Related Experiment Video
Updated: Jan 13, 2026

A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
Verification of Efficacy of Cognitive Function Test Using the Cognitive Impairment Screening Test
Hyuk Sung Kwon1, Jeewon Suh2, Geun U Park3
1Department of Neurology, Hanyang University Guri Hospital, Hanyang University College of Medicine, Guri, Korea.
Background And Purpose:
The Cognitive Impairment Screening Test (CIST) was developed to detect early-stage cognitive impairment. The clinical utility of this test has not yet been validated in real-world settings. This study compared the diagnostic performance of the CIST with that of the previously used screening test (PST).
Methods:
We used research data from the ANSYS database of the National Institute of Dementia, Republic of Korea. Individuals who underwent screening tests between 2017 and 2023 were analyzed. Scores were compared among participants stratified by age, sex, and education duration. The discriminative performance of both screening tests in detecting cognitive impairment was assessed. The equipercentile equating method was used to enable score conversion between the CIST and PST.
Results:
In total, 7,492,613 individuals (CIST: 3,195,129; PST: 4,297,484) were analyzed. The average screening score was higher for PST than for CIST. A larger proportion of cognitively impaired individuals was identified by the CIST (12.68%) than by the PST (10.72%, p<0.001). The CIST score (area under the receiver operating characteristic curve [AUC]=0.790, 95% confidence interval [CI]=0.70-0.850) outperformed the PST score (AUC=0.776, 95% CI=0.700-0.850) in predicting cognitive impairment based on diagnostic tests. CIST scores of 20, 15, and 5 were found to correspond to PST scores of 24, 20, and 10, respectively.
Conclusions:
This study supports the validity and utility of the CIST as a screening tool for cognitive impairment. The CIST outperformed the PST in predictive accuracy. However, direct score conversion between the two tests should be approached with caution.

