Anomaly Detection: Increase HCP Engagement and Discover Untapped Opportunities

When Covid-19 practically eliminated face-to-face interactions for pharmaceutical sales reps, the industry trend toward digital engagement rapidly accelerated. Before COVID-19, 64 percent of meetings with pharma sales reps were held in person. Today 65 percent of meetings are held virtually. As vaccines move us back to normality, this trend isn’t going away. Eighty-seven percent of HCPs want either all virtual or a mix of virtual and in-person meetings after the pandemic ends. To successfully navigate this brave new world of remote HCP engagement, every HCP interaction must be highly targeted. Otherwise, it’s impossible to stand out from the competition. But how do you ensure that sales reps have the right insights to make every HCP interaction successful? You need to deploy advanced analytics to surface valuable insights in your data. Anomaly detection helps sales reps more precisely target HCPs who are likely to prescribe their products and understand prescribing behaviors so that every interaction counts. Here’s why anomaly detection is the key to successful HCP engagement for life sciences companies in the digital era.

What is Anomaly Detection?

Briefly, an anomaly is an unexpected deviation from normal data. Traditionally, it has been used to find things like campaign donation irregularities, healthcare fraud, and credit card fraud. In the past few years, it has been a key part of medical breakthroughs that harness artificial intelligence (AI) to identify anomalies in the patient data from end-user medical devices (e.g., pacemakers, smartwatches, glucose monitors) and scans (X-rays, MRIs, ECGs). More recently, pharmaceutical and medical supply companies have begun using anomaly detection to identify sales opportunities to gain a competitive edge.

Anomaly Detection Improves Targeting for Better HCP Engagement

Underneath the surface, HCP engagement has always been complex. Even before the pandemic, sales reps grappled with the next best action because there are so many variables when it comes to attributing causes to HCP prescribing behaviors. For example, let’s say a doctor switches from your drug to a competitor’s. What happened?

• Did the doctor need a refresher on your drug’s benefits?
• Did a patient experience an adverse event?
• Did an insurance company drop coverage of your drug?

Whether it’s a formulary issue, or patient-related, or caused by a lack of education, sales reps need to understand HCP motivations. That way, they can use specific communications and actions to boost their effectiveness. The better the action, the better the engagement, and the better the outcome for HCPs, pharmaceutical companies, patients, and even sales reps.

Example: How Anomaly Detection Provides Insights

Typically, pharmaceutical sales are measured by the number of prescriptions written, and they’re counted in three ways: 1) new prescriptions (NRx), 2) new to brand prescriptions (NBRx) and 3) the total number of prescriptions (TRx). Theoretically, if more new patient prescriptions are written, then total prescriptions should increase. But what if some doctors switch from your drug to another drug, while at the same time other doctors start writing more prescriptions due to seasonal or local events? Looking at the top-line metrics, you may assume everything is fine. However, underneath the surface, a larger shift is happening that needs to be identified and corrected right away. In this case, an intelligent anomaly detection model might figure out that even though prescriptions are on track there are switches happening that will eventually lead to lower performance in the future. The territory rep can act on this insight by setting up calls with the HCPs.

Typical Approaches to Anomaly Detection

Over the years, traditional business analytics (BI) solutions have used several different anomaly detection techniques. Early on, anomaly detection was done by manually writing rules/algorithms. More recently, some BI analytics solutions have launched prewritten algorithm toolsets aimed at streamlining anomaly detection, enabling IT to tweak predefined algorithms for their use cases.

• The downside: Generic rules-based approaches typically surface anomalies that are oftentimes fairly generic and not actionable. Most business users find it to be noise rather than a signal.

More recently, AI and machine learning (ML) BI analytics solutions use advanced anomaly detection techniques to surface insights that are not anticipated. These solutions feature models that train on internal data. Over time, models grow smarter because they learn data patterns. For example, by scanning millions of sales data points AI will identify that your East Coast market share is generally between three and five percent. So, when your market share deviates from this pattern, the system will alert you. The model will auto-calibrate to new ranges as it sees new data

• The downside: Many AI- and ML-powered analytics solutions are limited because they aren’t trained on domain-specific data sets. Machine learning models take a long time to be effective and can only do so if trained on the right dataset with deep domain expertise.  

A Better Way to Approach Anomaly Detection for Life Sciences

The key to success for life sciences is using a domain-specific AI-powered analytics solution that features predefined, prepackaged models that are tailor-made for life sciences. Combining both AI with rules-based approaches means IT doesn’t need to train models or create new algorithms from scratch. WhizAI is the only AI-powered BI analytics solution that includes predefined models trained on life sciences industry data, like prescriptions, switches, formularies, and market access. Here’s why WhizAI anomaly detection works better for life sciences anomaly detection:

1. WhizAI is industry-specific and domain-specific. It’s not a general BI or augmented analytics solution that works for industries, like manufacturing, high tech, or finance.
2. WhizAI understands life sciences data, business terminologies, and calculations right out of the box from day one. There’s no need to train the AI or tweak algorithms toolsets. (This capability is available when needed.)
3. Customers can import their own models for customized functionality (“bring-your-own model”). ‍

Final Thoughts

The pandemic caused an abrupt shift toward HCP digital engagement that’s here to stay. So it’s imperative that life sciences businesses find a way to optimize every interaction. Anomaly detection is valuable because it helps pharmaceutical sales teams understand HCP prescribing behaviors and make every interaction count. Plus, when an HCP sees that a sales rep has done the work to really know their practice and patients, they are likely to feel more connected to your brand. Learn more about how WhizAI can help you leverage the most advanced anomaly detection techniques for life sciences. Get a demo today.