Heart failure (HF) is the most common cause of hospitalization among Americans over 65 years of age. A major goal of HF management is to maintain stability by predicting and preventing episodes of acute decompensated HF (ADHF). Body weight is currently used for non-invasive, at-home monitoring of volume status, and as an early warning for decompensation, by patients with HF. However, HF-related weight changes occur relatively close to the onset of symptoms and may not be detected in time to prevent an episode of decompensation. In contrast, the vocal folds consist of thin tissue layers that are impacted by body hydration levels and may therefore be especially sensitive to HF-related fluid retention. The amount of laryngeal edema required to change voice acoustics is expected to be small relative to the large amount of systemic edema needed to significantly increase body weight. Therefore, we hypothesize that clinicians will be able to detect and track HF-related volume overload more closely by monitoring voice metrics in addition to weight. The goal of this project is to investigate the potential of voice and speech characteristics as correlates of clinical improvement during treatment for ADHF. In a pilot study, acoustic voice and speech measures from ten HF patients undergoing diuresis for HF were analyzed. Several promising voice measures were identified for additional investigation, including measures reflecting higher pitch and increased vocal stability following successful ADHF treatment. In this project, the study size will be expanded to facilitate more rigorous and extensive statistical analysis techniques, including machine learning and multi-level modeling analyses. Additionally, the potential of a neck-surface vibration sensor for noise-robust, confidential monitoring of ADHF status will be evaluated. The vibration sensor has been used extensively for voice monitoring in our lab because it is minimally affected by environmental noise and does not record intelligible speech, making it ideal for recording voice while preserving patient privacy in noisy environments. The performance of microphone-based recordings in predicting patients? heart failure status will be compared with that of vibration-sensor-based recordings, with the hypothesis that the vibration sensor will work as well as, or better than, a conventional audio microphone. Completion of this project will result in a novel, voice-based method using wearable sensor technology to provide a monitoring system for HF patients recovering from decompensation, and will aid us in future work as we move towards developing an early warning system for patients at risk of ADHF.
The proposed project will investigate how voice and speech production change in response to treatment of acute decompensated heart failure (ADHF), which is characterized by an abnormal accumulation of fluid throughout the body. Excess fluid in the vocal folds and lungs is hypothesized to cause a number of changes in voice and speech, which can be detected with acoustic analysis of recordings from both a conventional microphone and a novel neck-skin vibration sensor. This research may provide clinicians and patients with a method of detecting impending ADHF earlier and more easily than is currently possible, which could save lives, reduce healthcare costs, and improve patients? quality of life.
