Identification and Comparison of Patient Characteristics for Those Hospitalized with COVID-19 versus Influenza Using Machine Learning in a Commercially Insured US Population

Figures & data

Table 1 Effect Estimates of the Twenty Most Influential Predictors from the Gradient Boosting Model

Covariates	Change in Predicted Probability of Hospitalization with COVID–19
Northeast region	11%
45–64 years	10%
Household crowding	Non-linear positive association (estimate:11%)
Chronic obstructive pulmonary disease	−8%
Percentage of population below federal poverty level	Non–linear positive association (estimate:6%)
Fever of unknown origin	7%
Other respiratory therapy	−8%
20–44 years	6%
Obesity	5%
Acute bronchitis	−7%
Medicare Advantage	3%
Substance-related disorders	−7%
GPI-8 distinct medication count	Non–linear negative association (estimate: −2%)
Urban	7%
65–84 years	1%
Female	−5%
Azithromycin	6%
Deyo-Charlson comorbidity score	Non-linear association (estimate:0.4%)
Hypertension	2%
Glucocorticosteroids	−4%

Notes: The marginal effect estimates of a single covariate on the predicted likelihood of hospitalization with COVID-19 were obtained using partial dependence plot. For a continuous covariate, the marginal effect estimate reported the change in predicted probability of hospitalization with COVID-19 between the lowest values of the covariate and the highest values of the covariate. A positive value indicated the increased predicted probability of hospitalization with COVID-19; while a negative value indicated the increased predicted probability of hospitalization with influenza.

Figure 1 Importance ranking of the twenty most influential predictors from the gradient boosting model.

Notes: Relative importance values range from 0 to 100 and represent the proportional contribution of a covariate in predicting the likelihood of hospitalization with COVID–19 as opposed to hospitalization with influenza. Household crowding and percentage of population below federal poverty level were estimated from the American Community Survey. The reference group for age is 0–19 years; the reference for urban/rural classification is rural; the reference group for region is Midwest.

Abbreviation: GPI, generic product identifier.