多准则决策分析
医学
排名(信息检索)
成对比较
风湿病
统计
机器学习
计算机科学
数学
运筹学
内科学
作者
Sara K. Tedeschi,Sindhu R. Johnson,Dimitrios T. Boumpas,David Daikh,Thomas Dörner,Betty Diamond,Søren Jacobsen,David Jayne,Diane L. Kamen,W. Joseph McCune,Marta Mosca,Rosalind Ramsey‐Goldman,Guillermo Ruiz‐Irastorza,Matthias Schneider,Murray B. Urowitz,David Wofsy,Josef S. Smolen,Raymond P. Naden,Martin Aringer,Karen H. Costenbader
标识
DOI:10.1136/annrheumdis-2018-214685
摘要
European League Against Rheumatism and are jointly supporting multiphase development of systemic lupus erythematosus (SLE) classification criteria based on weighted criteria and a continuous probability scale. Prior steps included item generation, item reduction and hierarchical organisation of candidate criteria using an evidence-based approach. Our objectives were to determine relative weights using multicriteria decision analysis (MCDA) and to set a provisional threshold score for SLE classification. An SLE Expert Panel (8 European, 9 North American) submitted 164 real, unique cases with a wide range of SLE probability in a standardised format. Using the candidate criteria, experts scored and rank-ordered 20 representative cases. At an in-person meeting, experts reviewed inter-rater reliability of scoring, further refined criteria definitions and participated in an MCDA exercise. Based on expert consensus decisions on pairwise comparisons of criteria, 1000minds software calculated criteria weights and rank-ordered the remaining 144 cases based on their additive scores. The score of the lowest-ranked case for which complete expert consensus was achieved defined the provisional threshold classification score. Inter-rater reliability of scoring cases with the candidate criteria was good. MCDA involved 74 pairwise decisions and was repeated for the arthritis and mucocutaneous domains when the initial ranking of some cases did not match expert opinion. After criteria weights and additive scores were recalculated once, experts reached consensus for SLE classification for all cases scoring>83. Using an iterative process, the candidate criteria definitions were refined, preliminary weights were calculated and a provisional threshold score for SLE classification was determined.
科研通智能强力驱动
Strongly Powered by AbleSci AI