Interpretable Machine Learning for COVID-19 Diagnosis Through Clinical Variables

Lucas  M. Thimoteo; Marley  M. Vellasco; Jorge  M. do Amaral; Karla  Figueiredo; Cátia  Lie Yokoyama; Erito  Marques

doi:10.48011/asba.v2i1.1590

Lucas M. Thimoteo Pontifícia Universidade Católica do Rio de Janeiro
Marley M. Vellasco Pontifícia Universidade Católica do Rio de Janeiro
Jorge M. do Amaral Universidade do Estado do Rio de Janeiro
Karla Figueiredo Pontifícia Universidade Católica do Rio de Janeiro
Cátia Lie Yokoyama Universidade Estadual de Londrina
Erito Marques Universidade do Estado do Rio de Janeiro

DOI: https://doi.org/10.48011/asba.v2i1.1590

Keywords: COVID-19 Diagnosis, Machine learning, Explainability, Interpretability, Shapley additive explanations

Abstract

This work proposes an interpretable machine learning approach to diagnose suspected COVID-19 cases based on clinical variables. Results obtained for the proposed models have F-2 measure superior to 0.80 and accuracy superior to 0.85. Interpretation of the linear model feature importance brought insights about the most relevant features. Shapley Additive Explanations were used in the non-linear models. They were able to show the difference between positive and negative patients as well as offer a global interpretability sense of the models.