Predicting early traumatic brain injury mortality with 1D convolutional neural networks and conventional machine learning techniques
Year: 2022
Type: article
Abstract: In this work, we compare the performance of a convolutional network with other conventional machine learning techniques, such as a multilayer perceptron neural network, decision tree, random forest, logistic regression and support vector machine, for predicting 14-day mortality in patients with traumatic brain injury, using a database obtained in a low- and middle-income country, with 529 records and 16 predictor variables. The missing values of predictor variables were filled in with techniques such as decision tree, random forest, k-nearest-neighbor and linear regression. Two methods for filling in predictor variables were proposed. In the first method, previous filled in variables were used for filling in other variables. In the second one, previous filled in variables were not used for filling in other variables. In the simulation of neural networks, several optimization methods were used, such as RMSProp, Adam, Adamax and SGDM. The best accuracy and area under the ROC, 0.859 and 0.911, respectively, were obtained with a multilayer perceptron network and with a convolutional neural network.
Source: Informatics in Medicine Unlocked
Institution: Universidade Federal do Amazonas
Language: English
Subfield: Health Information Management
Field: Health Professions
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $1,500
OpenAlex ID: W4283156164