Stochastic modeling of the human middle ear dynamics under pathological conditions

Area of Science:

• Bioacoustics
• Computational Auditory Neuroscience
• Medical Modeling

Background:

• Middle ear (ME) dynamics have been modeled since the mid-twentieth century.
• Stochastic approaches are recently being applied to ME modeling.
• This study utilizes a stochastic model to predict ME dynamics in healthy and pathological states.

Purpose of the Study:

• To develop and validate a stochastic model for middle ear (ME) dynamics.
• To predict ME behavior under both healthy and pathological conditions.
• To create a machine learning classifier for ME conditions based on simulated data.

Main Methods:

• A deterministic lumped-parameter model was randomized using a probabilistic non-parametric approach to create the stochastic ME model.
• The model was adapted to simulate pathological ME conditions.
• A machine learning algorithm was employed to develop a classifier for ME conditions using simulated data.

Main Results:

• The stochastic ME model showed good agreement with experimental data for healthy conditions.
• Simulated pathological conditions qualitatively matched experimental data.
• The developed classifier model demonstrated promising performance.

Conclusions:

• ME dynamics under different conditions can overlap across frequency ranges.
• Improvements are needed for the stochastic and classifier models.
• This research provides a foundation for developing tools for researchers and clinicians.