Neural Noise

Brain benefits of deep learning-based noise management in hearing aid users (fNIRS)

2 min read

  • fNIRS has been used to assess cortical effects of hearing interventions in experienced hearing aid users.1
  • Deep learning-based noise management in hearing aids shows measurable brain benefits when evaluated with fNIRS.1
  • The work links sensory neuroprosthetics (hearing) with neuroimaging and DL; Nature, tier-2.1 1

Weekly enrichment (2026-07-20)

  • The peer-reviewed source is a Scientific Reports study (s41598-025-25801-y) using fNIRS to measure prefrontal cortical oxygenation as an objective marker of listening effort in experienced hearing aid users.2
  • Twenty-six experienced hearing aid users repeated sentence-final words presented in noise while wearing individually prescribed aids in two conditions: a standard omnidirectional program and a deep-neural-network (DNN) program combining directional microphones with DNN-based noise management.2
  • The DNN program significantly lowered blood oxygenation in the left lateral prefrontal cortex, decreasing by roughly 18.97 units versus the standard program (b = −18.97, 95% CI [−30.86, −7.08], p = .0018), consistent with reduced cognitive/listening effort.2
  • No significant program effect appeared in the right lateral (p = .62) or lower medial prefrontal cortex (p = .93), indicating a regionally specific benefit rather than a global oxygenation shift.2
  • The interaction model (hearing aid program × prefrontal subregion) fit significantly better than the unconditional model (χ²(4) = 19.71, p = .0031) and than the main-effect-only model (χ²(4) = 10.8, p = .029).2
  • Manufacturer reporting on the same DNN feature (Phonak’s Spheric Speech Clarity/“Spheric Speech in Loud Noise,” in Audéo Sphere) cites an average 13.9% more words repeated correctly and roughly 0.94 fewer points of rated listening effort on a 7-point scale versus the standard program.3
  • fNIRS recorded three prefrontal regions (left lateral target area, lower middle, and right lateral), and handedness was a significant covariate, with right-handed participants showing ~13.45 units higher oxygenation overall.2
  • The work positions fNIRS as a practical, hearing-aid-compatible neuroimaging tool for revealing latent benefits of AI-driven signal processing even when behavioral speech scores appear similar, with relevance to sensory neuroprosthetics.2

Footnotes

  1. https://news.google.com/rss/articles/CBMiX0FVX3lxTE96VFpiN3Jjbmx5RWJxczUzZ2M5Q2V4U2hUdl9KQ2ktR0YyVF9NTmkzOURkMG03c25HZE8tWGxob1BQNjNtTERuNHFwYlBqUXM3clJKWG5Va2RTNnI5Q2xv?oc=5 ↩ ↩2 ↩3 ↩4

  2. https://www.nature.com/articles/s41598-025-25801-y ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7

  3. https://audiologyblog.phonakpro.com/what-can-brain-imaging-tell-us-about-hearing-aid-features-and-listening-effort/ ↩

Graph View