Search

This research investigates the application of machine learning to improve the diagnosis of tinnitus using high-frequency audiometry data. A Logistic Regression (LR) model was developed alongside an Artificial Neural Network (ANN) and various baseline classifiers to identify the most effective approach for classifying tinnitus presence. The methodology encompassed data preprocessing, feature extraction focused on point detection, and rigorous model evaluation through performance metrics including accuracy, Area Under the ROC Curve (AUC), precision, recall, and F1 scores. The main findings reveal that the LR model, supported by the ANN, significantly outperformed other machine learning models, achieving an accuracy of 94.06%, an AUC of 97.06%, and high precision and recall scores. These results demonstrate the efficacy of the LR model and ANN in accurately diagnosing tinnitus, surpassing traditional diagnostic methods that rely on subjective assessments. The implications of this research are substantial for clinical audiology, suggesting that machine learning, particularly advanced models like ANNs, can provide a more objective and quantifiable tool for tinnitus diagnosis, especially when utilizing high-frequency audiometry data not typically assessed in standard hearing tests. The study underscores the potential for machine learning to facilitate earlier and more accurate tinnitus detection, which could lead to improved patient outcomes. Future work should aim to expand the dataset diversity, explore a broader range of algorithms, and conduct clinical trials to validate the models' practical utility. The research highlights the transformative potential of machine learning, including the LR model and ANN, in audiology, paving the way for advancements in the diagnosis and treatment of tinnitus.

The Tinnitus Functional Index (TFI) was developed to be responsive to small treatment-related changes in the impact of tinnitus. Yet, no studies have integrated anchor-based and distribution-based techniques to produce a single Minimal Important Change (MIC) score. Here, we evaluated the responsiveness and interpretability of the TFI, determining for the first time a robust MIC score in a UK clinical population. Two-hundred and fifty-five patients with tinnitus participated in this prospective longitudinal validation study. Distribution-based estimates (Standard Error of Measurement, Smallest Detectable Change and Effect size) and anchor-based estimates of important change (minimal clinically important difference and Receiver Operator Curve optimal value) were calculated and then integrated using a visual anchor-based MIC distribution plot. A reduction in score of −14 was determined as the MIC estimate that exceeds the measurement error, most of the variability and reliably identifies patients demonstrating true improvement. It is therefore recommended that a reduction of 14 points should be used as a minimum change required when calculating statistical power and sample size in tinnitus intervention studies and assessing patients in clinical practice.

The tinnitus spectrum is a psycho-acoustic metric of tinnitus. Previous work found a tight relation between the spectrum and the tone audiogram. This suggests that the spectrum and the audiogram provide essentially the same information, and the added value of the spectrum is limited. In order to test whether the spectrum shows tinnitus characteristics that cannot be inferred from the audiogram, we re-examined the relation between the tinnitus spectrum and the tone audiogram, in a group of 80 tinnitus patients. We defined three subgroups of patients, using the shape of their tinnitus spectrum: (1) patients with a spectrum, monotonously increasing with frequency (2) patients with a distinct peak in their spectrum, (3) all other patients. Patients in group 3 typically showed low frequency tinnitus spectra. In all three groups, the largest hearing loss was at high frequencies (>2 kHz). The mean audiograms of group 1 and 2 were remarkably similar; group 3 had an additional hearing loss for the lower frequencies (