Robust nasality representation learning for cleft palate-related velopharyngeal dysfunction screening in real-world settings - Takeaways - MDSpire

Robust nasality representation learning for cleft palate-related velopharyngeal dysfunction screening in real-world settings

  • By

  • Weixin Liu

  • Bowen Qu

  • Amy Stone

  • Maria Powell

  • Shama Dufresne

  • Stephane Braun

  • Izabela Galdyn

  • Michael Golinko

  • Bradley Malin

  • Zhijun Yin

  • Matthew E. Pontell

  • July 16, 2026

Share

  • 1

    Velopharyngeal dysfunction (VPD) impairs speech due to inadequate closure, leading to hypernasality and reduced intelligibility.

  • 2

    Current VPD screening requires specialized expertise and controlled conditions, limiting access in low- and middle-income countries.

  • 3

    The study introduces a two-stage framework using nasality representation pre-training to enhance robustness in real-world recording scenarios.

  • 4

    The proposed method achieved ceiling performance in standardized clinical settings and improved out-of-domain performance compared to existing baselines.

  • 5

    Learning a nasality-focused representation can mitigate recording artifacts, supporting practical deployment of VPD screening tools.

Original Source(s)

Robust nasality representation learning for cleft palate-related velopharyngeal dysfunction screening in real-world settings

  • by Weixin Liu, Bowen Qu, Amy Stone, Maria Powell, Shama Dufresne, Stephane Braun, Izabela Galdyn, Michael Golinko, Bradley Malin, Zhijun Yin, Matthew E. Pontell

  • July 16, 2026

Related Content

Spatial transcriptomics reveals molecular differences associated with malignant transformation in oral epithelial dysplasia

Case Report: Melanoma of the hard palate presenting initially with gastrointestinal symptoms: the diagnostic pivotal role of 18F-FDG PET/CT

Videofluoroscopic swallow study in dysphagia stroke patients