1. Abstract
Objective: This study investigates the potential of voice analysis as a tool for prescreening or monitoring Type 2 Diabetes Mellitus (T2DM) by examining differences in voice recordings between non-diabetic and T2DM individuals.
Methods: A total of 267 participants (79 women and 113 men non-diabetic, 18 women and 57 men T2DM) were recruited in India. Using a smartphone application, participants recorded a fixed phrase up to 6 times daily for 2 weeks, resulting in 18,465 recordings. Fourteen acoustic features were extracted from each recording to analyze differences between groups and create prediction models for T2DM status.
Results: Significant differences were found between voice recordings of non-diabetic and T2DM men and women. The highest predictive accuracy was achieved by pitch, pitch SD, and relative average perturbation jitter for women, and intensity and 11-point amplitude perturbation quotient shimmer for men. Incorporating these features with age and BMI, the optimal prediction models achieved accuracies of 0.75±0.22 for women and 0.70±0.10 for men through 5-fold cross-validation.
Conclusion: Vocal changes occur in individuals with T2DM compared to those without T2DM. Voice analysis shows potential as a prescreening or monitoring tool for T2DM, particularly when combined with other risk factors associated with the condition.
2. Rapid Reading
a. Research Methods:
- Data Collection: Participants recorded their voices daily using a smartphone application for 2 weeks.
- Feature Extraction: Fourteen acoustic features were extracted from each recording, including pitch, intensity, harmonic noise ratio, shimmer, and jitter.
- Model Building: Logistic regression, Gaussian Naïve Bayes, and Support Vector Machine models were trained and evaluated using 5-fold cross-validation.
- Model Evaluation: The best performing models were tested on a separate test set.
Innovations and Improvements: - Data Collection: Utilizing smartphone applications for data collection makes the process more accessible and convenient.
- Feature Extraction: Using a fixed phrase for recording provides more consistent and comparable data.
- Model Evaluation: Incorporating demographic factors like age and BMI into the prediction models improves accuracy.
Problems Solved: - Challenges of early detection and screening of T2DM.
- Invasiveness and cost of traditional detection methods.
b. Experiment Process: - Dataset: 267 participants, including non-diabetic and T2DM patients.
- Experiment Design: Age and BMI matched samples and separate test set.
- Results: Significant differences in voice features between groups and good performance of prediction models.
- Significance: Voice analysis has potential as a tool for T2DM screening and monitoring.
c. Main Advantages: - Innovation: Use of smartphone applications and fixed phrase recordings.
- Practicality: Non-invasive, low-cost, and easy to use.
- Accuracy: Good performance of prediction models with demographic factor integration.
Innovative Points and Potential Impact: - Voice analysis as a novel T2DM detection method.
- Improved efficiency of early detection and screening of T2DM.
- New insights for T2DM prevention and management.
3. Summary
a. Contributions:
- Research Methods: Use of smartphone applications and fixed phrase recordings for data collection.
- Results: Significant differences in voice features between T2DM and non-diabetic individuals and good performance of prediction models.
- Conclusion: Voice analysis has potential as a tool for T2DM screening and monitoring.
b. Innovative Points: - Use of smartphone applications for data collection.
- Incorporation of demographic factors in prediction models.
c. Future Research Outlook: - Larger and more diverse cohorts for model validation.
- Exploration of advanced machine learning models.
- Integration with other biomarkers for improved accuracy.
- Development of voice-based T2DM detection applications.
Overall, this study demonstrates the potential of voice analysis for T2DM screening and monitoring, offering a non-invasive, low-cost, and accessible approach for early detection and management of this prevalent condition.
4、Mind Map
Links to papers：https://arxiv.org/pdf/2407.08683
