A0594
Title: Assessing intra- and inter-method agreement of functional data
Authors: Jeong Hoon Jang - Yonsei University (Korea, South) [presenting]
Abstract: Modern medical devices increasingly produce functional data whose sampling unit is a smooth continuous function defined over a time/spatial domain. A series of intraclass correlation coefficient (ICC) and concordance correlation coefficient (CCC) indices that can evaluate the reliability and reproducibility of medical devices producing functional data are proposed. Specifically, two versions of ICC and CCC indices are introduced. The first version consists of time-dependent ICC and CCC indices that can quantify the degrees of intra-method, inter-method and total (intra+inter) agreement that vary smoothly over time. The second version denotes their global counterparts, summarising agreement over the entire dime domain using a single measure. The proposed indices are formulated based on a multivariate multilevel functional model that represents indices in terms of truncated multivariate Karhunen-Loeve expansions, whose terms can be smoothly estimated by functional principal component analysis. Extensive simulation studies are performed to assess the finite-sample properties of the estimators. The proposed method is applied to Emory renal study data to evaluate the reliability and reproducibility of renogram curve data produced by a high-tech radionuclide image scan used to detect kidney obstruction non-invasively.
