An Algorithm for the Diagnosis of Behçet Disease Uveitis in Adults.
Carl P HerbortSumru OnalMiles StanfordMehmet AkmanJos W R TwiskMaarten BoersMerih OrayPinar Çakar ÖzdalSibel KadayifcilarRadgonde AmerSivakumar R RathinamRajesh VedhanayakiMoncef KhairallahYonca AkovaFatime Nilüfer YalçındağEsra KardesBerna BasarirÇigdem AltanYilmaz ÖzyazganAhmet GülPublished in: Ocular immunology and inflammation (2020)
Purpose: To develop an algorithm for the diagnosis of Behçet's disease (BD) uveitis based on ocular findings.Methods: Following an initial survey among uveitis experts, we collected multi-center retrospective data on 211 patients with BD uveitis and 207 patients with other uveitides, and identified ocular findings with a high diagnostic odds ratio (DOR). Subsequently, we collected multi-center prospective data on 127 patients with BD uveitis and 322 controls and developed a diagnostic algorithm using Classification and Regression Tree (CART) analysis and expert opinion.Results: We identified 10 items with DOR >5. The items that provided the highest accuracy in CART analysis included superficial retinal infiltrate, signs of occlusive retinal vasculitis, and diffuse retinal capillary leakage as well as the absence of granulomatous anterior uveitis or choroiditis in patients with vitritis.Conclusion: This study provides a diagnostic tree for BD uveitis that needs to be validated in future studies.
Keyphrases
- juvenile idiopathic arthritis
- ankylosing spondylitis
- machine learning
- deep learning
- optical coherence tomography
- diabetic retinopathy
- optic nerve
- disease activity
- rheumatoid arthritis
- cross sectional
- big data
- systemic lupus erythematosus
- neural network
- data analysis
- artificial intelligence
- interstitial lung disease