Using Machine Learning, Speech Production to Quantify Mild Cognitive Impairment: Alvaro Pascual-Leone, MD, PhD

WATCH TIME: 5 minutes

"If you want to capture overall cognitive performance with the greatest sensitivity, you don’t want to use only one metric. You want to capture multiple domains. That’s why the idea is to use multimodal digital biomarkers, not just voice and the hundreds of metrics that can be extracted from it."

Patients with mild cognitive impairment (MCI) are at risk of conversion to dementia, and thus, classifying their cognitive status may be beneficial in guiding treatment and prevention strategies. A study presented at the 2022 Alzheimer’s Association International Conference, July 31 to August 4, in San Diego, California, examined if speech production metrics—frequency, loudness, and voice quality—can increase the accuracy of classifying cognitively healthy individuals and MCI subtypes.¹

The study included several cognitive assessments such as the Rey Auditory Verbal Learning Test, Mini-Mental State Examination, Functional Activities Questionnaire, Digital Clock Test, Digital Clock and Recall, and physician diagnosis. With the inclusion of speech and voice analysis, the classification accuracy was improved, with 4 cognitive statuses identified: HC – cognitively impaired (CI), HC – probably Alzheimer disease (AD), HC-MCI, and MCI-AD.

To learn more about the background of the study and its direct clinical implications, NeurologyLive^® sat down with senior investigator Alvaro Pascual-Leone, MD, PhD. Pascual-Leone currently serves as the medical director of the Deanna and Sidney Wolk Center for Memory Health and a senior scientist at the Hinda and Arthus Marcus Institute for Aging Research at Hebrew SeniorLife, as well as a professor of neurology at Harvard Medical School. He also answered questions about the use of speech production metrics in the AD field, and whether there needs to be more attention given to these approaches.

Click here for more coverage of AAIC 2022.

REFERENCE
1. Banks R, Jannati A, Gomes-Osman J, et al. Enhanced machine learning classification of mild cognitive impairment with multimodal digital biomarkers. Presented at: 2022 Alzheimer’s Association International Conference; July 31-Aug 4; San Diego, CA. 60485