Developing a new intelligent system for the diagnosis of tuberculous pleural effusion

Chengye Li; Lingxian Hou; Bishundat Yanesh Sharma; Huaizhong Li; ChengShui Chen; Yuping Li; Xuehua Zhao; Hui Huang; Zhennao Cai; Huiling Chen

doi:10.1016/j.cmpb.2017.10.022

Developing a new intelligent system for the diagnosis of tuberculous pleural effusion

Comput Methods Programs Biomed. 2018 Jan:153:211-225. doi: 10.1016/j.cmpb.2017.10.022. Epub 2017 Oct 18.

Authors

Affiliations

¹ Department of Pulmonary and Critical Care Medicine,The First Affiliated Hospital of Wenzhou Medical University, Wenzhou 325035, China.
² Department of Neurology, Wenzhou Hospital of Integrated Traditional Chinese and Western Medicine, Wenzhou 325027, China.
³ Department of Pulmonary and Critical Care Medicine,The First Affiliated Hospital of Wenzhou Medical University, Wenzhou 325035, China; Jawaharlal Nehru Hospital, Rose Belle, Grand-Port District 00230, Mauritius.
⁴ Department of Computing, Lishui University, Lishui 323000, Zhejiang, China.
⁵ School of Digital Media, Shenzhen Institute of Information Technology, Shenzhen 518172, China.
⁶ College of Physics and Electronic Information Engineering, Wenzhou University, Wenzhou 325035, China.
⁷ College of Physics and Electronic Information Engineering, Wenzhou University, Wenzhou 325035, China. Electronic address: chenhuiling_jsj@wzu.edu.cn.

PMID: 29157454
DOI: 10.1016/j.cmpb.2017.10.022

Abstract

Background and objective: In countries with high prevalence of tuberculosis (TB), clinicians often diagnose tuberculous pleural effusion (TPE) by using diagnostic tests, which have not only poor sensitivity, but poor availability as well. The aim of our study is to develop a new artificial intelligence based diagnostic model that is accurate, fast, non-invasive and cost effective to diagnose TPE. It is expected that a tool derived based on the model be installed on simple computer devices (such as smart phones and tablets) and be used by clinicians widely.

Methods: For this study, data of 140 patients whose clinical signs, routine blood test results, blood biochemistry markers, pleural fluid cell type and count, and pleural fluid biochemical tests' results were prospectively collected into a database. An Artificial intelligence based diagnostic model, which employs moth flame optimization based support vector machine with feature selection (FS-MFO-SVM), is constructed to predict the diagnosis of TPE.

Results: The optimal model results in an average of 95% accuracy (ACC), 0.9564 the area under the receiver operating characteristic curve (AUC), 93.35% sensitivity, and 97.57% specificity for FS-MFO-SVM.

Conclusions: The proposed artificial intelligence based diagnostic model is found to be highly reliable for diagnosing TPE based on simple clinical signs, blood samples and pleural effusion samples. Therefore, the proposed model can be widely used in clinical practice and further evaluated for use as a substitute of invasive pleural biopsies.

Keywords: Moth flame optimization; Pleural effusions; Prediction; Support vector machine; Tuberculosis.

MeSH terms

Adolescent
Adult
Aged
Female
Humans
Male
Middle Aged
Pleural Effusion / diagnosis*
Support Vector Machine
Tuberculosis / pathology*
Young Adult