แอปพลิเคชันประเมินกลศาสตร์ของเสียง : การศึกษานำร่องในผู้ป่วยพาร์กินสัน

Abstract

Parkinson's disease (PD) is a degenerative nervous system disorder that tends to increase every year. In addition to movement disorders, such as tremors and rigidity, patients also commonly experience hypokinetic dysarthria, which is which leads to speech impairments. The purpose of this study is to develop an application for acoustic assessment, encompassing: 1) loudness, 2) voice quality (jitter and shimmer), and 3) pitch (including high-low frequencies and fundamental frequency), with the aim of improving accuracy and precision in laboratory settings. To investigate the accuracy and precision in the laboratory, as well as to assess concurrent validity and test-retest reliability of the developed application, a pilot study was conducted among 20 healthy volunteers and volunteers with Parkinson's disease. Application for acoustic assessment was developed based on a literature review and by referring to the acoustic assessment model. The accuracy percentage test results compared to the Praat program ranged from 98.42% to 99.59%. The result of precision yielded a percentage coefficient of variation ranging from 1.65% to 3.78%. These results demonstrate that the application for acoustic assessment provides high accuracy and precision in testing. Concurrent validity testing involves assessing the relationship between tools by comparing acoustic assessment application with Praat program. This comparison utilizes Spearman's correlation coefficient to analyze the connection between the tools. The testing, involves a group of 20 healthy volunteers aged between 50 and 75 years. The results of the study revealed significant findings regarding the variables of loudness, pitch (high-low frequencies and fundamental frequency), and voice quality (jitter and shimmer) assessed through pronouncing /a/. There was a very strong correlation between the tools in loudness (median difference: MD = 0.23, 95% confident interval: 95% CI = 0.84-0.97, rs = 0.94), jitter (MD = 0.03, 95% CI = 0.86-0.97, rs = 0.94), high frequency (MD = 13.10, 95% CI = 0.95-0.99, rs = 0.98), and low frequency (MD = 0.25, 95% CI = 0.95-0.99, rs = 0.98), while shimmer showed a strong relationship between the tools (MD = -0.47, 95% CI = 0.42-0.89, rs = 0.73). In terms of reading the article "Fonfaa," only the fundamental frequency exhibited a strong relationship between the tools (MD = -40.30, 95% CI = 0.71-0.95, rs = 0.88), while loudness displayed a moderate relationship between the tools (MD = -4.46, 95% CI = 0.10-0.79, rs = 0.54). Jitter demonstrated a weak relationship correlation in the opposite direction (MD = 0.79, 95% CI = -0.70-0.10, rs = -0.37), and shimmer showed a very weak correlation in the opposite direction (MD = -3.15, 95% CI = 0.71-0.95, rs = -0.10). When testing in a group of 20 volunteers diagnosed with Parkinson's disease, aged between 50 and 75 years. The results of the concurrent validity test for the pronunciation /a/ showed a very strong level of correlation for loudness (MD = 1.05, 95% CI = 0.88-0.98, rs = 0.95), high frequency (MD = 55.09, 95% CI = 0.85-0.97, rs = 0.94), and low frequency (MD = 23.96, 95% CI = 0.96-0.99, rs = 0.98) tests. Additionally, jitter and shimmer exhibited strong correlations (MD = 0.05, 95% CI = 0.66-0.94, rs = 0.85 and MD = -0.38, 95% CI = 0.59-0.93, rs = 0.82, respectively). While reading the article "Fonfaa" the fundamental frequency exhibits a moderate relationship correlation (MD = -86.66, 95% CI = 0.19-0.82, rs = 0.59), while loudness demonstrates a weak relationship (MD = -0.27, 95% CI = -0.27-0.60, rs = 0.20). Additionally, jitter show a weak correlation in the opposite direction (MD = -1.44, 95% CI = -0.77--0.04, rs = -0.49), and shimmer exhibits a very weak correlation in the opposite direction as well (MD = -8.66, 95% CI = -0.19-0.65, rs = -0.28). In the test-retest reliability of both pronunciation of /a/ and reading the article "Fonfaa" showed a very strong level of correlation for all parameters, except for jitter in reading the article, which had a strong level of correlation (r = 0.86). When comparing a group of healthy volunteers with a group of Parkinson's patients using acoustic assessment application, it was found that the healthy volunteers had better test results than the Parkinson's patients in every test, including both pronunciation of /a/ and reading passages." The results of the study found that the acoustic assessment application demonstrates high concurrent validity and test-retest reliability in the /a/ pronunciation test, making it suitable for clinical assessment. However, in the "Fonfaa " article reading test, it demonstrated low validity and reliability. This could be attributed to Parkinson's patients' challenges in maintaining their voice while reading lengthy passages aloud, which may be due to voice abnormalities. This underscores the necessity of revising and developing practices, such as selecting shorter articles for future study.