跳到正文
原文
Nature 健康护理主题·· 15 天前AI 评分42

采用声学特征的分层人工智能分类方法用于良性喉部疾病

Hierarchical AI classification of benign laryngeal disorders using acoustic voice features

AI 导读

该研究提出了一种分层机器学习框架,利用声学特征自动分类八种良性喉部疾病及其对照组。该框架在15,132个录音样本上进行了测试,展示了优于平面多类分类器和通用自监督模型(HuBERT, HeAR)的性能。通过结合深度频谱学习和可解释的声学特征,该方法为临床验证和外部测试提供了潜在的诊断工具。

正文

We’re sharing this article early to provide faster access to peer-reviewed, accepted research. It is citable and carries a permanent DOI. This version is subject to further edits and will be replaced automatically by the final Version of Record. All legal disclaimers apply.

Abstract

Benign laryngeal voice disorders affect nearly one in five individuals and often manifest as dysphonia, while also serving as non-invasive indicators of broader physiological dysfunction. We propose a clinically inspired hierarchical machine learning framework for automated classification of eight benign voice disorders alongside healthy controls using acoustic features extracted from short, sustained vowel phonations. Experiments were conducted on 15,132 recordings from 1,261 speakers in the Saarbrücken Voice Database, covering vowels /a/, /i/, and /u/ produced at neutral, high, low, and gliding pitches. Mimicking clinical triage workflows, the framework operates in three sequential stages: Stage 1 performs binary pathological screening by integrating convolutional neural network–derived mel-spectrogram representations with 21 interpretable acoustic features; Stage 2 groups voices into Healthy, Functional/Psychogenic, and Structural/Inflammatory categories; and Stage 3 performs fine-grained subtype classification by incorporating probabilistic outputs from preceding stages as auxiliary diagnostic features. The hierarchical framework demonstrated improved performance compared to flat multiclass classifiers and general-purpose self-supervised models (HuBERT, HeAR), suggesting the value of hierarchical decomposition for complex voice-disorder classification. A confidence-aware human-in-the-loop strategy further improved diagnostic performance by selectively referring low-confidence predictions for expert review while maintaining substantial automatic coverage, providing a practical pathway toward trustworthy clinical deployment and continual model refinement. But external validation on independent cohorts remains essential before clinical deployment. By combining deep spectral learning with interpretable voice features, the proposed framework demonstrates potential as a component of comprehensive diagnostic assessment pending clinical validation and external testing on diverse populations.

Acknowledgements

Anaïs Rameau was supported by a Paul B. Beeson Emerging Leaders Career Development Award in Aging (K76 AG079040) from the National Institute on Aging and by the Bridge2AI award (OT2 OD032720) from the NIH Common Fund.

Funding

The authors declare that this research received no external funding or financial support.

Author information

Authors and Affiliations

  1. Dalio Institute of Cardiovascular Imaging, Department of Radiology, Weill Cornell Medicine, New York, NY, USA

    Mohsen Annabestani

  2. Department of Speech Therapy, School of Rehabilitation, Tehran University of Medical Sciences, Tehran, Iran

    Samira Aghadoost

  3. Department of Otolaryngology-Head & Neck Surgery, Sean Parker Institute for the Voice, Weill Cornell Medicine, New York, NY, USA

    Anaïs Rameau

  4. Englander Institute for Precision Medicine, Department of Physiology and Biophysics, Weill Cornell Medicine, New York, NY, USA

    Olivier Elemento

  5. Brain Health Imaging Institute, Department of Radiology, Weill Cornell Medicine, New York, NY, USA

    Gloria Chia-Yi Chiang

Authors

  1. Mohsen Annabestani
  2. Samira Aghadoost
  3. Anaïs Rameau
  4. Olivier Elemento
  5. Gloria Chia-Yi Chiang

Corresponding authors

Correspondence to Mohsen Annabestani or Gloria Chia-Yi Chiang.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Annabestani, M., Aghadoost, S., Rameau, A. et al. Hierarchical AI classification of benign laryngeal disorders using acoustic voice features. Sci Rep (2026). https://doi.org/10.1038/s41598-026-79468-0

Download citation

  • Received: 09 April 2026

  • Accepted: 31 October 2026

  • Published: 05 November 2026

  • DOI: https://doi.org/10.1038/s41598-026-79468-0

Keywords

来源:Nature 健康护理主题 · nature.com