Deep Learning–Based Acoustic Screening for Penetration–Aspiration Events Using Short Voice Recordings
Dysphagia · 2026
DOI: 10.1007/s00455-026-10956-1 ↗Kurz gefasst
Multizentrische Studie in der Zeitschrift Dysphagia, 2026. Verfasst von Yong Jae Na und 13 weiteren Autor:innen. Beteiligte Einrichtung: Department of Physical Medicine & Rehabilitation, Chung-Ang University Gwangmyeong Hospital, Chung-Ang University College of Medicine, Gwangmyeong-si…. Hauptthemen laut MEDLINE: Deep Learning, Deglutition Disorders, Respiratory Aspiration. Die Arbeit nennt 28 Literaturangaben. Gefördert von: Korean DYSPHAGIA Society-KT.
Abstract (Originalsprache)
To evaluate the feasibility of a smartphone-based deep learning artificial intelligence (AI) tool for detecting post-swallow airway compromise through brief acoustic analysis of voice recordings obtained before and after swallowing. This multicenter prospective study employed a simple 1.5-second sustained phonation ("a~") recorded on a smartphone in patients referred for videofluoroscopic swallowing studies (VFSS). Cases were classified using the Penetration-Aspiration Scale (PAS), with PAS 1 defined as normal and PAS 2-8 as abnormal (penetration-aspiration events). An autoencoder-based anomaly detection model was trained on normal data (PAS 1) and validated using sensitivity, specificity, accuracy, and the area under the receiver operating characteristic curve (AUC). Among 208 participants, the AI model achieved a sensitivity of 90.9% and specificity of 87.5% in the validation set, with an accuracy of 90.4% and an AUC of 0.98. In the independent test set, sensitivity was 91.9%, specificity 50.0%, accuracy 85.2%, and AUC 0.76. A brief 1.5-second voice recording analyzed with a deep learning AI model showed promising internal performance for screening post-swallow airway compromise. This approach may serve as a practical and accessible adjunct to identify individuals requiring further instrumental assessment.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Physical Medicine & Rehabilitation, Chung-Ang University Gwangmyeong Hospital, Chung-Ang University College of Medicine, Gwangmyeong-si, Republic of Korea.
- Department of Rehabilitation Medicine, Asan Medical Center, University of Ulsan College of Medicine, 88, Olympic-Ro 43-Gil, Songpa-gu, Seoul, 05505, Republic of Korea.
- Department of Physical Medicine and Rehabilitation, Anam Hospital, Korea University, Korea University College of Medicine, Seoul, Republic of Korea.
- Cloud and AI Business Department, KT, Seoul, South Korea.
- Service Tech Lab, Institute of Convergence Technology, KT, Seoul, South Korea.
- Department of Rehabilitation Medicine, Gangnam Severance Hospital, Yonsei University College of Medicine, Seoul, Republic of Korea.
Förderung
- Korean DYSPHAGIA Society-KT
Zitieren
APA
Na, Y. J., Lee, J. H., Choi, E., Chang, J. Y., Seo, K. C., Lee, J., Ko, H., Kang, S. I., Yoo, M., Park, Y. G., Park, J., Kim, J., Lee, J. K., & Choi, K. (2026). Deep Learning–Based Acoustic Screening for Penetration–Aspiration Events Using Short Voice Recordings. Dysphagia. https://doi.org/10.1007/s00455-026-10956-1
Vancouver
Na YJ, Lee JH, Choi E, Chang JY, Seo KC, Lee J, et al. Deep Learning–Based Acoustic Screening for Penetration–Aspiration Events Using Short Voice Recordings. Dysphagia. 2026. doi:10.1007/s00455-026-10956-1
BibTeX
@article{Na2026,
author = {Na, Yong Jae and Lee, Jun Hyeok and Choi, Eunyoung and Chang, Jong Yoon and Seo, Kyung Cheon and Lee, Jaeho and Ko, Hunseok and Kang, Sang-Ick and Yoo, Myungeun and Park, Yoon Ghil and Park, Jinyoung and Kim, Jiyoung and Lee, Ju Kang and Choi, Kyounghyo},
title = {Deep Learning–Based Acoustic Screening for Penetration–Aspiration Events Using Short Voice Recordings},
journal = {Dysphagia},
year = {2026},
doi = {10.1007/s00455-026-10956-1}
}
EUVIMED