Explaining machine learning based diagnosis of COVID-19 from routine blood tests with decision trees and criteria graphs

Abstract

The sudden outbreak of coronavirus disease 2019 (COVID-19) revealed the need for fast and reliable automatic tools to help health teams. This paper aims to present understandable solutions based on Machine Learning (ML) techniques to deal with COVID-19 screening in routine blood tests. We tested different ML classifiers in a public dataset from the Hospital Albert Einstein, São Paulo, Brazil. After cleaning and pre-processing the data has 608 patients, of which 84 are positive for COVID-19 confirmed by RT-PCR. To understand the model decisions, we introduce (i) a local Decision Tree Explainer (DTX) for local explanation and (ii) a Criteria Graph to aggregate these explanations and portrait a global picture of the results. Random Forest (RF) classifier achieved the best results (accuracy 0.88, F1–score 0.76, sensitivity 0.66, specificity 0.91, and AUROC 0.86). By using DTX and Criteria Graph for cases confirmed by the RF, it was possible to find some patterns among the individuals able to aid the clinicians to understand the interconnection among the blood parameters either globally or on a case-by-case basis. The results are in accordance with the literature and the proposed methodology may be embedded in an electronic health record system.

Machine learning application for the prediction of SARS-CoV-2 infection using blood tests and chest radiograph

“…More recently several machine learning based approaches have been published demonstrating more broader applicability in COVID-19 related applications including triage assessment 27 , severity classifcaiton 28 , 29 , risk prognostication including mortality 30 as well as applying to multi-omics data 31 . For example, a similar approach was tried with similar findings also with an attempt for explanability similar to our study 32 . This study used decision trees and criteria graph whilst our study used SHAP analysis.…”

PDTX: A novel local explainer based on the Perceptron Decision Tree

“…The literature has presented a set of terminologies associated with the XAI field, some of them consolidated in the works of [4], [6], [11], [14].…”

“…In medicine, for example, the utilization of some ML/DL approaches is still sensitive because the decisions might affect people's lives and health. Hence, the clinicians must be able to understand why the model made such a prediction [4]. Besides that, there is also a movement of the governments in the creation of new rules and laws to protect user's data and measure the impacts and consequences of AI-based decisions.…”

Feature-extraction and analysis based on spatial distribution of amino acids for SARS-CoV-2 Protein sequences

“…The SARS-CoV-2 is classified as -CoV [ 10 ] and has received widespread research attention across the world [ [11] , [12] , [13] ]. Every day, new genome sequences, as well as primary protein sequences of SARS-CoV-2, are being added to databases, such as the NCBI virus database [ 14 , 15 ] As of this writing, no antiviral drugs with proven efficacy nor vaccines for CoV2 prevention have been reported [ 16 , 17 ], while researchers have yet to attain a complete understanding of the molecular biology of SARS-CoV-2 infection [ 18 , 19 ]As a result, COVID-19 cases increase and have reached a global pandemic level, thus urgently requiring in-depth knowledge, infection mechanism, and other aspects of the virus-like forecasting its progression [ 18 , 20 ]. Although various protein-protein interactions (PPIs) of the virus and host are known, its viral infection mechanism is not fully understood [ 21 , 22 ]Therefore, identifying interactions between the SARS-CoV-2 virus proteins and host proteins will largely help to understand this mechanism and further develop treatments and vaccines [ 23 ].…”