Prediction of Change in Overall Performance for Patients with Huntington's Disease (HD) Using a Multilevel Functional Principal Component Analysis (MFPCA) Model

• Activity Number: 265
• Type: Contributed
• Date/Time: Monday, August 10, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Mental Health Statistics Section
• Author(s): Zhi Pan* and Yuanjia Wang and Jeff Goldsmith
• Companies: and Columbia University and Columbia University
• Keywords: MFPCA ; Huntington's disease ; cognitive evaluation ; overall performance prediction

Abstract

Huntington's disease causes progressive cognitive & motor impairment along with behavioral & psychiatric disorders. Our purposes are to determine association among cognitive tests and to predict overall performance of patients using key signals extracted from cognitive tests. To address the objective, MFPCA is applied to assess association of five major cognitive tests: symbol digit modalities test (SDMT), three STROOP tests (color, word & interference) and frontal system behavior scale. Then PC scores of MFPCA are used to predict change in overall performance monitored through scoring total function capacity (TFC). Education year & baseline age are added to adjust prediction model. Our results show MPFCA integrates information from all of subjects and cognitive tests to capture two main modes of variation at subject & test levels. The 1st two PC scores at both levels are used as predictors for they are able to shrink high dimensional data while retaining most original cognitive information. Furthermore, MPFCA is a better methodology in HD prediction than standard benchmark analysis, and adjusted PC scores related to SDMT & STROOP color tests have significant impact on TFC change.