An Interpretable Model-Based Prediction of Severity and Crucial Factors in Patients with COVID-19

Abstract

This study established an interpretable machine learning model to predict the severity of coronavirus disease 2019 (COVID-19) and output the most crucial deterioration factors. Clinical information, laboratory tests, and chest computed tomography (CT) scans at admission were collected. Two experienced radiologists reviewed the scans for the patterns, distribution, and CT scores of lung abnormalities. Six machine learning models were established to predict the severity of COVID-19. After parameter tuning and performance comparison, the optimal model was explained using Shapley Additive explanations to output the crucial factors. This study enrolled and classified 198 patients into mild ( n = 162 ; 46.93 ± 14.49 years old) and severe ( n = 36 ; 60.97 ± 15.91 years old) groups. The severe group had a higher temperature ( 37.42 ± 0.99 °C vs. 36.75 ± 0.66 °C), CT score at admission, neutrophil count, and neutrophil-to-lymphocyte ratio than the mild group. The XGBoost model ranked first among all models, with an AUC, sensitivity, and specificity of 0.924, 90.91%, and 97.96%, respectively. The early stage of chest CT, total CT score of the percentage of lung involvement, and age were the top three contributors to the prediction of the deterioration of XGBoost. A higher total score on chest CT had a more significant impact on the prediction. In conclusion, the XGBoost model to predict the severity of COVID-19 achieved excellent performance and output the essential factors in the deterioration process, which may help with early clinical intervention, improve prognosis, and reduce mortality.

Active pulmonary tuberculosis and coronavirus disease 2019: A systematic review and meta-analysis

“…We identified 3,375 publications from our literature search ( Fig 1 ). We finally selected 43 studies, describing 236,863 patients with COVID-19, for data synthesis [ 35 – 77 ]. Thirty-three (76.7%) of them provided information on one or more of the adverse clinical outcomes of interest ( Table 1 ).…”

“…Twenty studies with 24,371 COVID-19 patients, of whom 161 (0.7%) had tuberculosis, provided information on severe COVID-19 [ 35 , 38 , 42 , 44 , 46 , 47 , 49 , 51 , 53 – 58 , 70 , 73 – 77 ]. All, except four (20.0%), of these publications were from China [ 35 , 42 , 49 , 58 ].…”

“…All, except four (20.0%), of these publications were from China [ 35 , 42 , 49 , 58 ]. Severe COVID-19 was defined based on World Health Organization guidance in three (15.0%) studies [ 35 , 49 , 58 ], recommendations from international professional bodies in two (10.0%) studies [ 44 , 70 ], national guidelines in 11 (55.0%) studies [ 38 , 42 , 46 , 47 , 51 , 53 , 56 , 57 , 73 , 74 , 76 ], and institutional policy in four (20.0%) studies [ 54 , 55 , 75 , 77 ]. Only three (15.0%) studies were considered high quality [ 42 , 54 , 58 ].…”

“…Of the 3431 patients with severe disease in the included cohorts, 36 (1.0%) had underlying tuberculosis. Only four (20.0%) studies reported a RR for severe COVID-19 that significantly exceeded 1.0 ( Fig 3 ) [ 42 , 46 , 56 , 77 ]. COVID-19 patients who also had tuberculosis were 1.46 (95% CI 1.05–2.02) times more likely to develop severe COVID-19 as compared to COVID-19 patients without tuberculosis ( Fig 3 ).…”

Comparative Analysis of Machine Learning Methods to Predict Growth of F. sporotrichioides and Production of T-2 and HT-2 Toxins in Treatments with Ethylene-Vinyl Alcohol Films Containing Pure Components of Essential Oils

“…In the literature, it has been successfully used to predict mortality of patients with acute kidney injury in intensive care unit [ 69 ]. Moreover, after comparison with ML algorithms, e.g., logistic regression (LR), kNN, decision tree (DT), SVM, and RF, the XGBoost algorithm combined with clinical information, laboratory tests, and other features, aimed at predicting the possibility of COVID-19 patients becoming severe and critically ill, provided excellent performance [ 70 ]. In another study, various ML algorithms (SVM, NN, RF, DT, LR, and kNN) have been compared to predict the mortality rate in patients with COVID-19 using 10-fold cross-validation; all models provided accurate results in the range of 86.87–89.98%, although the NN ranked first, followed by kNN, SVM and RF [ 71 ].…”