Skip to content

Latest commit

 

History

History
136 lines (104 loc) · 5.54 KB

README.md

File metadata and controls

136 lines (104 loc) · 5.54 KB
Raw

RRWaveNet: A Compact End-to-End Multiscale Residual CNN for Robust PPG Respiratory Rate Estimation

Official implementation of RRWaveNet from IEEE Internet of Things Journal (IOTJ).

Paper authors: Pongpanut Osathitporn, Guntitat Sawadwuthikul, Punnawish Thuwajit, Kawisara Ueafuea, Thee Mateepithaktham, Narin Kunaseth, Tanut Choksatchawathi, Proadpran Punyabukkana, Emmanuel Mignot, and Theerawit Wilaiprasitporn

Abstract

DOI

alt text

Figure: Composed of three modules, RRWaveNet involves the multiscale convolution (left), the deep spatial-temporal residual blocks (center), and the RR estimator (right). Each layer’s title is abbreviated at the top row for simplicity and the shape of the output tensor after each layer is specified below its title. For example, conv1dk32s5, the leftmost layer in the center module, refers to a 1-D-convolutional layer with a kernel of size 32 and a stride of 5, resulting a (10 W, 1) tensor.

Respiratory rate (RR) is an important biomarker as RR changes can reflect severe medical events, such as heart disease, lung disease, and sleep disorders. Unfortunately, standard manual RR counting is prone to human error and cannot be performed continuously. This study proposes a method for continuously estimating RR, RRWaveNet. The method is a compact end-to-end deep learning model which does not require feature engineering and can use low-cost raw photoplethysmography (PPG) as input signal. RRWaveNet was tested subject-independently and compared to baseline in four data sets (BIDMC, CapnoBase, WESAD, and SensAI) and using three window sizes (16, 32, and 64 s). RRWaveNet outperformed current state-of-the-art methods with mean absolute errors at an optimal window size of $1.66 \pm 1.01$, $1.59 \pm 1.08$, $1.92 \pm 0.96$, and $1.23 \pm 0.61$ breaths per minute for each data set. In remote monitoring settings, such as in the WESAD and SensAI data sets, we apply transfer learning to improve the performance using two other ICU data sets as pretraining data sets, reducing the MAE by up to 21%. This shows that this model allows accurate and practical estimation of RR on affordable and wearable devices. Our study also shows feasibility of remote RR monitoring in the context of telemedicine and at home.

Usage

Download and install the dependencies

# Clone the repository
git clone https://github.com/IoBT-VISTEC/RRWaveNet

# Install dependencies. Recommend to use a separate environment.
pip install -r requirements.txt

Prepare the data

The data must be accessible at /path/to/this/repo/data/{dataset_name}/{file}_{winsize}.npy where file must be both X and RR (CO2 is also required to run RespNet). For example:

data
└── bidmc
    ├── CO2_16.npy
    ├── RR_16.npy
    └── X_16.npy

The structure of each X_*.npy and CO2_*.npy file should

  • Be iterable
  • Consist of n_subjects elements, all iterable
  • Each of the n_subjects should contain n_windows elements
  • Each of the n_windows elements should be a 1D np.ndarray which contains the data of PPG or CO2 of 1 minute in length with a shape of (60 * sampling_rate, )

For RR_*.npy files, they should

  • Be iterable
  • Consist of n_subjects elements, all iterable
  • Each of the n_subjects should contain n_windows floats, each number indicating the BPM of the corresponding window
# Example
x = np.load(f"data/bidmc/X_16.npy", allow_pickle=True)
y = np.load(f"data/bidmc/RR_16.npy", allow_pickle=True)

print(len(x))           # 50
print(len(x[0]))        # 141
print(len(x[1]))        # 150
print(type(x[0][0]))    # <class 'numpy.ndarray'>
print(x[0][0].shape)    # (2000,)

print(len(y))           # 50
print(len(y[0]))        # 141
print(type(y[0][0]))    # <class 'numpy.float64'>

Run the experiment

python main.py --model [model_name] --dataset [dataset_name] --winsize [window_size]

Valid options are

  • --model: DeepLearning, RespNet, RespWatch, or RRWaveNet.
  • --dataset: Depends on the available data you prepared, we used bidmc, capnobase, and wesad for the experiments in the paper.
  • --winsize: Depends on the available data you prepared, we used 16, 32, and 64 for our experiments.

The implementation of AR is not provided in this repository. We recommend the reproducing steps as defined in the original paper.

When the experiment finishes, the average and standard deviation of the MAEs across all subjects will be printed.

Should you encounter any issue or bug, please create an issue with details here.

Citation

Plaintext

P. Osathitporn et al., "RRWaveNet: A Compact End-to-End Multiscale Residual CNN for Robust PPG Respiratory Rate Estimation," in IEEE Internet of Things Journal, vol. 10, no. 18, pp. 15943-15952, 15 Sept.15, 2023, doi: 10.1109/JIOT.2023.3265980.

BibTeX

@ARTICLE{10098530,
  author={Osathitporn, Pongpanut and Sawadwuthikul, Guntitat and Thuwajit, Punnawish and Ueafuea, Kawisara and Mateepithaktham, Thee and Kunaseth, Narin and Choksatchawathi, Tanut and Punyabukkana, Proadpran and Mignot, Emmanuel and Wilaiprasitporn, Theerawit},
  journal={IEEE Internet of Things Journal}, 
  title={RRWaveNet: A Compact End-to-End Multiscale Residual CNN for Robust PPG Respiratory Rate Estimation}, 
  year={2023},
  volume={10},
  number={18},
  pages={15943-15952},
  doi={10.1109/JIOT.2023.3265980}}