乌鸦传媒

Acoustic signals picked up during phone calls may help identify patients with congestive heart failure (CHF) who are at risk of death or hospitalization within months, one group reported.

From telephone conversations recorded between patients and nurses as part of a hospitals' telemedicine service, a vocal fingerprint identified was associated with increased risk of death (23%, 29%, 38%, and 54% across quartiles of implicated vocal features, P<0.001) and hospitalization (54%, 58%, 63%, and 70%, P<0.001) over 20-month follow-up.

On multivariate analysis, each standard deviation increase in vocal biomarker features was tied to increased risk of death (adjusted HR 1.32, 95% CI, 1.24–1.41) and hospitalization (adjusted HR 1.21, 95% CI 1.15-1.28), according to a group led by Elad Maor, MD, PhD, of Sheba Medical Center Hospital-Tel Hashomer in Ramat Gan, Israel.

"While data on the effectiveness and the exact role of telemedicine in heart failure [HF] are conflicting, the vocal biomarker holds the potential to assist in identification of high-risk subjects in areas where access to services such as physical examination, blood tests, and cardiovascular imaging are limited," the group wrote in the .

"Our study therefore supports the use of vocal biomarkers, which are noninvasive and can be incorporated to any smartphone or even landline phone, for risk assessment of HF patients in telemedicine settings," the authors concluded.

They suggested that the vagus nerve, which participates in voice production and is critical for autonomic control of the heart, is one possible link between voice and HF.

Phone recordings used for the study came from an Israeli hospital that provides telemedicine services to complex patients with various chronic conditions. A machine learning model was trained on audio recordings from 8,316 non-CHF patients and tested on a separate CHF cohort of 2,267 patients (median age 77, 63% men).

In the end, the vocal biomarker, identified by the machine learning algorithm, incorporated 223 acoustic features extracted from the recordings (all in Hebrew or Russian).

This vocal fingerprint signaled risk of death in all subgroups.

However, those with low estimated glomerular filtration rate had an attenuated association between the biomarker and risk of death (P=0.001 for interaction), whereas non-obese individuals had significantly more pronounced risk compared to obese peers (P=0.01 for interaction).

Maor and colleagues acknowledged that their results had unknown generalizability to other populations. Statistical adjustment was also imperfect given that some lab data, New York Heart Association functional class, and echocardiographic ejection fraction were not available for many study participants.

"Lastly, this is an observational study and therefore the consistency of our findings including the validation and incremental prognostic value of the vocal biomarker need to be addressed in future controlled studies," the investigators said.

Nevertheless, voice signal characteristics have been tied to various disorders, including dyslexia, attention deficit hyperactive disorder, and Parkinson's disease, the authors noted.

Maor and colleagues said they had previously linked coronary artery disease to certain vocal characteristics.

Comment