![]() In addition, studies advocated using vocal biomarkers recorded via smartphones to classify patients with coronary artery disease and pulmonary hypertension 15, 16, 17, 18. Another study using Gaussian mixture model system reported discriminating vocal fold disorders with 99% accuracy 14. Recently, machine learning (ML) and deep learning methods have been used to better predict voice disorders, achieving accuracy levels as high as 90% by using the acoustic parameters of jitter, shimmer, and noise-to-harmonic ratio (NHR) 13. An acoustic analysis can increase the sensitivity of detecting voice changes objectively and help in clinical evaluation 11, 12. While speech-language pathologists and other experts can reliably detect pathological voice changes, non-experts can be inaccurate 10 or miss them 5. A study has demonstrated that 90% of aspirators exhibit dysphonic vocal quality 6, and specific voice patterns may indicate aspiration 7, 8, 9. Among these clinical signs, voice change is associated with aspiration and penetration 5. Impaired gag reflex, dysphonia, weak cough, and choking after swallowing are known predictive factors 4. Early screening and prevention of these respiratory events may also affect prognosis a recent study found that any previous episode of aspiration pneumonia resulted in poor stroke outcomes 3.Ĭonsequently, efforts have been made to develop a screening test that can safely and quickly predict aspiration pneumonia. These respiratory complications occur in approximately one-third of the post-stroke dysphagia population and are associated with high mortality and morbidity 2. Voice features may be considered as viable digital biomarkers in those at risk of respiratory complications related to post-stroke dysphagia.ĭisturbed swallowing, or oropharyngeal dysphagia, commonly occurs after cerebrovascular disease, and may result in malnutrition, dehydration, and aspiration pneumonia 1. In both cases, voice features proved to be the strongest contributing factors in these models. The eXtreme gradient boosting multimodal models that included abnormal acoustic features and clinical variables showed high sensitivity levels of 88.7% (95% CI 82.6–94.7) and 84.5% (95% CI 76.9–92.1) in the classification of those at risk of tube feeding and at high risk of respiratory complications respectively. A total of 449 samples were obtained, with 234 requiring tube feeding and 113 showing high risk of respiratory complications. Subjects that required tube feeding were further classified to high risk of respiratory complication, based on the voluntary cough strength and abnormal chest x-ray images. Voice samples from patients referred for swallowing disturbance in a university-affiliated hospital were collected prospectively using a mobile device. ![]() This study was aimed to determine whether machine learning algorithms with voice recorded via a mobile device can accurately classify those with dysphagia at risk of tube feeding and post-stroke aspiration pneumonia and be used as digital biomarkers. ![]() Abnormal voice may identify those at risk of post-stroke aspiration. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |