Toward Accessible Auditory Health: A Cloud-Based fNIRS Solution for Auditory Training and Assessment

Auditory training (AT) is a proactive intervention for managing auditory health and preventing hearing loss. However, in its current form, it requires significant financial and time resources. As the excellent performance of functional near-infrared spectroscopy (fNIRS) in the medical field has led to its gradual application in auditory health, we aim to combine machine learning with fNIRS data to enhance accessibility and general applicability of AT. In this study, fNIRS was used to collect brain data related to auditory tasks and six machine learning methods were applied to classify different AT outcomes. Among these algorithms, AdaBoost demonstrated the best performance, achieving an accuracy of 88%. Based on the results, we propose a novel cloud-based framework that integrates AT with the assessment of training outcomes for individuals with hearing loss. The framework has been validated for its generalizability, and the evaluation results are not influenced by subjective experience.