Nonlinear Z‐score modeling for improved detection of cognitive abnormality

Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring

Nadine Tatton¹  Susan Dickinson¹  Rodney Pearlman²  Brian Appleby³  Edward D. Huey⁴  Masood Manoochehri⁴  Jill Goldman⁴  Mario Mendez⁵  Yvette Bordelon⁵  Miranda Maldonado⁵  Giovanni Coppola⁵  Eliana Marisa Ramos⁵  Lilah Besser⁶  Jaya Padmanabhan⁷  Bradford Dickerson⁷  Bonnie Wong⁷  Diane Lucente⁷  Scott McGinnis⁷  Alexander Pantelyat⁸  Chiadi Onyike⁸  Ann Fishman⁸  Arthur Toga⁹  Brad Boeve^1,10  David Knopman^1,10  Sara Farmer^1,10  Maria Lapid^1,10  Ruth Kraft^1,10  Amy Rindels^1,10  Leah Forsberg^1,10  Christina Dheel^1,10  David Jones^1,10  Ralitza Gavrilova^1,10  Danielle Brushaber^1,10  Kejal Kantarci^1,10  Jeremy Syrjanen^1,10  Julie Fields^1,10  Walter Kremers^1,10  Jonathan Graff‐Radford^1,10  Deb Gearhart^1,10  Neill Graff‐Radford^1,11  Zbigniew Wszolek^1,11  Rosa Rademakers^1,11  Dana Haley^1,11  Len Petrucelli^1,11  Walter Kukull^1,12  Madeline Potter^1,13  Kelley Faber^1,13  Tatiana Foroud^1,13  Codrin Lungu^1,14  Margaret Sutherland^1,14  John Hsiao^1,15  Ian M. Grant^1,16  Sandra Weintraub^1,16  Emily Rogalski‐Miller^1,16  Hiroko Dodge^1,17  Patrick Brannelly^1,18  Irene Litvan^1,19  Anna Karydas^2,20  Katherine Rankin^2,20  John Kornak^2,20  Bruce Miller^2,20  Reilly Dever^2,20  Adam Boxer^2,20  Hilary W. Heuer^2,20  Peter Ljubenkov^2,20  Ping Wang^2,20  Joanne Taylor^2,20  Adam M. Staffaroni^2,20  Howard Rosen^2,20  Jamie Fong^2,20  Joel Kramer^2,20  ARTFL/LEFFTDS Consortium^2,20  Diana Kerwin^2,21  Erik D. Roberson^2,22  Emily McKinley^2,22  Ian Mackenzie^2,23  Pheth Sengdy^2,23  Robin Hsiung^2,23  Jessica Ferrall^2,24  Daniel Kaufer^2,24  Murray Grossman^2,25  Les Shaw^2,25  Katya Rascovsky^2,25  Sophia Dominguez^2,25  Jessica Bove^2,25  David Irwin^2,25  John Trojanowski^2,25  Behnaz Ghazanfari^2,26  Carmela Tartaglia^2,26  Namita Multani^2,26  Kimiko Domoto‐Reilly^2,27  Christina Caso^2,27  Lynne Jones^2,28  Nupur Ghoshal^2,28 

[1] Association for Frontotemporal DegenerationRadnorPAUSA;Bluefield Project to Cure FTDSan FranciscoCAUSA;Case Western Reserve UniversityClevelandOHUSA;Columbia UniversityNew YorkNYUSA;Department of PsychiatryDavid Geffen School of Medicine, UCLALos AngelesCAUSA;Florida Atlantic UniversityBoca RatonFLUSA;Harvard University/MGHBostonMAUSA;Johns Hopkins UniversityBaltimoreMDUSA;Laboratory of Neuroimaging (LONI), USCLos AngelesCAUSA;Mayo Clinic RochesterRochesterMNUSA;Mayo ClinicJacksonvilleFLUSA;National Alzheimer Coordinating Center (NACC), University of WashingtonSeattleWAUSA;National Cell Repository for Alzheimer's Disease (NCRAD)Indiana UniversityIndianapolisINUSA;National Institute of Neurological Disorders and Stroke (NINDS)BethesdaMDUSA;National Institute on Aging (NIA)BethesdaMDUSA;Northwestern UniversityChicagoILUSA;Oregon Health and Science UniversityPortlandORUSA;Tau ConsortiumRainwater Charitable FoundationFort WorthTXUSA;UCSDSan DiegoCAUSA;UCSFSan FranciscoCAUSA;UTSWDallasTXUSA;University of Alabama at BirminghamBirminghamALUSA;University of British ColumbiaVancouverBritish ColumbiaCanada;University of North CarolinaChapel HillNCUSA;University of PennsylvaniaPhiladelphiaPAUSA;University of TorontoTorontoOntarioCanada;University of WashingtonSeattleWAUSA;Washington UniversitySt. LouisMOUSA;

关键词: Generalized additive models;    Heterogenous variance modeling;    Neuropsychological testing scores;    Nonlinear Z‐score correction;    Shape constrained additive models;   

DOI  :  10.1016/j.dadm.2019.08.003

来源: DOAJ

【 摘 要 】

Abstract Introduction Conventional Z‐scores are generated by subtracting the mean and dividing by the standard deviation. More recent methods linearly correct for age, sex, and education, so that these “adjusted” Z‐scores better represent whether an individual's cognitive performance is abnormal. Extreme negative Z‐scores for individuals relative to this normative distribution are considered indicative of cognitive deficiency. Methods In this article, we consider nonlinear shape constrained additive models accounting for age, sex, and education (correcting for nonlinearity). Additional shape constrained additive models account for varying standard deviation of the cognitive scores with age (correcting for heterogeneity of variance). Results Corrected Z‐scores based on nonlinear shape constrained additive models provide improved adjustment for age, sex, and education, as indicated by higher adjusted‐R2. Discussion Nonlinearly corrected Z‐scores with respect to age, sex, and education with age‐varying residual standard deviation allow for improved detection of non‐normative extreme cognitive scores.

【 授权许可】

Unknown