A Predictive Coding Model-based Sound Therapy for Patients With Subjective Tinnitus: A Pilot Study

This study investigates the therapeutic effects of a novel predictive coding model-based sound therapy using Shepard tones on patients with chronic subjective tinnitus. We hypothesized that the therapy would reduce tinnitus symptoms by minimizing prediction errors in auditory processing.

This single-centre, randomized, single-blind, 3-arm trial recruited 71 adults with chronic subjective tinnitus from National Cheng Kung University Hospital. The participants were randomized into 3 groups: an experimental group receiving Shepard tone therapy, control group A receiving down-shifted Shepard tone therapy, and control group B receiving matched-noise therapy. The primary outcome measure was Tinnitus Handicap Inventory (THI) score, with secondary outcomes including changes in tinnitus frequency and intensity assessed before and after 2 weeks of 1-hour daily therapy sessions.

The experimental group showed trends of reductions in THI score, tinnitus frequency, and intensity, although these changes were not statistically significant. There was a significant reduction in THI score ( P = 0.032) in control group B, while the change in control group A was minimal. Baseline tinnitus parameters were negatively correlated with post-therapy changes in the experimental group, indicating greater improvements in patients with more severe tinnitus.

The results of this study suggest that customized Shepard tone therapy may offer therapeutic benefits, particularly in individuals with severe tinnitus. These findings warrant further investigation in larger, long-term clinical trials.