Abstract

MORALES MUNOZ, Luis; QUINTANA, Gerardo  and  NINO, Luis Fernando. Computational model for the identification of endophenotypes and classification of rheumatoid arthritis patients from genetic, serological, and clinical data using computational intelligence techniques. Rev.Colomb.Reumatol. [online]. 2015, vol.22, n.2, pp.90-103. ISSN 0121-8123.  https://doi.org/10.1016/j.rcreu.2015.05.005.

Objective: To use computational intelligence models for the classification and identification of endophenotype (relationships between phenotype and genetic markers) in patients with rheumatoid arthritis and healthy controls from genetic information, primarily the DRB1 HLA (human leukocyte antigen) and the shared epitope theory. This refers to the association between rheumatoid arthritis and the HLA-DRB1 alleles mainly containing amino acid common motif sequences QKRAA, RRRAA, QRRAA or at positions 70 to 74 DRB1 chain, which have been associated to susceptibility of this disease. Methods: Computational models were developed for classification using computational intelligence techniques, such as neural networks, Bayesian networks, and methods such as k-means. The input data consisted of variables such as: rheumatoid factor, anti-citrullinated protein antibody, C-reactive protein, number of swollen and tender joints, morning stiffness, age, gender, history of comorbidities, and the information on the HLA-DRB1. Bioinfor-matics techniques were used to search for amino acid sequences related to disease severity. Results: Promising results for the diagnosis of the disease were obtained, as well as its categorisation as potential application in personalised medicine for individuals suffering from this disease. Computational models were designed for the classification, in which the neural network using 5 variables obtained a sensitivity of 92.3% with a specificity of 86.66%, and the Bayesian network obtained a sensitivity of 92.3% and a specificity of 93.33%. The neural network using 11 variables had a sensitivity of 84.61% and a specificity of 93.33%, while the Bayesian network obtained a sensitivity of 92.3% with a specificity of 93.33%. K-means clustering method type was used to divide all patients and controls into two groups of data. It also managed to obtain two patient groups to define severity. Finally, a tree distance was obtained between amino acid sequences of the different alleles HLA DRB1, which allows genetic proximity groups to be visualised and to determine and ensure that there are may be more groups outside the proposed different theories. Conclusion: The proposed method can be used to provide better stratification of the disease in relation to the predicted phenotypes, and the potential for primary prevention of this disease.

Keywords : HLA human leukocyte antigen; Epitope shared; Computational intelligence; Rheumatoid arthritis.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )