KEY TERMS AND DEFINITIONS

End Stage Renal Disease:

Stage 5 of Chronic Kidney Disease synonymous with Glomerular Filtration Rate of less than 15 mL/min/1.73m². The patient is close to or already included in renal replacement therapy (hemodialysis or peritoneal dialysis.

Anemia:

Qualitative or quantitative deficiency of hemoglobin, a molecule found inside red blood cells. Since hemoglobin carries oxygen from the lungs to the tissues, anemia leads to hypoxia (lack of oxygen) in organs.

Erythropoietin:

Glycoprotein hormone that estimulates erythropoiesis, i.e. red blood cell production. Erythropoietin is a protein present in human blood whose characteristics involve an inverse relationship between the Hemoglobin (Hb) concentration and the EPO concentration in blood. Therefore, the use of EPO and the associated diminution of its concentration in blood, involves an increase of the Hb concentration. This is key reason that justifies the relevance of EPO in anaemia treatment.

Neural Networks:

Non-linear processing models that carry out a mapping of an input space onto an output space using a set of parameters (synaptic weights). Synaptic weights vary their value according to a certain learning algorithm.

Multilayer Perceptron:

It is the most widely used neural network. It can be applied to modeling, classification and time series problems. Its name comes from the peculiar arrangement of neurons in different layers.

Self-Organizing Map:

Neural network mainly used for visualization purposes. It carries out a non-linear processing whose result is the low (usually two) dimensional representation of high dimensional data while preserving topological relationships among data in the high dimensional space.

Clustering:

Procedure targeted at finding prototypes of the data in those areas where there is a (relatively) huge density of data. These areas are known as clusters. Usually clusters are separated each other by areas with a (relatively) low density of data. It can be used to reduce the number of patterns as well as the dimensionality in data mining problems.