Table 2: Modalities of electronic health record (EHR) data used for predicting mortality due to COVID-19.

For each modality of EHR data, the description and number of features are included in the table.

Modality	# features	Description
Admission	23	Baseline information collected from patients at hospital admission, including age, race/ethnicity, sex, and some vital signs, such as oxygen saturation and body temperature.
Co-morbidities	22	Simultaneous presence of other common conditions in patients, such as asthma and obesity, that can affect COVID-19 trajectory and outcomes.
Vital signs	9	Clinically appropriate maximum and/or minimum values of heart rate, body temperature, respiratory rate, diastolic and systolic blood pressure, and oxygen saturation level that define a patient’s clinical state during hospitalization.
Laboratory tests	44	Values of laboratory tests, such as the number of white blood cells and platelets in blood, conducted on samples taken from patients during the course their hospitalization to monitor their internal health.