Skip to main content
. Author manuscript; available in PMC: 2024 Mar 1.
Published in final edited form as: J Biomed Inform. 2023 Jan 5;139:104269. doi: 10.1016/j.jbi.2022.104269

Fig. 1.

Fig. 1.

Different levels of EHR data. a) displays Level 0 EHR data (i.e., data still in EHR system without any pre-processing). b) displays Level 1 EHR data after having been extracted from the EHR system and after minimal processing such as curation and harmonization. Each box in the figure represents a single medical event in the patient history. Note that Level 1 EHR data are in sequential form with structured, codified, and unstructured data and no explicit missing values. c) displays Level 2 EHR data that have been pre-processed into matrix form. Going from Level 1 EHR data to Level 2 EHR data requires significant effort spent pre-processing such as feature extraction/engineering through chart review or other means and may lead to substantial information loss. There are explicit missing values in Level 2 data due to its matrix form.