v3551G / estimateBP

An implementation of 'Continuous Systolic and Diastolic Blood Pressure Estimation Utilizing Long Short-term Memory Network' (Frank P.-W. et al.) paper

Continuous Systolic and Diastolic Blood Pressure Estimation Utilizing Long Short-term Memory Network

This respository was created to implement this paper which title is
'Continuous Systolic and Diastolic Blood Pressure Estimation Utilizing Long Short-term Memory Network' by Frank P.-W. Lo, Charles X.-T. Li, Jiankun Wang, Jiyu Cheng and Max Q.-H. Meng, Fellow, IEEE

The network model of this method is seq2seq which I developed using Keras. (for more information please visit link)

Dataset

You can download the dataset here. After Downloaded the dataset put it in data/.

Test on

• python 3.7.4
• keras 2.3.1
• tensorflow 2.0.0
• numpy 1.17.2

Run the model

To run the code please open jupyter notebook:

1. Run preprocess.ipynb
2. Run train.ipynb

Experimental Results

The results will be different from the paper because I used 84 subjects, so the result will be a little bit similar to
Long-term Blood Pressure Prediction with Deep Recurrent Neural Networks which also used 84 subjects.

My Results are:

Overall RMSE Systolic: 6.053 (mmHg)
Overall RMSE Diastolic: 3.402 (mmHg)

For future modification to Improve Model Performance

You can modify the network model in model.py and test it in model_eval.ipynb

References

• Continuous Systolic and Diastolic Blood Pressure Estimation Utilizing Long Short-term Memory Network by Frank P.-W. Lo, Charles X.-T. Li, Jiankun Wang, Jiyu Cheng and Max Q.-H. Meng, Fellow, IEEE
• Keras implementation of a sequence to sequence model for time series prediction using an encoder-decoder architecture. by Luke Tonin
• Long-term Blood Pressure Prediction with Deep Recurrent Neural Networks by Peng Su, Xiao-Rong Ding, Yuan-Ting Zhang, Jing Liu, Fen Miao, Ni Zhao