In today's digital landscape, capturing and retaining consumer attention has become an uphill battle. With over $1.1 billion spent monthly on pharmaceutical advertising alone, brands are scrambling to find innovative ways to personalize customer interactions across multiple channels. This is where machine learning (ML) comes into play, offering a powerful tool to unravel the complexities of omnichannel consumer engagement.
At this year's PMSA conference, we had the privilege of collaborating with Regeneron to present a case study that explored the role of explanatory machine learning in ethical and responsible artificial intelligence (AI) deployment. Our presentation delved into the intricacies of predicting features that drive brand engagement, uncovering valuable insights into consumer behavior and preferences.
As we navigate the ever-evolving landscape of consumer engagement, machine learning emerges as a powerful ally… However, it is essential to approach ML with a holistic and responsible mindset, ensuring ethical deployment, stakeholder acceptance, model validation, and accountability.
Karin Hayes, SVP, Analytics & Insights, OptimizeRx
The Case Study: Enhancing Patient Engagement in Ophthalmology
Ophthalmology is a sensitive and personal healthcare specialty, where patients often experience fear and anxiety about potential blindness or disability. Building trust and providing reassurance through brand messaging is crucial. To achieve this, we collaborated with stakeholders across the organization to utilize de-identified patient-level claims data to build meaningful features related to outcomes of interest.
Through an iterative approach, we refined cohorts, tuned models, and adjusted feature combinations, optimizing model performance. By incorporating diverse metrics, we gained a nuanced understanding of model strengths and weaknesses, enabling informed decision-making and continuous improvement.
After selecting our highest-performing AI models, we assessed different methods to explain how they worked, considering their strengths and weaknesses. We employed a combination of techniques, including descriptive analyses, patient journey profiling, model feature importance evaluation, and advanced ML techniques. This allowed us to understand how key features, such as patient demographics, comorbidities, and specific inflection points along the treatment pathway, impacted the machine-learning model.
The insights gained were then incorporated into a hyperlocal targeting strategy, dividing populations into microneighborhoods of approximately 10-15 households. By factoring in demographics, medical conditions, healthcare needs, and other relevant data, we facilitated highly targeted and pertinent direct-to-consumer (DTC) communication and engagement, leading to improved health outcomes for the target audience.
Key Learnings and Takeaways
Throughout this project, we uncovered several valuable lessons that can guide future endeavors in leveraging machine learning for personalized consumer engagement:
- Clear Business Objectives and Thoughtful Cohort Development: Understanding brand goals and creating well-thought-out cohorts and features are crucial for developing high-quality and explanatory models. Regularly evaluating and refining cohorts based on model performance can improve accuracy.
- Key Stakeholder Collaboration: Regular meetings and open communication with key stakeholders ensured alignment, informed decision-making, and a smooth project flow.
- Confirmation and Validation: While seeking new insights is important, confirming existing knowledge and validating assumptions are equally valuable, especially when working with successful, mature brands.
- Overlap in Important Attributes: When important attributes overlap across models, it suggests homogeneity in medical conditions or patient profiles, guiding targeted strategies and efficient model deployment.
- Application of Insights: Insights gained from machine learning can be applied across various aspects of the campaign strategy, from design to execution, optimizing outcomes throughout the campaign lifecycle.
- ML's Predictive Capabilities and Patient Experience: Machine learning's predictive abilities can enhance the patient experience and improve healthcare delivery, but it requires human intelligence, validation, and thoughtful application to maximize its benefits.
As we navigate the ever-evolving landscape of consumer engagement, machine learning emerges as a powerful ally, offering brands the ability to personalize interactions, build trust, and ultimately improve health outcomes. However, it is essential to approach ML with a holistic and responsible mindset, ensuring ethical deployment, stakeholder acceptance, model validation, and accountability.
By embracing explanatory machine learning and leveraging its insights judiciously, we can unlock the true potential of omnichannel consumer engagement, forging deeper connections with consumers and delivering personalized experiences that resonate on a profound level.
