Related Experiment Video
Updated: Jul 8, 2026
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
An Electronic Health Record Algorithm's Performance to Identify Cognitive Impairment in Primary Care.
Christine E Kistler1, Molly E Lynch2, Feng-Chang Lin3
1Division of Geriatric Medicine, School of Medicine, University of Pittsburgh, Pittsburgh, PA, USA; Department of Family Medicine, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA.
An electronic health record (EHR) algorithm effectively identifies older adults with dementia or mild cognitive impairment in primary care settings. The algorithm demonstrated high specificity and acceptable sensitivity, though International Classification of Diseases, 10th Revision (ICD-10) codes were often nonspecific.
Area of Science:
- Geriatric Medicine
- Health Informatics
- Cognitive Neurology
Background:
- Primary care settings are crucial for diagnosing and managing dementia and mild cognitive impairment (MCI).
- Electronic health records (EHRs) offer potential for identifying patients with cognitive decline.
- Accurate identification of dementia/MCI in primary care is essential for timely intervention.
Purpose of the Study:
- To evaluate the performance of an EHR algorithm designed to detect dementia or MCI in older adults within primary care.
- To assess the algorithm's accuracy using International Classification of Diseases, 10th Revision (ICD-10) codes and prescribed dementia medications.
Main Methods:
- Retrospective cohort study involving 525 adults aged 65+ from 59 primary care clinics.
- Dementia/MCI presence confirmed by trained staff using gold standard chart review.
- Algorithm evaluated using 114 cognitive impairment-related ICD-10 codes and dementia medications.
Main Results:
- The cohort (n=525) was 59% female, 81% White, with an average age of 74.5 years.
- Dementia or MCI was identified in 8.6% (n=45) of the cohort.
- The algorithm achieved 98.1% specificity, 60.0% sensitivity, 0.75 positive predictive value, and 0.96 negative predictive value (F-1 score: 0.667).
Conclusions:
- An optimized EHR algorithm combining ICD-10 codes and dementia medications shows high specificity and acceptable sensitivity for identifying dementia/MCI in primary care.
- Nonspecific ICD-10 codes, particularly 'unspecified dementia,' were frequently used, highlighting a limitation.
- Further refinement is needed to improve the specificity of diagnostic coding within EHR systems for cognitive impairment.
