Table 2.

Comparison between the old relational system (RDB) and Graph4Med with evaluations from our end users. The figure references relate to implementation examples

Feature	RDB	Graph4Med	User comments
Overview & general cohort statistics	✗	✓ (Fig. 7)	In the RDB, the DB administrator queries for the desired information summarizing a patient cohort in a table from which plots are generated manually with third-party tools. This usually takes several weeks in our practice, especially, if the query has to be adapted after seeing the first version of the plot.
Visualization of cohort/subpopulation	✗	✓ (Fig. 8)
Visualization of relationships among patients	✗	✓ (Fig. 9a)
Navigate through individual cases	✓	✓
Gather a patient profile & visualize various data of individual cases	✗	✓ (Figs. 9b and 10)	In the old system, users have to browse multiple tables to gather a profile of an individual case and there is no possibility of visualizing relationships among various data in such a profile (cf. Fig. 10)
Eliminate redundant answers	✗	✓
Aggregate results	✗	✓
Fusion-based search	✗	✓ (Figs. 8b and 9a)	Gene fusions are the most important drivers for leukemia, and also the main bases for patient stratification, thus, searching for patients with the same fusion is very useful.
Patient similarity search	✗	✓ (Fig. 10)