Title: Binary particle swarm optimisation and the extreme learning machine for diagnosing paraquat-poisoned patients

Authors: Xuehua Zhao; Xin Tian; Zhen Li; Xu Tan; Qian Zhang; Huiling Chen; Lufeng Hu; Shuangyin Liu

Addresses: Shenzhen Institute of Information Technology, Shenzhen, 518172, China ' Cancer Hospital Chinese Academy of Medical Sciences, Shenzhen Hospital, Shenzhen, 518000, China ' Liaocheng Dongchangfu District Maternal and Child Health Hospital, Liaocheng, 252000, China ' Shenzhen Institute of Information Technology, Shenzhen, 518172, China ' Shenzhen Institute of Information Technology, Shenzhen, 518172, China ' Department of Computer Science, Wenzhou University, Wenzhou, 325035, China ' Department of Pharmacy, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, 325000, China ' College of Information Science and Technology, Zhongkai University of Agriculture and Engineering, Guangzhou, 510225, China

Abstract: The diagnosis of paraquat-poisoned patients is one of the important problems in the medical diagnosis field. Current methods identify the paraquat-poisoned patients mainly depending on paraquat content in the body. However, the lack of such methods is treating paraquat-poisoned patients as a healthy person when there is little paraquat content in the body. Here, a new diagnostic method for paraquat-poisoned patients is proposed, which fuses gas chromatography-mass spectrometry, binary particle swarm optimisation and extreme learning machine together. In the proposed method, the data is collected by gas chromatography-mass spectrometry, the binary particle swarm optimisation is adopted to select the excellent feature sets and the extreme learning machine is adopted to identify the paraquat-poisoned patients. In contrast to current methods, the proposed method still can accurately identify the paraquat-poisoned patients even if there is little paraquat content in the body. In our experiments, two measures, which are accuracy and sensitivity, are used to evaluate our method. The accuracy and sensitivity get to 93.90% and 94.54%, respectively. We also made comparisons with four algorithms and the experimental results show that our method has better performance than the other four methods.

Keywords: medical diagnosis; paraquat-poisoned patients; feature selection; extreme learning machine; ELM; particle swarm optimisation; PSO.

DOI: 10.1504/IJAAC.2021.116419

International Journal of Automation and Control, 2021 Vol.15 No.4/5, pp.427 - 443

Received: 14 May 2019
Accepted: 09 Oct 2019

Published online: 23 Jul 2021 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article