MULTIVARIATE IMPUTATION TANDA VITAL PASIEN UNIT PERAWATAN INTENSIF MENGGUNAKAN CONVOLUTIONAL NEURAL NETWORK
Total View This Week0
Total View This Week0
Institusion
Universitas Sriwijaya
Author
AIS'SY, WIDYA ROHADATUL
firdaus, firdaus
Tutuko, Bambang
firdaus, firdaus
Tutuko, Bambang
Subject
Q334-342 Computer science. Artificial intelligence. Algorithms. Robotics. Automation.
Datestamp
2023-05-22 08:18:46
Abstract :
Missing data is a common problem in vital sign datasets and is one of the most complex topics in computer science. The large number of missing values in the data makes it more difficult to process. Therefore, one way to address missing data is through imputation. Vital sign data imputation is the process of filling in missing or incomplete data in a patient's medical records. Machine learning for data imputation has been frequently applied, but tends to perform poorly with datasets that have high levels of missing values. Hence, deep learning methods are used, as they have the ability to unearth and capture hidden information in data that leads to progress in data imputation. One of the methods used for vital sign data imputation is the Autoencoder and Convolutional Neural Network (CNN). Recent studies have shown that CNN and Autoencoder are two methods that are quite effective in performing vital sign data imputation. In this study, the performance between Autoencoder and CNN methods in imputing vital sign data in the medical information mart for intensive care III (MIMIC III) database will be compared. The CNN resulted in the smallest RMSE value of 0.0306 while the Autoencoder resulted in the smallest RMSE value of 0.0642. The results of vital sign data imputation using CNN showed superior performance compared to Autoencoder.
Missing data is a common problem in vital sign datasets and is one of the most complex topics in computer science. The large number of missing values in the data makes it more difficult to process. Therefore, one way to address missing data is through imputation. Vital sign data imputation is the process of filling in missing or incomplete data in a patient's medical records. Machine learning for data imputation has been frequently applied, but tends to perform poorly with datasets that have high levels of missing values. Hence, deep learning methods are used, as they have the ability to unearth and capture hidden information in data that leads to progress in data imputation. One of the methods used for vital sign data imputation is the Autoencoder and Convolutional Neural Network (CNN). Recent studies have shown that CNN and Autoencoder are two methods that are quite effective in performing vital sign data imputation. In this study, the performance between Autoencoder and CNN methods in imputing vital sign data in the medical information mart for intensive care III (MIMIC III) database will be compared. The CNN resulted in the smallest RMSE value of 0.0306 while the Autoencoder resulted in the smallest RMSE value of 0.0642. The results of vital sign data imputation using CNN showed superior performance compared to Autoencoder.
File :
RAMA_56201_09011281924035.pdf
RAMA_56201_09011281924035_TURNITIN.pdf
RAMA_56201_09011281924035_0221017801_0620086501_01_front_ref.pdf
RAMA_56201_09011281924035_0221017801_0620086501_02.pdf
RAMA_56201_09011281924035_0221017801_0620086501_03.pdf
RAMA_56201_09011281924035_0221017801_0620086501_04.pdf
RAMA_56201_09011281924035_0221017801_0620086501_05.pdf
RAMA_56201_09011281924035_0221017801_0620086501_06_ref.pdf
RAMA_56201_09011281924035_0221017801_0620086501_07_lamp.pdf
Download
book
BibTex
Latex, Jabref
cloud_download
Original Resource
url resource Institution
