JMIR Publications

ABSTRACT

Background:

Heart failure (HF) contributes greatly to morbidity, mortality, and healthcare costs worldwide. This study describes a novel HF diagnostic technology using audio recordings from a standard cellular telephone.

Objective:

This is a prospective study of acoustic microphone recordings obtained at two different clinical sites used to create algorithms with physics-based models.

Methods:

The analysis matched acoustic data to ejection fraction (EF) as found by echocardiographic interpretations and stroke volume (SV) as evaluated by echocardiogram Teichholtz method. Recordings were obtained from 115 subjects. None were excluded.

Results:

Subjects had diverse racial and body mass indices (28-29)/ body surface areas. Reliable echocardiogram data for EF was available in 113 patients and for SV in 65 patients. The EF cohort had mean age 66.3 years and was 38.3% female. Using an EF cutoff of ≤40% vs. >40%, the model performed with area under the receiver operating characteristic of 0.955, sensitivity of 0.952, specificity of 0.958, and accuracy of 0.956. The SV cohort had mean age 65.5 years and was 33.8% female. Using a SV cutoff of <50ml vs >50ml, the model performed with area under the receiver operating characteristic of 0.922, sensitivity of 1.000 specificity of 0.844, and accuracy of 0.923.

Conclusions:

This work describes the use of recordings obtained with unaltered cellular phones to reproduce estimates of EF and SV with impressive accuracy. The research team is developing a phone application that could bring screening and monitoring of HF to remote and underserved areas across the globe with phones that patients already own and in situations where no other testing options exist.