Uncovering clinical risk factors and prediction of severe COVID-19: A machine learning approach based on UK Biobank data

Abstract

Background: COVID-19 is a major public health concern. Given the extent of the pandemic, it is urgent to identify risk factors associated with severe disease. Accurate prediction of those at risk of developing severe infections is also important clinically. Methods: Based on the UK Biobank (UKBB data), we built machine learning(ML) models to predict the risk of developing severe or fatal infections, and to evaluate the major risk factors involved. We first restricted the analysis to infected subjects, then performed analysis at a population level, considering those with no known infections as controls. Hospitalization was used as a proxy for severity. Totally 93 clinical variables (collected prior to the COVID-19 outbreak) covering demographic variables, comorbidities, blood measurements (e.g. hematological/liver and renal function/metabolic parameters etc.), anthropometric measures and other risk factors (e.g. smoking/drinking habits) were included as predictors. XGboost (gradient boosted trees) was used for prediction and predictive performance was assessed by cross-validation. Variable importance was quantified by Shapley values and accuracy gain. Shapley dependency and interaction plots were used to evaluate the pattern of relationship between risk factors and outcomes. Results: A total of 1191 severe and 358 fatal cases were identified. For the analysis among infected individuals (N=1747), our prediction model achieved AUCs of 0.668 and 0.712 for severe and fatal infections respectively. Since only pre-diagnostic clinical data were available, the main objective of this analysis was to identify baseline risk factors. The top five contributing factors for severity were age, waist-hip ratio(WHR), HbA1c, number of drugs taken(cnt_tx) and gamma-glutamyl transferase levels. For prediction of mortality, the top features were age, systolic blood pressure, waist circumference (WC), urea and WHR. In subsequent analyses involving the whole UKBB population (N for controls=489987), the corresponding AUCs for severity and fatality were 0.669 and 0.749. The same top five risk factors were identified for both outcomes, namely age, cnt_tx, WC, WHR and cystatin C. We also uncovered other features of potential relevance, including testosterone, IGF-1 levels, red cell distribution width (RDW) and lymphocyte percentage. Conclusions: We identified a number of baseline clinical risk factors for severe/fatal infection by an ML approach. For example, age, central obesity, impaired renal function, multi-comorbidities and cardiometabolic abnormalities may predispose to poorer outcomes. The presented prediction models may be useful at a population level to help identify those susceptible to developing severe/fatal infections, hence facilitating targeted prevention strategies. Further replications in independent cohorts are required to verify our findings.

Machine Learning-Based Model to Predict the Disease Severity and Outcome in COVID-19 Patients

“…A number of studies focused on prediction of severity or mortality have noted that the age is one of the top features that helps to predict the severity of cases [10][11][12][13]. In our study, age was ranked among top 10 features across all 25 features used in our prediction model.…”

“…Similarly, Wong and So [11] also used XGB with another dataset to predict the severe and the death cases and identify the risk factors associated with COVID-19. e dataset was retrieved from United Kingdom Biobank (UKBB) and includes 93 different variables collected between 16 March 2020 and 19 July 2020.…”

Exploring drugs and vaccines associated with altered risks and severity of COVID-19: a UK Biobank cohort study of all ATC level-4 drug categories

“…An advantage of using ML models is that non-linear and complex interactions can be considered, which may improve predictive performance over logistic models. We employed the same set of predictors as our previous work, and followed the same analysis strategy of hyper-parameter tuning and cross-validation to obtain predicted probabilities (please refer to 18 for details). Beta-calibration 19 was performed and the resulting average AUC was 0.622.…”

“…We took reference to the approach described in 16 to analyze the data with IPW. Following our recent work 18 which aims to predict COVID-19 severity with machine learning (ML), here we also employed an ML model (XGboost) to predict Pr(tested) based on a range of factors. An advantage of using ML models is that non-linear and complex interactions can be considered, which may improve predictive performance over logistic models.…”

“…We employed the same set of predictors as our previous work, and followed the same analysis strategy of hyper-parameter tuning and cross-validation to obtain predicted probabilities (please refer to 18 for details).…”