/arrhythmia-alarms

The objective of this work is to detect false arrhythmia alarms using convolutional neural networks as proposed in Reducing False Arrhythmia Alarms in the ICU: the PhysioNet/Computing in Cardiology Challenge 2015. This project was the code from my undergraduate thesis to complete my computer engineering degree.

Primary LanguagePythonMIT LicenseMIT

Detecting false arrhythmia alarms in the ICU using Convolutional Neural Networks

The objective of this work is to detect false arrhythmia alarms using convolutional neural networks as proposed in Reducing False Arrhythmia Alarms in the ICU: the PhysioNet/Computing in Cardiology Challenge 2015. This project was the code from my undergraduate thesis to finish my computer engineering degree.

Project Organization

├── LICENSE
├── README.md                  <- This file
├── reports
|   └── figures                <- Figures used in the final report
|   └── results                <- Tables with results of the 3-step grid-search training
│
├── requirements.txt           <- pip requirements
├── requirements-conda.txt     <- conda requirements
│
├── src
│   ├── data
│   │   ├── make_dataset.py     <- code to download, create and prepare database
│   │   ├── plot_dataset.py     <- auxiliar functions to create the plots used as input
│   │   ├── prepare_dataset.py  <- auxiliar code to prepare dataset
│   │   └── resample_dataset.py <- random oversampling code
│   │
│   ├── features
│   │   └── build_features.py   <- bottleneck features generation
│   │
│   ├── models
│   │   ├── metrics_callback.py <- class for using as callback on keras models
│   │   ├── models.py           <- models used in training
│   │   └── train_model.py      <- code for model training
│   └── experiment.py           <- experiment run

Project based on the cookiecutter data science project template. #cookiecutterdatascience