Deterministic or predictive profiles? How to optimize your HCP promotional activities across channels
We’re excited to share part three of our blog series about one of the most promising methodologies for life science brands to harness: real world clinical diagnostics data to power HCP omni-channel marketing. We’ve broken up our series into five tranches to explore how early adopters are synchronizing all promotion activity based on clinical diagnostic insights for competitive advantage. If you missed the kick-off blog, check it out here!
Once clinical data sources are harmonized into one central engine to segment at the patient level and link to the treating HCP, life science marketers have two core paths to generate clinical profiles that match a brand’s indications.
Path One: Deterministic – The attributes from your clinical data sources can be queried with exact boolean logic phrases that can be matched to your therapy’s indication.
- Example: Show me all patients who had an uncontrolled A1c score above 7.5 and currently on a second line therapy this week?
Clinical diagnostic insight firms like Prognos have these exact attributes to write this rule and then link to the treating HCP. This deterministic profile (segment), could give marketers the insights required to deliver relevant branded or unbranded messages at the right time.
Path Two: Predictive – The attributes of a longitudinal clinical data source(s) can be harnessed as a training dataset to set up predictive experiments.
- Example: Show me all patients who are currently on a second line therapy for diabetes and will probabilistically move to a third line therapy one month from now?
Since scalable firms like Prognos have the deterministic data historically, the same deterministic dataset can be used to train a forward looking model to predict when a patient will more than likely require a third line therapy. Again, another tactic to help intervene at the right time.
There’s no direct answer to which path to choose. All depends on the use case and the outcomes a life science marketer is seeking to achieve. Focus on outcomes first, then work your way backwards to the best solution.
Traditionally, the majority of HCP segmentation has been designed from a deterministic perspective. Clinician-based rules that are specifically aligned to the brand’s indications. Coupled with a scalable registry or clinical information, this is an exceptional solution. Brands and entire portfolios can design multiple profiles that link by indication, by creative message, and by channel.
On the predictive side, life science firms are harnessing machine learning tools to predict when certain events may occur, such as below:
Diagnosis of a certain disease —> Treatment start or change —> Disease progression
Prognos, for instance, works with many rare disease therapeutics. A big challenge in this market is diagnosis. Bringing awareness to the disease and/or ensuring HCPs are aware of certain lab testing to treat patients is paramount.
Disease progression is another use case for applying machine learning to your diagnostic datasets. Below is an example of a third line therapy progression.
The goal of this therapy was to predict three-months in advance prior treatment change. As in all data experiments, probabilistic scores are generated (nothing is 100% in the predictive world!). But that’s ok, when this use case makes sense life science marketers can allocate reach and efficiency by marketing channel. The least efficient segment (4.6% in this example), send to your programmatic digital advertising HCP media channel. You’d want to send a higher scoring segment to your field force (the cost of people is much more expensive than sending a couple digital ads to those HCPs!).
Contact us for more information and share your omnichannel strategy with our experts.
About the Author: Matt Apprendi has been a fixture in the NYC technology scene for more than a decade, where he’s focused on conceptualizing, building and commercializing intelligent data products. As VP of Digital Solutions at Prognos, Matt is responsible for unlocking the value of clinical diagnostic data to power intelligent digital applications across the greater healthcare ecosystem.