Voice Biomarker Hypoglycemia Detection

Type: diagnostic

Status: Research

Developer: University of Bern

Breakthrough Summary

University of Bern study: ML model detects hypoglycemia from 30-second voice recordings with clinically significant accuracy. 540 recordings analyzed. Could enable passive monitoring via smartphone.

Mechanism of Action

ML model detects hypoglycemia from voice pitch, jitter, and shimmer changes in 30-second recordings

Year: 2026