Predictive models have been used to target and measure healthcare professional (HCP) marketing in the pharmaceutical industry for years. Oftentimes, these tasks occur at different times of the year using different data sources.  Methodologies are misaligned across different brands, lifecycles, and levels of marketing investment.

This lack of continuity can not only create a suboptimal solution and lower-than-desired ROI, but it can also create organizational inefficiencies and disjointed decision-making.

In the approach developed, we bring together these separate analyses using artificial intelligence and machine learning to maximize the expected Rx value of each marketing communication. We set out to create a machine-learned solution to identify the most valuable HCPs, the optimal channels to target them by, and the most influential messaging for each. This will allow us to deploy a brand’s marketing budget and resources in a will maximize the ROI across all sales and marketing efforts.

Furthermore, the algorithms adjust their approach as brands transition from their nascent state to maturity. Below is a high-level graphic to showcase how the utilization and purpose of the models shift throughout the lifecycle of a brand.

This process creates a methodology that is more accurate in identifying high ROI targets. It also has a granular level that creates channel and messaging recommendations for each HCP. The algorithms combine previously disparate analyses and extract important data points to create one final normalized weighted score for each HCP across each channel. An ensemble of three components, firstly, prescription habits ‘the past’. Secondly, the expected response to marketing content, ‘the present’. Thirdly, the propensity to engage, ‘the future’.

• ‘The past’ behaviors of a doctor provide a basis for their proclivity to write RX for a given brand. For nascent brands, the machine learning algorithms create an ‘RX score’. The basis for this ‘score’ is an analysis of each doctor’s initial adoption of like brands when those brands were within their launch stages. As the brand ages, the model shifts to model the known and predictable patterns of HCP writing habits.
• The present’ is a component that determines how responsive a doctor will be to marketing efforts. The methodology to measure responsiveness is to create a promo mix response model. A promo mix model is a regression-based approach that estimates the incremental impact that prior marketing efforts have had on the prescribing of each HCP. Utilizing this score within the final model provides an estimated Rx-level impact of future marketing efforts on each doctor.
• ‘The future’ is an analysis of the doctor’s propensity to engage with the brand across the various sales and marketing channels. Called an engagement score, the core classification methodology is the same for each brand’s lifecycle. However, with young brands, the data utilized comes through several proxy variables that identify an HCP propensity to engage. With brands in their latter stages, engagement is estimated through past participation with the brand and the likelihood to continue.

The next stage of the process is to create the machine-learned RX-value score is to combine the scores of the Past, Present, and Future. As in the prior stages, there is flexibility in how much weight you give each analysis as the strategy and for the targeting changes.

The final normalized scores can be a mechanism to create a refined target list as well as provide strategic insight on how to reach each target, deciding which doctors are better reached through emails, or digital marketing can decrease wasteful rep details. Using unsupervised learning techniques, we can then segment the available HCP universe into macro and micro-segments that align to their RX-Value by channel.

The final product is a target list created by artificial intelligence that identifies the highest potential engagers, the most effective marketing channel, and the necessary effort to convert. Most importantly it is easily replicable and has the potential to bring a much stronger ROI to marketing efforts.