Preprocessing of Pure Tone Audiometry Data and Design of Machine Learning Models for Hearing Loss Classification

In this research, we conducted a preprocessing step on pure-tone audiometry image data to determine the presence or absence of hearing loss in patients, and designed machine learning models for hearing loss classification using the preprocessed data. The dataset utilized consisted of a total of $\mathbf{1 8, 5 3 0}$ data $(\mathbf{9, 2 6 5}$ normal, 9,265 hearing loss), and detailed parameter configuration was performed using randomly extracted training and validation data. This research involved converting image data into a CSV file format during the preprocessing stage, and the preprocessed data was then utilized to design and construct the Logistic Regression and Decision Tree Classifier machine learning models for patient classification. These models achieved an accuracy of $85.61 \%$ and $\mathbf{8 5. 2 5 \%}$, respectively, in automatically determining the presence or absence of hearing loss.