Addressing Diversity with Abnorms: Implications For Detecting Mild Cognitive Impairment and Alzheimer's Disease
Date of Award
Campus Access Dissertation
Doctor of Philosophy (PhD)
Objective: Normative bases in clinical neuropsychological are limited when working within demographically diverse clinical settings by their inflexibility, that is, specificity to the location, time, and demographic and clinical variables accounted for at the time of norming. This feasibility study addresses this need with a novel protocol that retrieves neuropsychological test data from a hospital electronic medical record (EMR) to create a clinical case series of patients with mild cognitive impairment (MCI) who were found eventually to convert to Alzheimer's disease. Patients with the same referral diagnosis are compared to the case series “abnorms," while allowing practicing neuropsychologists to match the patient with nuanced demographic variables in the reference group, such as country of origin. This protocol is intended to improve diagnostic clarity for diverse patients who are poorly represented in available norms.
Participants and Methods: Logistic regression models (PCA, LASSO, and ridge logistic regression) were compared on N = 66 patients’ neuropsychological reports from 2015 to 2017 were retrieved from the EMR of a Boston-area hospital. Three groups are represented in the sample: MCI patients (N=22) who were confirmed in the EMR to convert to AD at least a year later; AD patients (N = 11); and controls (N = 33) referred for memory complaints but never converted. The total sample was separated into two conditions, in which MCI was first compared with general controls, and then compared with general controls combined with the AD subjects. Due to lack of a sufficient sample size for Spanish-speaking patients in the total sample, an English-speaking cohort was selected for the analysis to test the general applicability of this protocol.
Results: The PCA model outperformed LASSO and ridge logistic regression in terms of accuracy and AUC with 10-fold cross validation across conditions and levels of cross validation. Sensitivity and specificity was consistently above the .8 cutoff for the PCA model only.
Conclusions: Abnorms represent a promising direction for modifying standard practices in neuropsychological assessment. In addition to this protocol, novel procedures are yet to be developed to make clinical use of this flexible and extensive reference base to improve clinical care.
Gable, Samuel C., "Addressing Diversity with Abnorms: Implications For Detecting Mild Cognitive Impairment and Alzheimer's Disease" (2019). Graduate Doctoral Dissertations. 495.
Free and open access to this Campus Access Dissertation is made available to the UMass Boston community by ScholarWorks at UMass Boston. Those not on campus and those without a UMass Boston campus username and password may gain access to this dissertation through resources like Proquest Dissertations & Theses Global or through Interlibrary Loan. If you have a UMass Boston campus username and password and would like to download this work from off-campus, click on the "Off-Campus UMass Boston Users" link above.