Hearing Assessment: Lessons learned developing a modular solution for audiometric testing

Authors: Chambers, Robert D., Creare Inc.; Finger, William H., Creare Inc.; Norris, Jesse A., Creare Inc.; Cl

Track: Posters

Hearing loss is a leading public health concern, with 17% of American adults reporting some degree of impairment. To support research into differentiating the multiple sources of hearing impairment, we developed a handheld USB- and Bluetooth-enabled device which we term the Creare Hearing Assessment (CHA) system. The CHA includes an embedded DSP that handles signal generation and processing, as well as high-level algorithmic control, for a variety of hearing exams (e.g., speech-in-noise, audiometric thresholds, and otoacoustic emissions). This approach separates the time-critical exam algorithms from the user-interface and post-analysis tasks. To efficiently implement a rich and full-featured front-end to the CHA, we use Python and the Enthought Tool Suite.

Methods: We use TraitsUI and the Tasks framework for UI generation, Chaco for plotting, and Envisage as a plug-in architecture. We employ Traits throughout. Each audiometric exam type is implemented as an optional Envisage plug-in that provides a GUI for running the exam; Traits/TraitsUI classes for browsing and visualizing results; and a corresponding database schema. A custom subsystem transparently syncs between Traits objects (for data browsing and visualization) and SQLAlchemy objects (for database storage). For signal processing and data analysis, we use NumPy/SciPy. Finally, a Python scripting subsystem allows users to define extremely powerful and flexible assessment protocols which can respond dynamically to testing results.

Results: Challenges include packaging TraitsUI-based software as a standalone executable; addressing bottlenecks due to large datasets and ill-defined data dependencies; and implementing automated testing in a primarily graphical application. However, relative to previous C++-based implementations, the current approach has exceeded expectations in terms of development time, functionality, usability, and stability. It is currently in use in three separate research centers to evaluate the effects of noise exposure on hearing.