Simulation-driven deep learning for the diagnosis of middle ear pathologies using wideband acoustic immittance

Clinical Snapshot

Key Findings

• P Value: Not reported

• Effect Size: Macro-F1: 96.30% on simulated test set; 87.76% on external clinical dataset. Balanced accuracy: 96.29% (simulated); 88.07% (clinical)

• Primary Outcome: Multi-class diagnostic accuracy of WAIHybrid CNN for four middle ear conditions (normal, ossicular chain discontinuity, ossicular chain fixation, otitis media with effusion)

• Nnt Or Sensitivity: Per-class sensitivity and specificity not reported. Macro-F1 of 87.76% on clinical data represents aggregate performance across four classes; class-level diagnostic accuracy unavailable from abstract

• Confidence Interval: Not reported

Clinical Application

WAI is an established clinical tool available in tertiary audiology and ENT centres. The WAIHybrid model is described as lightweight, suggesting potential for integration into existing WAI hardware platforms. However, clinical deployment requires prospective validation, regulatory approval, and standardisation of WAI acquisition protocols across devices and operators WAI is not yet a standard-of-care diagnostic tool in Australian primary care; tympanometry remains the predominant middle ear assessment in RACGP-aligned practice. Otitis media with effusion is a significant burden in Australian Indigenous children, where access to specialist ENT and audiology is limited — a validated, automated WAI tool could have meaningful public health impact in this population if robustly validated. The TGA would require prospective clinical validation data substantially beyond the current 206-ear dataset before regulatory clearance as a medical device software (SaMD). No PBS listing or MBS item number exists for WAI-based computer-aided diagnosis. RACGP guidelines do not currently reference WAI-based automated diagnosis. Adults with suspected middle ear pathology undergoing WAI testing in audiology or ENT settings; paediatric applicability unconfirmed. Conditions evaluated: normal ear, ossicular chain discontinuity, ossicular chain fixation, otitis media with effusion

Abstract

Wideband acoustic immittance (WAI) provides comprehensive frequency-dependent information for diagnosing middle ear pathologies. However, the scarcity of clinical data and complex response patterns significantly hinder automated diagnosis, particularly in data-limited scenarios. To address this issue, this study proposes a simulation-driven computer-aided diagnosis framework for WAI based on finite element (FE) modeling. Latin hypercube sampling was employed to systematically perturb key physiological parameters of the human ear FE models, generating a standardized virtual WAI dataset comprising 12 000 samples across the 0.2-6 kHz frequency range. The dataset includes four middle ear conditions: normal ear, ossicular chain discontinuity, ossicular chain fixation, and otitis media with effusion. Based on this dataset, a lightweight convolutional neural network tailored for multi-channel WAI inputs, termed WAIHybrid, was developed. It was benchmarked against traditional feature-based machine learning models and five representative deep learning architectures on simulated data and subsequently evaluated on an external clinical dataset comprising 206 ear-level WAI records. WAIHybrid achieved a macro-F1 of 96.30% and a balanced accuracy of 96.29% on an independent simulated test set. On the external clinical dataset, the corresponding values were 87.76% and 88.07%, respectively. Response-level comparisons, learned-representation analyses, and Integrated Gradients maps identified partial class-related correspondence between the simulated and clinical data, residual simulation-to-clinical discrepancy, and class-dependent channel-frequency attribution patterns. These findings support a simulation-driven proof of concept for automated WAI analysis in data-limited middle ear assessment. Further evaluation in larger, more balanced, and clinically heterogeneous cohorts is needed.

References

1. 1.Jiang, B., Li, C., Guo, W., Chen, Y., Zhao, Y., Jiang, W., Chen, W., Liu, W., & Liu, H. (2026). Simulation-driven deep learning for the diagnosis of middle ear pathologies using wideband acoustic immittance. Biomedical Physics & Engineering Express. https://doi.org/10.1088/2057-1976/ae885d