Accuracy of DCTclock to Detect Cognitive Impairment: Ali Jannati, MD, PhD

WATCH TIME: 4 minutes

"From other efforts in artificial intelligence, we know that if there is limitation in the representativeness of the original datasets that researchers use to build these AI models, they can be bias when they want to be applied to other cultural factors, educational factors that might vary across the different sections of the populations.”

FDA-listed, the DCTclock is a digital version of the traditional paper-based clock drawing test that identifies cognitive impairment before clinical symptoms arise. Developed by Linus Health, the test prompts participants to both draw from memory and copy an analog clock face using either a digital pen with paper or mobile device. By capturing the entire process of task performance, rather than solely the end product, the DCTclock takes into account the cognitive function of an individual performing the test.

At the 2023 American Academy of Neurology (AAN) Annual Meeting, held April 22-27, in Boston, Massachusetts, a presentation looked at the reliability, validity, learning effects, and psychometrics of the DCTclock. Co-author Ali Jannati, MD, PhD, and others at Linus Health compared multiple datasets of cognitively normal (CN) and cognitively impaired (CI) individuals, with DCTclock scores showing good test-retest reliability (r = 0.8) in CN and CI participants, and an independent CN sample (r = 0.7). Overall, clock-element classification accuracy was 99.6% and 99.3% for CN and CI groups, respectively.

To learn more about the analysis, NeurologyLive^® sat down with Jannati, manager, Clinical Development-Research, Linus Health. Jannati provided insight on the significance of the data, as well as why validation across large cohorts supports the clinical effectiveness of the DCTclock.

Click here for more coverage of AAN 2023.

REFERENCE
1. Helbig T, Toro-Serey C, Thompson K, et al. Reliability of tablet-based Digital Clock Drawing Task (DCTclock) for automated detection of cognitive impairment. Presented at: 2023 AAN Annual Meeting; April 22-27; Boston, MA.