Back to Evidence Hub
Case Study S027
2020 Release

Suresh 2020: SeismoNet End-to-End DL

Prithu Suresh et al.
Access Paper

Quick Conclusion: S027 (SeismoNet) is a key technical reference for OpenSCG's early-stage ML development. It represents the shift from manual feature engineering to end-to-end deep learning, proving that raw SCG vibrations can be directly mapped to cardiac events using modern AI architectures.


📊 Key Accuracy Metrics

MetricResult
ModelSeismoNet (Deep Convolutional Neural Network)
FunctionTransforms SCG into interpretable heart rate indices
Repositorygithub.com/prithusuresh/SeismoNet


🔍 Study Analysis

Objective & Population

Deep Learning Development / End-to-End DCNN. Cohort: Preprocessed CEBS dataset (20 healthy subjects) (N=20).

What it Supports

The study supports the use of end-to-end deep convolutional neural networks (SeismoNet) to robustly extract heart activity from SCG signals. It provides an open-source framework for transforming raw accelerometer data into interpretable cardiac waveforms.

What it Does Not Support

The study does not provide evidence for clinical diagnosis of specific diseases, as its focus is on the signal processing and feature extraction framework.


🛠 Technical Context

  • DOI: arXiv:2010.05662
  • Authors: Prithu Suresh et al.
  • Confidence Tier: Supporting

Featured Illustration

No caption found automatically.

Study Snapshot

Metadata Summary

Target Population

Preprocessed CEBS dataset (20 healthy subjects)

N

Sample Size

20 Subjects

Validated Metric

SeismoNet (Deep Convolutional Neural Network)

Critical Appraisal
supporting

Demonstrated the use of end-to-end DCNN for robust SCG waveform transformation.