| Issue/Objective |
Mild Cognitive Impairment (MCI) represents an early stage of cognitive decline, making accurate and accessible diagnostic tools essential. Speech analysis has emerged as a promising, non-invasive approach for detecting subtle cognitive changes. This review aimed to synthesize the evidence on the accuracy of speech-based biomarkers for identifying MCI. |
| Methodology/Approach |
A systematic review and meta-analysis were conducted following the PRISMA guidelines. PubMed, Scopus, Ovid Medline, and PsycINFO databases were searched up to March 2025 without restrictions on language, article status, or year. |
| Results |
Of the 55 included articles, meta-analysis of 24 reporting Area Under the Curve (AUC) values revealed a pooled estimate of 76.0% for distinguishing MCI from cognitively unimpaired individuals. Meta-analysis of 18 articles reporting accuracy values showed a pooled estimate of 78.0%. For both AUC and accuracy analyses, Egger's regression tests indicated no publication bias (p≥ 0.748), and the I² statistic indicated no heterogeneity across studies (I²= 0.00%, p= 1.00). Five studies examined individuals with subjective cognitive decline, reporting significant differences in certain speech features compared to control peers. |
| Discussion/Conclusion |
Speech analysis demonstrates a good ability to distinguish MCI from control peers based on lexical and acoustical parameters. Speech-based biomarkers represent a potentially valuable, accessible, and cost-effective tool for early cognitive screening. Further research is needed to standardize protocols, validate findings in diverse populations, and explore the utility of speech analysis for detecting cognitive decline at even earlier stages. |
| Presenters and affiliations |
Zahra Jafari Dalhousie University |
