Figure 2.

Each matrix entry, M(i, j), represents the conditional probability $Pr(patient=j|patient=i)$, ie., the probability that a patient is diagnosed with a disease j, given that they already suffer from disease i. The left panel plots data for all patients, while the right shows only those patients are considered who were diagnosed with both the diseases i and j. In other words, given that a patient has been diagnosed with both diseases i and j, how likely is that the patient was diagnosed with disease i before the patient was was diagnosed with disease j.