Application of Machine Learning Based on fNIRS in Predicting Acupuncture's Efficacy in Treating Tinnitus

This study will recruit 500 subjects with tinnitus. Functional near-infrared spectroscopy (fNIRS) will be employed to examine specific brain regions, and the corresponding fNIRS imaging data from all detection channels will be extracted. Subsequently, the subjects will undergo a course of acupuncture treatment. Based on the recovery status of tinnitus at the conclusion of the acupuncture course, all subjects will be categorized into a "good prognosis group" and a "poor prognosis group" according to relevant efficacy criteria. The entire dataset will then be randomly divided into a training set (70%) and a test set (30%) following a 7:3 ratio.

Inclusion criteria

• Bilateral tinnitus that meets the diagnostic criteria for chronic subjective tinnitus.
• Male and female, aged between 18 and 60 years.
• Right-handed subjects who are able to comply with the study protocol and sign written informed consent.
• Not participating in other clinical trials concurrently.

Exclusion criteria

• Participants with objective tinnitus.
• Participants have nervous system diseases or neuropsychiatric diseases that can significantly affect brain blood oxygen metabolism assessed by fNIRS.
• Participants with severe cardiovascular and cerebrovascular diseases, malignant liver and kidney diseases, and other serious diseases.
• Participants have any contraindications for acupuncture (such as a bleeding tendency).
• Pregnant or lactating women.
• Participants have received tinnitus treatment with drugs or other therapies in the last four weeks.