*Presented at MDS Congress 2026

BACKGROUND

Problem: Speech changes occur early in Parkinson's disease, but therapy often starts too late.

Status Quo: Subjective ratings are the gold standard, but they sometimes reach their limits.

Innovation: Digital speech analysis as an objective, sensitive, and robust tool for early detection.

METHOD

Recording booth vs. living room? Assessing the technical equivalence and scalability of speech recordings in real-world settings.

DATA PROCESSING & ANALYSIS

Measures quantifying audio recording quality and noise characteristics were extracted using SIGMA, ki:elements’ proprietary speech processing pipeline. 

Cross-setup robustness was evaluated via paired t-tests (Table 1), with effect sizes shown in Figure 1. Outliers beyond ± 2.5 SD were excluded.

CONCLUSIONS

Acoustics: Smartphone auto-processing suppresses low frequency background noise and prevents clipping, but increases broadband noise compared to professional equipment.

Scalability: Smartphone apps provide a reliable, device-independent solution for remote speech monitoring at scale without losing information.