# Integrating data from a wideband acoustic immittance database to develop machine-learning models that characterize pathology in auditory measurements

> **NIH NIH R15** · SMITH COLLEGE · 2024 · $423,986

## Abstract

PROJECT SUMMARY / ABSTRACT
This work provides fundamental knowledge for the development of wideband acoustic immittance (WAI)
measures as a noninvasive auditory diagnostic tool, where WAI refers to the collection of measurements that
include absorbance, power reflectance, impedance, and related quantities. Potential uses of these measures
include (1) detection of fluid in newborn and young infant ears where tympanometry is less successful and
early diagnosis can be critical to supporting normal speech and language development, (2) identification of the
cause of a conductive hearing loss (e.g., fluid, disarticulated ossicle, fixed ossicle), and (3) monitoring changes
in middle-ear stiffness that result from intracranial pressure changes. The proposed work continues to expand
and update the world’s only online WAI database and corresponding website, which was developed during
the first two cycles of this grant; the database now includes more than 4.9 million rows of WAI data from 9903
subjects and 41 publications that collectively include both normal and diseased ears from subjects of all ages.
This database has provided and will continue to provide additional measurements that the hearing community
can utilize to (1) define features of normative measures for clinical application of WAI measurements, (2)
determine how specific pathologies affect WAI, and (3) train systems that apply machine learning for the
interpretation of WAI measurements. The proposed work expands upon the work in previous grant cycles
to include (1) controlled laboratory-based WAI measurements that will inform criteria for the development of
WAI measurement validity criteria by systematically studying acoustic leaks and probe placements along the
canal and (2) the development of machine-learning models to classify WAI measurements made on normal
and abnormal ears and to ultimately classify specific types of pathologies. A second emphasis of the proposed
work is the research-based education of undergraduate students at Smith College, an all-women’s liberal arts
college. Undergraduate engineering, computer science, statistics, data science, and mathematics students will
be actively involved in all areas of the proposed work, with the goal of encouraging them to continue their
education at the graduate level where they can contribute to health-related research.

## Key facts

- **NIH application ID:** 10973867
- **Project number:** 2R15DC014129-03A1
- **Recipient organization:** SMITH COLLEGE
- **Principal Investigator:** SUSAN E VOSS
- **Activity code:** R15 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2024
- **Award amount:** $423,986
- **Award type:** 2
- **Project period:** 2014-06-05 → 2027-06-30

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10973867

## Citation

> US National Institutes of Health, RePORTER application 10973867, Integrating data from a wideband acoustic immittance database to develop machine-learning models that characterize pathology in auditory measurements (2R15DC014129-03A1). Retrieved via AI Analytics 2026-05-27 from https://api.ai-analytics.org/grant/nih/10973867. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*