Numerical prediction for Systolic Blood Pressure in Intradialytic Hypotension Using Time-relevant RNN Models

During hemodialysis (HD), intradialytic hypotension (IDH) is a serious complication and a major risk factor of mortality. This study aimed to use machine learning to predict IDH occurrence to improve prevention. In the proposed model in this study, we conducted Gated Recurrent Units, Deep Neural Net...

Full description

Saved in:
Bibliographic Details
Published in2021 IEEE 3rd Eurasia Conference on Biomedical Engineering, Healthcare and Sustainability (ECBIOS) pp. 57 - 59
Main Authors Tung, Nai-Yun, Hu, Hsiang Wei, Chi, Hsin-Yin, Chen, Kuan-Yu, Sung, Junne-Ming, Liu, Kuan-Hung, Boyce, Zachary, Lin, Chou-Ching, Law, David, Yu, Chang-Chia, Chen, Chen-Ying, Lin, Hsuan-Ming
Format Conference Proceeding
LanguageEnglish
Published IEEE 28.05.2021
Subjects
Biological system modeling
Intradialytic Hypotension
Long-Short Term Memory
Medical services
Physiology
Predictive models
Rated Recurrent Unit
Recurrent neural networks
Sensitivity
Wearable computers
Online AccessGet full text

Cover

More Information
Summary:During hemodialysis (HD), intradialytic hypotension (IDH) is a serious complication and a major risk factor of mortality. This study aimed to use machine learning to predict IDH occurrence to improve prevention. In the proposed model in this study, we conducted Gated Recurrent Units, Deep Neural Networks, and Long-Short-term Memory models to predict SBP values. For predicting IDH, a binary classification model was established. The results showed an accuracy of 90% with a difference between the predicted and actual value of 25mm- Hg for SBP value prediction. Also, the binary classification model had a threshold of 90mm-Hg with a accuracy of 93% and a specificity of 97%.
DOI:10.1109/ECBIOS51820.2021.9510228