OP0339 Development of a predictive model of radiological damage in patients with rheumatoid arthritis based on artificial intelligence. (12th June 2018)

• Record Type: Journal Article
• Authors: Lezcano, J.M.; Ivorra-Cortes, J.; López Mejías, R.; Martín, J.; Fernandez-Gutiérrez, B.; González-Gay, M.A.; Balsa, A.; González-Álvaro, I.; Salazar, F.; Abásolo, L.; Rodriguez-Rodriguez, L.
• Is Part Of: Annals of the rheumatic diseases. Volume 77(2018)Supplement 2
• Page Start: 215
• Page End: 215
• Publication Date: 2018-06-12
• Subjects: Rheumatism -- Periodicals; 616.723005
• DOI: 10.1136/annrheumdis-2018-eular.6801
• Languages: English
• ISSNs: 0003-4967

Abstract

Abstract : Background: Rheumatoid arthritis (RA) is a chronic inflammatory disease associated with increased mortality and disability. Although different factors associated with prognosis have been identified, it is still difficult to predict the evolution of a specific patient. Objectives: Our objective is to train and validate a predictive model of disease severity using radiological damage as a surrogate marker, based on Artificial Intelligence techniques, and using clinical and genetic data. Methods: Four independent cohorts were included (892 patients with 1667 hand X-rays). Radiological damage was measured with the Sharp/van-der-Heijde score (SvdH). The variables to be predicted [total value of SvdH, erosion component (ES) and joint narrowing (NS)] were logarithmically transformed. As clinical predictors, age at onset of symptoms, sex, duration of the disease at the time of each radiograph, year of onset of symptoms and presence of rheumatoid factor were used. As genetic variables, the single nucleotide polymorphism data obtained from the Immunochip genotyping platform (Illumina) were used. In addition, an interaction between each polymorphism and the duration of the disease was introduced. Three cohorts were used for the selection of variables, generation of predictive models and internal validation. The fourth cohort was used to perform the external validation of the models. Regression trees with random effects were generated using the R package 'REEMtree'. The goodness of fit of the models was measured using the root mean squared error (RMSE) and the intraclass correlation coefficient (ICC). Results: In the cohorts where the predictive models were developed, the RMSEs for total SvdH, ES and NS were 3.16, 1.02 and 2.29 units of the Sharp/van-der-Heijde score, respectively. The ICCs were 0.96, 0.87 and 0.95, respectively. In the external validation cohort, the RMSEs were 7.13, 3.53 and 4.81 units of the Sharp/van-der-Heijde score, respectively. The ICCs were 0.90, 0.78 and 0.88. For the total SvdH, the best fit model contained the variables 'age of onset of the symptoms of RA' and the interaction between duration of the disease and 3 polymorphisms: rs10752907, rs4405161 and rs2501617. For the ES, it contained the variables 'age of onset of AR symptoms', the polymorphism rs7769752 and the interaction between disease duration and 6 polymorphisms: rs12410412, rs117029499, rs72925969, rs869186, rs11258464, rs4781952. For the NS, it contained the variables 'age of onset of AR symptoms', 'gender', and the interaction between disease duration and 9 polymorphisms: rs3814055, rs1020822, rs13157991, rs152294, rs2914190, rs10824537, rs2637229, rs114136906 and rs4958241. Conclusions: It is possible to generate predictive models of radiological damage of great precision using Artificial Intelligence techniques. This could allow early stratification of patients according to prognosis. It is necessary to validate these models in other populations. Disclosure of Interest: None declared