Artificial intelligence approach towards assessment of condition of COVID-19 patients - Identification of predictive biomarkers associated with severity of clinical condition and disease progression

Abstract

Background and objectives: Although ML has been studied for different epidemiological and clinical issues as well as for survival prediction of COVID-19, there is a noticeable shortage of literature dealing with ML usage in prediction of disease severity changes through the course of the disease. In that way, predicting disease progression from mild towards moderate, severe and critical condition, would help not only to respond in a timely manner to prevent lethal results, but also to minimize the number of patients in hospitals where this is not necessary.

Methods: We present a methodology for the classification of patients into 4 distinct categories of the clinical condition of COVID-19 disease. Classification of patients is based on the values of blood biomarkers that were assessed by Gradient boosting regressor and which were selected as biomarkers that have the greatest influence in the classification of patients with COVID-19.

Results: The results show that among several tested algorithms, XGBoost classifier achieved best results with an average accuracy of 94% and an average F1-score of 94.3%. We have also extracted 10 best features from blood analysis that are strongly associated with patient condition and based on those features we can predict the severity of the clinical condition.

Conclusions: The main advantage of our system is that it is a decision tree-based algorithm which is easier to interpret, instead of the use of black box models, which are not appealing in medical practice.

Keywords: COVID-19; Clinical condition assessment; Personalized model; Predictive blood biomarkers; Rule-based machine learning.

Graphical abstract

Fig. 1

Graphical representation of clinical data which consist of demographic data, symptoms and blood analysis.

Fig. 2

Schematic representation of ML methodology for determination of the severity of clinical condition.

Fig. 3

Dependence between clusters and values of different blood biomarkers.

Fig. 4

Schematic representation of principles which was used for database organization.

Fig. 5

Importance scores of ten best features.

Fig. 6

Comparison between actual and predicted values of Hgb.

Fig. 7

Comparison between actual and predicted values of CRP.

Fig. 8

Comparison between actual and predicted values of creatinine.

Fig. 9

Comparison between actual and predicted values of LDH.

Fig. 10

a) Receiver operating characteristic curve (ROC) and b) Area under the precision-recall curve (AUPR).

Fig. 11

Confusion matrix with normalized (left) and regular (right) values of patients.

Fig. 12

An example of tree which is a part of XGBoost classification model.