Predicting Individual Response to Acupuncture in Sensorineural Tinnitus Using Integrated Functional Near-Infrared Spectroscopy and Machine Learning: Protocol for a Model Development and Validation Study.

Background: Research indicates that around 20% of adults experience chronic tinnitus, with about a fifth of these cases being severe. Although various treatments are available for tinnitus, their effectiveness is often limited, and each treatment has clinical constraints. While acupuncture has shown promise in treating tinnitus, individual responses vary significantly. Identifying methods to predict acupuncture's effectiveness on Sensorineural tinnitus (SNT) patients in advance remains a critical clinical challenge. Purpose: This study aims to develop and validate a machine learning model based on functional near-infrared spectroscopy (fNIRS) data to predict acupuncture treatment outcomes in SNT patients. Methods and Analysis: This study will enroll 500 subjects with SNT, with sample size determined via established machine learning feature-to-sample ratio method. Specific brain regions will be scanned using fNIRS pretreatment, collecting data from multiple temporal and frontal lobe channels. Subjects will receive standardized acupuncture over four weeks. Outcomes will be evaluated using validated measures including Tinnitus Severity Grading and Tinnitus Handicap Inventory. Based on treatment responses, subjects will be categorized into "favorable prognosis" or "poor prognosis" groups. The dataset will be randomly split into training (70%) and test (30%) sets. Support Vector Machine (SVM) algorithms will identify features and develop models, with performance evaluated through accuracy, sensitivity, specificity, and AUC. Anticipated Results: It is anticipated that the fNIRS-based machine learning model will be able to distinguish between patients who will have a favorable response to acupuncture and those who will not, with acceptable accuracy. We expect to identify specific neurofunctional features from the temporal and frontal lobes that are predictive of treatment success. If so, this will provide an objective tool to aid in clinical decision-making and the personalization of tinnitus treatment. Conclusion: This study represents the first attempt to integrate fNIRS detection with machine learning techniques for predicting acupuncture efficacy in SNT treatment. The methodology addresses several key challenges in acupuncture research through comprehensive data collection and advanced analytical approaches. These findings could potentially enable more personalized treatment approaches for SNT patients and provide a foundation for future studies combining neuroimaging and machine learning in acupuncture research. Trial Registration: Clinical trials registry (identification code NCT06364670).