Identification of Variable Importance for Predictions of Mortality From COVID-19 Using AI Models for Ontario, Canada

Abstract

The Severe Acute Respiratory Syndrome Coronavirus 2 pandemic has challenged medical systems to the brink of collapse around the globe. In this paper, logistic regression and three other artificial intelligence models (XGBoost, Artificial Neural Network and Random Forest) are described and used to predict mortality risk of individual patients. The database is based on census data for the designated area and co-morbidities obtained using data from the Ontario Health Data Platform. The dataset consisted of more than 280,000 COVID-19 cases in Ontario for a wide-range of age groups; 0–9, 10–19, 20–29, 30–39, 40–49, 50–59, 60–69, 70–79, 80–89, and 90+. Findings resulting from using logistic regression, XGBoost, Artificial Neural Network and Random Forest, all demonstrate excellent discrimination (area under the curve for all models exceeded 0.948 with the best performance being 0.956 for an XGBoost model). Based on SHapley Additive exPlanations values, the importance of 24 variables are identified, and the findings indicated the highest importance variables are, in order of importance, age, date of test, sex, and presence/absence of chronic dementia. The findings from this study allow the identification of out-patients who are likely to deteriorate into severe cases, allowing medical professionals to make decisions on timely treatments. Furthermore, the methodology and results may be extended to other public health regions.

Design of an artificial neural network to predict mortality among COVID-19 patients

“…According to the literature, the ANN model [ [57] , [58] , [59] , [60] , [61] , [62] , [63] , [64] ] has the greatest performance in predicting COVID-19 mortality. The results of other reviewed studies also showed that ensemble ML (hybrid) models [ [65] , [66] , [67] , [68] , [69] ] and RF [ 58 , 61 , 70 , 71 ] algorithms are the most widely used and effective models for predicting COVID-19 mortality. So far, most efforts have targeted the application of ANNs and their comparison with other techniques for mortality prediction in patients with COVID-19.…”

“…In previous studies, different ML methods were trained to predict COVID-19 outcomes such as disease progression and deterioration [ 45 , 46 ], ICU hospitalization [ [46] , [47] , [48] , [49] , [50] ], and mortality [ 47 , 48 , [51] , [52] , [53] , [54] , [55] , [56] ]. The most important of these algorithms can be listed as ANN [ [57] , [58] , [59] , [60] , [61] , [62] , [63] , [64] ], ensemble models (boosting algorithms) [ [65] , [66] , [67] , [68] , [69] ], decision trees, in particular random forests (RF) [ 6 , 58 , 61 , 70 , 71 ], support vector machine (SVM) [ 58 , 61 ], and Naive Bayes (NB) [ 72 ]. According to the literature, the ANN model [ [57] , [58] , [59] , [60] , [61] , [62] , [63] , [64] ] has the greatest performance in predicting COVID-19 mortality.…”

Machine-Learning Model to Predict the Intradialytic Hypotension Based on Clinical-Analytical Data

“…The results for the areas under the curves (AUROC and AUPRC) observed in Table 6 show a very good performance of the XGBoost model compared to other developed models. Additionally, there was a statistically significant difference (p < 0.0001) in the AUROC, meaning that this model was able to correctly identify and distinguish between the two classes (IDH and non-IDH) [75]- [77].…”

Longitudinal Study of Therapeutic Adherence in a Cystic Fibrosis Unit: Identifying Potential Factors Associated with Medication Possession Ratio

“…Our results show that using machine learning algorithms, especially RF, is a promising methodology for analzying cross-sectional studies, showing robust predictive power and the ability to identify predictors of major importance. So far, this methodology has previously been used to evaluate factors with the greatest impact on high HIV viral load, COVID mortality, or presence of Bovine Viral Diarrhoea Virus [ 30 , 31 , 32 ], among others, but this is the first study of its application in CF and TA.…”