Authors:
Ziyue Yu
1
;
Lihua He
1
;
Wuman Luo
2
;
1
;
Rita Tse
2
;
1
and
Giovanni Pau
3
;
1
;
4
Affiliations:
1
School of Applied Sciences, Macao Polytechnic Institute, Macao, SAR, China
;
2
Engineering Research Centre of Applied Technology on Machine Translation and Artificial Intelligence of Ministry of Education, Macao Polytechnic Institute, Macao, SAR, China
;
3
Department of Computer Science and Engineering, University of Bologna, Bologna, Italy
;
4
UCLA Computer Science Department, Los Angeles, U.S.A.
Keyword(s):
Covid-19, Deep Learning, Blood Test, CNN+BI-GRU.
Abstract:
The COVID-19 pandemic is highly infectious and has caused many deaths. The COVID-19 infection diagnosis based on blood test is facing the problems of long waiting time for results and shortage of medical staff. Although several machine learning methods have been proposed to address this issue, the research of COVID-19 prediction based on deep learning is still in its preliminary stage. In this paper, we propose four hybrid deep learning models, namely CNN+GRU, CNN+Bi-RNN, CNN+Bi-LSTM and CNN+Bi-GRU, and apply them to the blood test data from Israelta Albert Einstein Hospital. We implement the four proposed models as well as other existing models CNN, CNN+LSTM, and compare them in terms of accuracy, precision, recall, F1-score and AUC. The experiment results show that CNN+Bi-GRU achieves the best performance in terms of all the five metrics (accuracy of 0.9415, F1-score of 0.9417, precision of 0.9417, recall of 0.9417, and AUC of 0.91).