Exploring the impact of type II diabetes mellitus on voice quality.
Level 4 - case-series / case-control
Cross-sectional case-control comparison of 30 patients with type 2 diabetes and 30 matched controls
PubMed 38319369 · doi:10.1007/s00405-024-08485-4
What was done
A cross-sectional study in Iran (February 2020 to September 2023) compared 30 participants with type II diabetes mellitus (T2DM) to 30 non-diabetic controls matched by birth year across six age categories. Participants recorded speech elicitation tasks using WhatsApp on smartphones. Seven acoustic features were extracted and analyzed via Praat software: fundamental frequency, jitter, shimmer, harmonic-to-noise ratio (HNR), cepstral peak prominence (CPP), voice onset time (VOT), and formants (F1-F2). Group differences and predictors were evaluated using t-tests, two-way ANOVA, and binary logistic regression.
What was found
Significant differences were observed between T2DM and control groups for fundamental frequency, jitter, shimmer, CPP, and HNR (p < 0.05). No significant differences were detected for formants or VOT (p > 0.05). Binary logistic regression identified shimmer as the most significant predictor of T2DM group status. A significant interaction was also reported between diabetes status and age for CPP. Exact numerical means, effect sizes, and regression coefficients were not provided in the abstract.
Why it matters
This study provides exploratory evidence that type II diabetes is associated with measurable acoustic voice alterations, pointing toward potential non-invasive acoustic prescreening approaches.
Limits
The sample size is small (30 participants per group) from a single geographic setting. Audio was gathered via a messaging app with unstandardized microphones, recording environments, and audio compression. The abstract does not report specific values, effect sizes, diagnostic accuracy metrics (sensitivity, specificity, AUC), or clinical confounders such as glycemic control level, neuropathy status, or smoking history.