Figure 1 | Scientific Reports

Figure 1

From: Use of machine learning to assess the prognostic utility of radiomic features for in-hospital COVID-19 mortality

Figure 1

Results from predictive analysis of in-hospital mortality. (a) Average feature importance of clinical and imaging features based on one hundred testing datasets with standard errors, sorted by highest feature importance in ensemble averaging. (b) Kaplan–Meier curves for in-hospital mortality, stratified by patient age and risk group (defined by the median risk score; high risk = solid, low risk = dashed); risk scores defined either by clinical or clinical plus imaging features within each age group. (c) Kaplan–Meier curves for in-hospital mortality, stratified by comorbidity burden and risk group (defined by the median risk score; high risk = solid, low risk = dashed); risk scores defined either by clinical or clinical plus imaging features within each comorbidity burden group.

Back to article page