Figure 1 | Scientific Reports

Figure 1

From: Preparing for the next pandemic via transfer learning from existing diseases with hierarchical multi-modal BERT: a study on COVID-19 outcome prediction

Figure 1

An illustration of multi-modal data sources observed over the course of a COVID-19 patient’s stay in the hospital. The colors indicate diagnosis (purple), drugs (green), procedures (gray), and numeric lab measurements (blue bars). Different data modalities are observed at varying frequency in raw patient data, with lab measurements being the most sparse across patients and across time. \({\textsc {TransMED}}\) reduces the impact of sparsity by utilizing all modalities of data in a given time interval (e.g., 24 h), creating more informed patient state snapshots in time.

Back to article page