Listening to Hypoglycemia: Voice as a Biomarker for Detection of a Medical Emergency Using Machine Learning
Authors' conclusion
An ML approach exclusively based on voice data allows for noninvasive hypoglycemia detection, corroborating the potential of ML-based approaches to infer acute health states through voice.
Study context
- Population
- People with T1D; 11 female, age 37.3 ± 12.4 years, HbA1c 7.1 ± 0.5%; 540 smartphone voice recordings
- Participants
- 22 people540 recordings
- Comparison
- Machine learning (ML) voice-based detection of hypoglycemia by speech task
- Outcome
- Noninvasive detection of hypoglycemia from voice features (AUC)
Methodological detailsmachine learning, two sequential clinical studies, standardized euglycemia and hypoglycemia
- Machine learning
- Two sequential clinical studies
- Standardized euglycemia and hypoglycemia
Performance
As reported in the abstract
Hypoglycemia detection, AUC
| Test or model | Value | AUC |
|---|---|---|
| Reading text aloud | 0.90 ± 0.12 | 0.90 ± 0.12 |
| Diadochokinetic task | 0.87 ± 0.15 | 0.87 ± 0.15 |
Value
Reading text aloud0.90 ± 0.12
Diadochokinetic task0.87 ± 0.15
AUC
Reading text aloud0.90 ± 0.12
Diadochokinetic task0.87 ± 0.15
Original publication
Listening to Hypoglycemia: Voice as a Biomarker for Detection of a Medical Emergency Using Machine Learning. Diabetes Care. Published 18 November 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →