Efficacy of voice therapy associated with photobiomodulation in individuals with behavioral dysphonia: a randomized triple-blind clinical trial

Abstract

OBJECTIVE: To compare the efficacy of vocal therapy associated with photobiomodulation with isolated voice therapy in patients with behavioral dysphonia. METHODS: This was a randomized, prospective, triple-blind clinical trial, with random allocation of participants in two intervention groups: VT-LLLT (Vocal Therapy Associated with Low-Level Laser Therapy), VT-PLT (Vocal Therapy Associated with Placebo Laser Therapy). VT-LLLT received low-level laser therapy (10mW power, 808nm infrared, point, 30J) associated with the Comprehensive Vocal Rehabilitation Program (CVRP) for 8 sessions; VT-PLT received laser placebo before performing the CVRP for 8 sessions. 39 women with behavioral dysphonia between the ages of 18 and 60 years attended the Speech Therapy Clinic at the Federal University of Paraíba (UFPB) and the State University of Bahia to participate. The subjects who agreed to participate in the study signed a Free and Informed Consent Form (ICF). Videolaringostroboscopy were assessed using the Laryngeal Vibration and Structural Assessment Form (LVSAF); vocal self-assessment tools were also applied [The Voice Symptom Scale (VoiSS), Voice Handicap Index (VHI-10) and Voice Severity Rating Scale (VSRS)]; computerized acoustic assessment of the voice samples was performed with extraction of Jitter, Shimmer, Cepstral Peak Prominence Smoothed (CPPS), difference between the amplitude of the first and second harmonics (H1-H2), Harmonic-to-Noise Ratio (HNR), Glottal-to-Noise Excitation Ratio (GNE), Acoustic Voice Quality Index (AVQI), Acoustic Breathness Index (ABI), using the PRAAT and Voxplot softwares; Overall Severity of dysphonia (OS) was extracted through Visual Analog Scale (VAS) using the software VOXMore after choosing the most reliable judge. All evaluations were performed before and after intervention in subjects of all groups. The intervention lasted 8 weeks with weekly monitoring by the researcher and team in the proposed procedures. Statistical analysis to compare all evaluation outcomes before and after intervention, as well as the group differences that occurred, applying parametric and non-parametric tests with Statistical Package Software for Social Sciences (SPSS) 29.0.2.0 and R Studio 2025.09.1 programs, considering a confidence level of 95%. Machine Learning classifier results were generated using Weka 3.8.6 to recognize voice parameter patterns in the acoustic and self-assessment post-intervention data and classify the groups. RESULTS: ENT and self-assessment variables did not show statistical difference between the groups, only related to the time factor; CPPS (p=0.049), and AVQI (p=0.014) demonstrated a group factor statistically significant with better values on the VT-LLLT group after intervention; OS also showed a statistically significant effect related to the group factor with a better progress in the VT-LLLT group (p=0.002). Hybrid machine learning classifiers showed good accuracy (84,62%) on recognizing patterns and classifying both groups, but the VT-LLLT stood up. CONCLUSION: The results suggest that the LLLT combined with vocal therapy is more effective than isolated vocal therapy on BD treatment.