Voice-based algorithm predicts type 2 diabetes in USA adults
Author Information
Author(s): Elbéji Abir, Pizzimenti Mégane, Aguayo Gloria, Fischer Aurélie, Ayadi Hanin, Mauvais-Jarvis Franck, Riveline Jean-Pierre, Despotovic Vladimir, Fagherazzi Guy
Primary Institution: Luxembourg Institute of Health
Hypothesis
Can a voice-based algorithm effectively predict type 2 diabetes status in adults?
Conclusion
The study suggests that voice-based algorithms could serve as a more accessible and non-invasive screening tool for type 2 diabetes.
Supporting Evidence
- The voice-based algorithms demonstrated good predictive capacity with AUC scores of 75% for males and 71% for females.
- Performance improved in females aged 60 years or older and individuals with hypertension.
- The algorithms showed over 93% agreement with the American Diabetes Association risk score.
Takeaway
This study shows that we can use voice recordings to help find out if someone has type 2 diabetes, which could make testing easier and less painful.
Methodology
The study analyzed voice recordings from 607 participants, using machine learning algorithms to predict diabetes status based on voice features.
Potential Biases
Potential selection bias if certain groups were more likely to participate.
Limitations
The study relied on a sample of English speakers only and did not include an external validation dataset.
Participant Demographics
Participants included 323 females and 284 males, primarily white, with varying ages and health conditions.
Statistical Information
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website