Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
December 10, 2021Applied SciencesOpen Access

Reducing the Heart Failure Burden in Romania by Predicting Congestive Heart Failure Using Artificial Intelligence: Proof of Concept

View Full Paper
Ask AI
Bookmark
Share

Why the study?

With population aging increasing chronic heart failure cases, the study aimed to implement a noncontact system to predict heart failure exacerbation through vocal analysis.

Does machine-learning-based vocal analysis accurately classify NYHA stage in patients hospitalized with acute pulmonary edema?

Population

16 hospitalized patients admitted with cardiogenic acute pulmonary edema

Comparison

Evaluation across different machine-learning algorithms (ANN, SVM, KNN) using voice recordings

Design

Proof-of-concept observational study

Follow-up

From day one of hospitalization until discharge

Authors

MPMaria-Alexandra PanăȘBȘtefan BusnatuLȘLiviu Ionuț Șerbănoiu

Discussion

Loading...

Member takes

Overview

Noncontact vocal analysis shows promise for remote HF monitoring; leaves open validation in larger prospective cohorts before clinical use.

Structured PICO

Does machine-learning-based vocal analysis accurately classify NYHA stage in patients hospitalized with acute pulmonary edema?

P
Population
16 hospitalized patients (9 men, 7 women, aged 65–91 years old) admitted with cardiogenic acute pulmonary edema.
I
Intervention
Vocal analysis using machine-learning algorithms (Artificial Neural Networks, Support Vector Machine, K-Nearest Neighbors) based on voice recordings taken twice daily via smartphone.
O
Outcome
Classification accuracy of NYHA stage based on voice recording.

Machine-learning analysis of voice recordings shows high preliminary accuracy in classifying heart failure severity, offering a potential non-contact monitoring tool.

Limitations

  • Small sample size

Cite This Study

Pană et al. (2021) studied this question.

synapsesocial.com/papers/6a70c2183ce530166bc2ea25https://doi.org/10.3390/app112411728
View Full Paper
Ask AI
Bookmark
Share