Evidence-based lab test critical value discovery for ICU patients
In clinical research, a very common task is to predict the patients’ potential critical conditions in future using the time series data collected from the patients. Recently, due to the growth of deep learning, recurrent neural network (RNN), a traditional deep learning model, is widely used to mode...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
2018
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/74389 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | In clinical research, a very common task is to predict the patients’ potential critical conditions in future using the time series data collected from the patients. Recently, due to the growth of deep learning, recurrent neural network (RNN), a traditional deep learning model, is widely used to model time series data in clinical research. In this project, we proposed a novel architecture for RNN. It allows the neural network to make prediction at each time step based not only on its current input, but the previous prediction and the actual observed result of the previous time step. In our experiment, we focused on predicting the acute kidney injury for patients in ICU. And we found that our proposed methods help to improve the prediction accuracy of RNN. |
|---|