Point-of-care aid-to-diagnosis for heart failure using artificial intelligence based on seismocardiography acquired with a smartphone in the emergency department
Executive Summary
This study evaluates the diagnostic performance of a smartphone-based seismocardiography (SCG) system combined with artificial intelligence (AI) for detecting heart failure (HF) in emergency department patients presenting with dyspnea. Using cardiac kinetic energy (CKE) metrics derived from SCG, the AI achieved an AUROC of 0.82, sensitivity of 82.5%, and negative predictive value (NPV) of 79.3%, demonstrating its potential as a non-invasive, point-of-care diagnostic tool.
Answer Machine Insights
Q: What was the primary diagnostic metric used in the study?
Cardiac kinetic energy (CKE) metrics, specifically ΔiKdiastolic, derived from SCG recordings.
From these vibrations, cardiac kinetic energy (CKE) metrics are derived, including diastolic gradient kinetic energy (ΔiKdiastolic).
Q: How effective was the AI system in diagnosing HF?
The AI system achieved an AUROC of 0.82, sensitivity of 82.5%, and NPV of 79.3%.
The AI approach resulted in an AUROC of 0.82 (0.80–0.89), sensitivity of 82.5% (72·7–90·2), and NPV of 79.3% (77·5–82·6).
Key Results
AI achieved an AUROC of 0.82 for HF classification.
ΔiKdiastolic was significantly lower in HF patients (0.06 vs. 0.35, p<0.005).
Clinical Snapshot
Evidence Rating
Relevance
high Priority