Related Experiment Video
Updated: Aug 5, 2026
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
Curation of Mini Mental State Examination (MMSE) Scores in the VA Million Veteran Program (MVP): Applications for
Francesca V Lopez1,2,3, Margaret Gillis4, Sophia Lee4
1Research & Psychology Services, VA San Diego Healthcare System (VASDHS), San Diego, CA, USA.
Background:
Electronic health record (EHR)-linked biorepositories provide opportunities to advance epidemiological research in Alzheimer's disease (AD) and related dementias.
Objective:
Evaluate the extraction, curation, and associative validity of Mini Mental State Examination (MMSE) scores from the VA EHR for participants in the VA Million Veteran Program (MVP).
Methods:
The sample (N = 49,555; 7.4% women) included a multiethnic cohort (European [68.3%], African [20.4%], Hispanic [9.0%]) with EHR-extracted MMSE scores; 30.7% were apolipoprotein E (APOE) e4 carriers, and 25.8% had multiple scores. Linear regressions examined cross-sectional associations between e4 dosage (0, 1, 2) and first and lowest MMSE scores. MMSE scores were also evaluated against MVP dementia diagnostic algorithms in participants aged ≥65 years.
Results:
Among participants of European ancestry, there was a significant e4 dose-response relationship (ps < .001) with MMSE scores. Homozygote carriers scored lower than heterozygote carriers (Mdiff: first = -0.5; lowest = -0.9), who scored lower than non-carriers (Mdiff: first = -0.4; lowest = -0.6). Among Veterans of African and Hispanic ancestry, no dose-response relationship was observed, although e4 carriers had lower scores than non-carriers (ps ≤ .04). MMSE scores corresponded strongly with dementia case/control status across phenotypes: mild impairment on the MMSE was strongly associated with AD (odds ratio [OR] = 11.48), with more severe MMSE impairment showing stronger associations (moderate OR = 17.95; severe OR = 27.83).
Conclusion:
This study demonstrated MMSE scores can be systematically extracted and curated from the VA EHR. Findings offer a scalable framework for future studies on risk stratification, highlighting the potential for harnessing MVP to explore genetic and clinical factors contributing to cognitive and dementia outcomes in diverse samples.
