Would Artificial Neural Networks Implemented in Clinical Wards Help Nephrologists in Predicting Epoetin Responsiveness?

Luca Gabutti; Nathalie Lötscher; Josephine Bianda; Claudio Marone; Giorgio Mombelli; Michel Burnier

Disclosures

BMC Nephrology 

In This Article

Conclusion

Compared with linear models, computer assisted tools based on artificial neural networks predict the mean erythropoietin dose required for an individual patient significantly better. Surprisingly using non linear correlations the most important variable influencing the epoetin requirement expressed in units/kg/week is the weight. Even if a particular attention should be reserved to both the pre-dialysis pH and the epoetin administration route, a prediction tool built with other variables known to influence the epoetin responsiveness will not increase in a quantitatively significant way the prediction power of the model.

As expected the model built to predict the dose adjustments is mainly influenced by the historical haemoglobin levels and even using only the data of the previous 2 months would allow compared with nephrologists, with a specificity of 92%, to detect a further 23% of the patients treated with an insufficient dose of erythropoietin.

Thus, implementing computer assisted tools that help predict the ideal erythropoietin dose, allowing timely and appropriate prescription adjustments, is an important challenge that will have relevant consequences on the patients' quality of life and should be further encouraged.

Pre-publication History

The pre-publication history for this paper can be accessed here: https://www.biomedcentral.com/1471-2369/7/13/prepub

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.

processing....