Patient Flow

Patient-level data allows marketers to track patients over time and across different points of care, recording the identity of the prescribing physicians as well as the drug, strength and duration of treatment.  Such data can be shown as a graph where each node represents a physician and each arc the number of times a patient moves from one physician to visit another.  The specialty of the physician is relevant when establishing if a patient movement constitutes a referral or not.  This method of viewing patient-level data is a significant departure from mainstream analyses in that the patient is regarded as a mere thread that connects physicians as opposed to being the focal point of the analysis as is the case with compliance, persistency, switching, dosing, new therapy starts, and other conventional targeting processes.

Graph Building - Step 1

Take one patient and construct a graph that tracks the physicians that patient visits.

Graph Building - Step 2

Repeat the process with the other patients. Add a new node to the graph if the physician in question is a new physician, otherwise increment the arc by one to denote the fact that yet another patient has visited that physician.

Graph Building - Step 3

The graph below is pruned to get rid of stark instances of non-referrals!

Prune non-referrals, politely called self-referrals, from the graph.

Results - Three Types of Influencers

1. Referred-to-physicians: Get a lot of referrals from other physicians(high-caliber professionals)
2. Referring physicians: Send a lot of physicians to other physicians that may favor or shun your product (average Joe – extremely valuable since they can deflect the outflow of patients, yet difficult to identify otherwise)
3. Wheeler-dealers: score high on both counts.

For additional information about this product please send an e-mail to igor@bayser.com