Using ML, AI and RWD for Precision Healthcare Marketing

Real World Data increasingly informs the life sciences industry across various functions, for instance in pharmacoeconomic analysis, outcomes research and peak revenue forecasting, but often with a temporal lag that doesn’t allow for continued optimization of marketing execution. Swoop.ai has developed the industry’s first HIPPA-certified and NAI accredited system of engagement that utilizes machine learning, artificial intelligence and longitudinal analysis in conjunction with an enriched, tokenized and granular-level data pool of over 300 million de-identified patients and a behavioral data stream of over 65 billion anonymous consumer transactions. We utilize this technology to define and uncover exclusive audiences based on client-specific market definitions and strategies, and then use the outputs to fuel optimal cross-channel marketing strategies that engage the ideal patient with the ideal message through the ideal channel. The result is greater performance over conventional targeting approaches.

Let’s illustrate the system in action as exemplified by a market leader for a high prevalence indication such as diabetes or RA. No matter the scenario, when using machine learning, artificial intelligence and real world evidence, there are three major tasks required to optimize digital marketing and media activation: Define and Find your Ideal Patient Population; Engage these Ideal Patients and their Healthcare Providers; and Convert these Ideal Patients to Achieve Your Defined Performance Goals.

In our example, the branded market leader is positioned behind a first-line generic and it’s primary goal is to simply get in front of as large as possible diagnosed patient population and increase the number of patients seeking diagnosis or treatment. The easiest and most efficient solution is to advertise to diagnosed visitors - a cost per unique diagnosed visitor campaign to reach the diagnosed patient population as cost-effectively as possible– is an excellent option. First, we need to define and find the ideal patient population which, in this case, is fairly straightforward –anyone who is highly likely to have been diagnosed with the condition target by the therapy. A secondary patient population could be a model of likely pre-diagnosed individuals such as patients who may not have a diagnosis of the condition itself but who fit a model that implies the condition is likely to be diagnosed in the future. Another alternative could be specifically individuals who fit a model implying diagnosed and untreated.

Once we know the ideal patients, it’s time to engage them and their associated health care providers directly. Engaging HCPs is traditionally accomplished via the personal promotions team, but being able to effectively and efficiently activate both patients and HCPs via digital channels can dramatically increase the efficacy of a campaign.

Swoop closes this digital gap by advertising to real-world performance metrics. We determined earlier that the ideal patient population is likely all individuals with a diagnosis. Swoop can natively run campaigns optimized towards these diagnosed individuals. This would be a campaign which maximizes marketing dollars by reaching diagnosed patients as cost-effectively as possible.

In addition, because the system associates diagnosed patients with their healthcare providers, Swoop serves ads to not only an existing target list of HCPs by NPI, but to any HCP who has patients diagnosed with the condition. Messages can also be differentiated by patient or HCP type: HCPs who write scripts for the company’s therapy receive one message, HCPs who write scripts for a competitor’s therapy another; patients who are diagnosed and prescribed the company’s therapy receive adherence messaging, while patients who are diagnosed and prescribed a competing therapy receive switching messaging. Most importantly, engagement vehicles should optimize towards a real-world business goal – in this case cost per unique diagnosed visitor. Again, Swoop performs this optimization natively and is the only solution that can ingest real-world health data and perform real-time algorithmic optimization against it.

Comment Form