Voice Analysis Using the LPC (Linear Predictive Coding)Method for the Prediction of Aspiration

Voice Analysis Using the LPC (Linear Predictive Coding)Method for the Prediction of Aspiration;Prospective, Positive-controlled, Single-blind, Single Cohort Clinical Study of Efficacy

It was well known that aspiration is the primary cause of an aspiration pneumonia among the dysphagia patients. In order to predict the degree of an aspitation,video-fluoroscopic-swallowing study and fiberoptic-endoscopic swallowing evaluation have been developed, but those evaluations are invasive and allowed only in a limited condition. In this study, as a precedent study for the non-linear analysis, voice analysis using LPC method will be evaluated. The superiority of LPC method will be compared with a spectral analysis using Jitter and Simmer as a positive control. This clinical study will be performed prospective, positive-controlled, single-blind (i.e. assessor-blind),and single cohort clinical trial for efficacy

Voice sampling is executed before and immediately after drinking of 20ml N/S and linear voice analyses using LPC method are performed to calculate LPC index, residual variance, coefficiency mean, coeffiency variance, and coeffiency skewness.

Voice sampling is executed before and immediately after drinking of 20ml N/S and spectral voice analyses using Dr. Speech program are performed to calculate HNR (dB), RAP (%: Jitter), Mean and SD Fundamental Frequency (Hz: Fo), Shimmer (%) and SNR (dB).

VFSS was perfomred one day before or after the voice sampling in the same patients according to the guideline of modified Longeman's protocol.

Voice sampling before and immediately after drinking of 20ml N/S and linear analysis using LPC method

Voice sampling is performed before and immediately after drinking of 20ml N/S and HNR (dB), RAP (%: Jitter), Mean and SD Fundamental Frequency (Hz: Fo), Shimmer (%) and SNR (dB) are calculated using Dr. Speech program.

Voice sampling is performed before and immediately after drinking of 20ml N/S and residual variance, coefficiency mean, coeffiency variance, and coeffiency skewness are computed using LPC analysis program.

According to Penetration-aspiration scale, patients are classified into high-risk patients (grade IV-VIII) and low-risk patients (grade I-III) by using the data derived from videofluoroscopic swallowing study

Inclusion Criteria: Patients who were referred for VFSS Exclusion Criteria: Patients cannot phonate d/t tracheostomy Preexisting oro-laryngo-pharyngeal disease Patients cannot follow 1 step command Patients cannot maintain a phonation for 3 seconds Refusal of participation Patients who were not classified for the candidate by physicians Patients who coughs following drinking of 5ml normal saline