Skip to content

Customize AI models for protytype ECG heart rate device, wearable AI, for startup Areteus.us

Notifications You must be signed in to change notification settings

JennEYoon/ECG-transform

Repository files navigation

ECG-transform repo, README.md

Github: https://github.com/JennEYoon/ECG-transform/edit/main/README.md

Author: Jennifer E Yoon
Date Range: Nov 2024 - 2025

1. Quick Project Description:

This repo contains a deep learning project for Areteus.us. (See https://www.areteus.us/)
Areteus is developing a novel ECG heart rate detection wearable device. It's novel factors are: (1) continuously wearable at home and (2) comparable to hospital machines (12 or more signals). This AI project will customize ECG classification model(s) for Areteus's novel device. Visualization app showing live-feed ECG signals is also planned.

2. Device Information:

2.A) Areteus Device: See _Wearable_ECG_Device.md
https://github.com/JennEYoon/ECG-transform/blob/main/_Wearable_ECG_Device.md

2.B) to write, hospital ECG devices

2.C) at-home ECG devices

3. Literature and Public Datasets research:

3.A) Review ECGTransform paper, original paper provided by Areteus:
Paper review, datasets review.

  • MIT dataset
  • PTB dataset

3.B) Research other ECG papers, training models, simpler models:

  • Simple CNN
  • ResNet
  • RandomForest

3.C) Research other large datasets:
Chinese datasets
PTB-XL large dataset
Physionet challenge as data sources, paper sources.

3.D) ECG data splitting and processing methods research:

Currently here.

4. AI Model Development and Testing:

5. Key Findings:

6. App and Website Hosting, Report Writing:

  • documentation started
  • organization started
  • Embedding visualization (notebooks, plots) on website research
  • Interactive python viz libraries research

About

Customize AI models for protytype ECG heart rate device, wearable AI, for startup Areteus.us

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published