Heg.IA: an intelligent system to support diagnosis of Covid-19 based on blood tests
Heg.IA: an intelligent system to support diagnosis of Covid-19 based on blood tests
复制标题
DOI：
10.1007/s42600-020-00112-5
复制
发表时间：
2021-01-07
影响因子：
--
通讯作者：
dos Santos WP
中科院分区：
文献类型：
--
作者：
de Freitas Barbosa VA;Gomes JC;de Santana MA;Albuquerque JE;de Souza RG;de Souza RE;dos Santos WP
一种新型冠状病毒SARS-CoV-2引发了本世纪最大的大流行。Covid-19已造成100多万人死亡。因此，快速准确的诊断测试是必要的。目前的黄金标准是带有DNA测序和鉴定的RT-PCR，但其结果需要很长时间才能获得。已经使用了基于IgM/IgG抗体的测试，但它们的敏感性和特异性可能很低。许多研究已经证明Covid-19对血液参数的影响。本工作提出了一种基于血液检测的新型冠状病毒诊断智能系统。从血象图和生化测试中获得的实验室参数被定义为支持临床诊断的标准，作为输入特征。然后，我们使用粒子群优化、进化算法和基于成本最小化的人工选择来选择最显著的特征。我们测试了几种机器学习方法，获得了较高的分类性能：总体准确率为95.159%±0.693，kappa指数为0.903±0.014，灵敏度为0.968±0.007，精密度为0.938±0.010，特异性为0.936±0.011。这些结果是使用经典和低计算成本分类器实现的，其中贝叶斯网络是最好的。此外，只需要进行24次血液检查。这表明了一种新的低成本快速测试的可能性。该系统的桌面版本功能齐全，可供免费使用。
A new kind of coronavirus, the SARS-CoV-2, started the biggest pandemic of the century. More than a million people have been killed by Covid-19. Because of this, quick and precise diagnosis test is necessary. The current gold standard is the RT-PCR with DNA sequencing and identification, but its results take too long to be available. Tests base on IgM/IgG antibodies have been used, but their sensitivity and specificity may be very low. Many studies have been demonstrating the Covid-19 impact on hematological parameters. This work proposes an intelligent system to support Covid-19 diagnosis based on blood testing. Laboratory parameters obtained from the hemogram and biochemical tests defined as standards to support clinical diagnosis were used as input features. Afterward, we used particle swarm optimization, evolutionary algorithms, and manual selection based on cost minimization to select the most significant features. We tested several machine learning methods, and we achieved high classification performance: overall accuracy of 95.159% ± 0.693, kappa index of 0.903 ± 0.014, sensitivity of 0.968 ± 0.007, precision of 0.938 ± 0.010, and specificity of 0.936 ± 0.011. These results were achieved using classical and low computational cost classifiers, with Bayes Network being the best of them. In addition, only 24 blood tests were needed. This points to the possibility of a new rapid test with low cost. The desktop version of the system is fully functional and available for free use.