Read from Anita Moser, Michael Davis, Dharmendra Sahay as they outline UCB's measured approach to scaling our AI capabilities. The article was originally published in PharmExec.
 

Today’s business environment is increasingly complex, with a growing volume of data available for guiding decisions. When UCB was preparing to launch a new epilepsy product three years ago, we recognized the market had changed considerably. We could no longer rely solely on prior institutional knowledge or intuition: Healthcare had become more focused on the challenge of delivering patient-centric value, despite a decrease in access to physicians. At the same time, data analytics and computing power had grown significantly. The key to success was to pursue an agile, data-driven decision-making model that would guide our business strategy and execution.

At the time, pharma’s most common method for making this pivot was to develop sophisticated data and advanced analytics capabilities first. However, this required commitment, vision, and discipline to see the change through, which often made organizations wary of pursuing this kind of innovation in the first place. At UCB, our approach is to focus on what patients value and to enable people living with severe diseases to access the medicines they need. We believe that to achieve this goal for patients, it takes data-driven decision making that will help us reach the right physicians. To successfully scale the AI, we decided to pursue a measured approach.
 

• Starting small: Understanding holistic business performance drivers by leveraging all our available data was critical to helping us prioritize focus areas. Our first step was to integrate different primary and secondary data sources to create a 360-degree view of our customers’ needs, attitudes, behaviors, potential, and access, among other factors. Next, we focused on developing an algorithm through a supervised machine-learning approach that blended parametric modeling and tree-based methods. This algorithm showed us areas where physicans were treating patients in a similar fashion based on more than 100 different criteria. The results were eye-opening. For example, the algorithm showed that while our initial hypothesis of patient access to medicine was important, the main concern was really affordability. The results also offered a natural road map to the next stream of analytics that could improve the trajectory of our business performance.
   

• Paving the way to enabling machine learning for critical decisions: The analysis of our business performance drivers built confidence in the power of machine learning among UCB senior leadership. Next, we aimed to embed analytics in our day-to-day execution and expand the approach across the entire portfolio. We focused on using algorithms to turn reactive, one-off analysis into proactive insights that would support a host of interconnected decisions. This integrated infrastructure greatly reduced our lead time to generate insights and provided a framework for high-priority focus areas. Over time we were able to optimize promotional spend based on ROI, refine product messaging based on what we were seeing in the ecosystem, and enhance targeting strategy. As we continued to look for performance drivers and successfully address existing issues, the algorithm grew smarter, which showed us the process was working.
  
  In this step, the momentum of data-driven decision-making also began to reach the level of executive decisions, and soon leadership meetings were structured around insights gleaned from advanced analytics. Suddenly, the idea of being a data- and analytics-driven organization that deployed machine learning on a routine basis had gone from visionary to reality.
   

• Replicating the success across the portfolio and with new product launches: When UCB was preparing to launch another product, we now had a framework to drive commercial excellence through advanced analytics that we could apply to the launch. The new product was for a rare disease with no one clear definition for treatment, which made it difficult to identify in the data. Instead, we leveraged the algorithms to parse physician-entered notes that identified event and treatment markers associated with the disease. The primary insights and secondary data evidence helped us target the appropriate patient population with reasonable confidence, understand drivers of early adoption, and define a go-to-market model around patient needs and point-of-care locations.
   

• Making the leap to real-time decision-making: By 2019, we had seen tremendous progress in using advanced analytics and machine learning to enable business decisions. Our new north star was using analytics to drive actions. Our next opportunity was in leveraging analytics at the point of action, where it could have the most impact. We set a goal to replace our process for  quarterly targeting and call plan with dynamic targeting that could provide just-in-time recommendations based on predicted patient events, disease severity, the location and the propensity of physicians to prescribe our products. After four months of iterative learning, our proof of concept included a robust algorithm that recommended who to call, when to call, and what to say in real-time. By putting UCB in front of physicians with the right messaging at the right time, we were able to help those physicians better serve their patients, delivering on our core patient-centric objective. Furthermore, the algorithm continued to learn from usage and feedback, evolving into a more sophisticated AI tool.
   

By starting small, targeting the right business problems, and embedding data-driven decision making into how we operate every day, we’ve seen success. Enabling AI-based analytics capabilities and continuing to refine our algorithms and dynamic targeting has allowed us to get our medicines to the patients who need them as quickly as possible. Regardless of approach, the journey requires perseverance and patience. The most important thing we learned is that AI works if we create the right path.  

Anita Moser is UCB’s head of assets and optimization for US Neurology. Michael Davis leads UCB’s US Neurology Patient Value Unit. Dharmendra Sahay leads ZS’s integrated analytics practice.