Patients Developed Using a Machine Learning Tool: The UniverSel Study.

There is a need for a reliable and validated method to estimate dietary potassium intake in chronic kidney disease (CKD) patients to improve prevention of cardiovascular complications. This study aimed to develop a clinical tool to estimate potassium intake using 24-h urinary potassium excretion as a surrogate of dietary potassium intake in this high-risk population. Data of 375 adult CKD-patients routinely collecting their 24-h urine were included to develop a prediction tool to estimate potassium diet. The prediction tool was built from a random sample of 80% of patients and validated on the remaining 20%. The accuracy of the prediction tool to classify potassium diet in the three classes of potassium excretion was 74%. Surprisingly, the variables related to potassium consumption were more related to clinical characteristics and renal pathology than to the potassium content of the ingested food. Artificial intelligence allowed to develop an easy-to-use tool for estimating patients' diets in clinical practice. After external validation, this tool could be extended to all CKD-patients for a better clinical and therapeutic management for the prevention of cardiovascular complications.