Back to Evidence Hub
Case Study S028
2025 Release

Korsgaard 2025: Beat-to-beat Delineation

T. Korsgaard et al.
Access Paper

Quick Conclusion: Introduces and validates a robust deep-learning framework (SeismoTracker) for automated cardiomechanical signal analysis.

📊 Key Accuracy Metrics

MetricResult
Positive Predictive Value (PPV)0.809 to 1.000
Sensitivity0.843 to 0.918
Total beats annotated42,452

🔍 Study Analysis

Objective & Population

Deep Learning Development / U-Net Segmentation. Cohort: 198 subjects (Both healthy and with cardiac diseases) (N=198).

What it Supports

Demonstrates high-reliability automated identification of 11 SCG fiducial points (PPV up to 1.0) across a diverse population (N=198).

What it Does Not Support

The study does not support perfect accuracy in pathological cases, where morphology is irregular.

🛠 Technical Context

Featured Illustration

No caption found automatically.

Study Snapshot

Metadata Summary

Target Population

198 subjects (Both healthy and with cardiac diseases)

N

Sample Size

198 Subjects

Validated Metric

0.809 to 1.000

Critical Appraisal
cornerstone

“Validated a robust deep-learning model for beat-to-beat segmentation of 11 SCG fiducial points across diverse populations.”