A deep learning approach for acoustic-based identification of muscle tension dysphonia and spasmodic dysphonia
-
By
-
Zhou Zhou
-
Yuan Cheng
-
Qingyi Ren
-
Yike Li
-
Xu Yuanyue
-
Jing Kang
-
Cheng Lu
-
Pingjiang Ge
-
July 9, 2026
-
Clinical Scorecard: Utilizing Deep Learning for Acoustic Identification of Muscle Tension Dysphonia and Spasmodic Dysphonia
At a Glance
| Category | Detail |
|---|
| Condition | Spasmodic Dysphonia and Muscle Tension Dysphonia |
| Key Mechanisms | Differentiation based on acoustic analysis using deep learning models. |
| Target Population | Patients with voice disorders, specifically SD and MTD. |
| Care Setting | Otolaryngology and speech-language pathology |
Key Highlights
- AI model achieved 89.5% accuracy in distinguishing healthy voices from disordered ones.
- Ternary classification accuracy of 71.6% for Healthy, MTD, and SD voices.
- AI performance surpassed human experts in both binary and ternary classifications.
- Model integrates explainable AI techniques for clinical trust.
- Designed for rapid processing compatible with clinical workflows.
Guideline-Based Recommendations
Diagnosis
- Utilize AI-based acoustic analysis for differential diagnosis of SD and MTD.
Management
- Consider botulinum toxin injections for SD and voice therapy for MTD.
Monitoring & Follow-up
- Regular assessment of voice characteristics using AI tools.
Risks
- Misdiagnosis can lead to delayed treatment and unnecessary procedures.
Patient & Prescribing Data
Individuals diagnosed with SD or MTD.
Differentiation between SD and MTD is crucial for appropriate treatment pathways.
Clinical Best Practices
- Implement AI tools for preliminary screening of voice disorders.
- Integrate AI diagnostics into routine clinical practice for enhanced accuracy.
Related Resources & Content
