By using website you agree to our use of cookies as described in our cookie policy. Learn More

TDWI Upside - Where Data Means Business

Decision Intelligence Is the Key to AI/ML Adoption in Life Sciences

Technologies such as AI and ML have the potential to radically transform the way life sciences businesses research, develop, and deliver life-saving medical treatments and devices.

A wide variety of organizations are pursuing digital transformation using artificial intelligence (AI) and machine learning (ML). However, despite years of sustained investment, many industries are struggling to realize justifiable ROI. This includes the highly complex and regulated life sciences sector. In fact, in a survey of life sciences organizations, 30 percent of respondents said their top challenge was identifying viable AI/ML use cases and 18 percent said they face challenges to prove the business value of their AI/ML solutions.

For Further Reading:

CEO Perspective: Your Enterprise and the Future of AI

Why Emotion AI Is the Key to Mental Health Treatment

How Process Mining Can Aid Your Digital Transformation

Data strategy leaders have a clear imperative to overcome this disconnect within their organizations. Technologies such as AI and ML have an impact on administrative efficiency, from optimizing clinical trial planning and design to patient-trial matching and enhancing safety in post-market surveillance. As a result, they have the potential to radically transform the way life sciences businesses research, develop, and deliver life-saving medical treatments and devices. To accomplish this, leaders must demonstrate the personalized value of AI/ML in their individual organizations, as well as their respective capability to reliably scale the technology over time.

Fortunately, the growing field of decision intelligence (DI) has emerged as a tool to help achieve this.

What is Decision Intelligence?

Derived from Google Research and Development, DI combines behavioral research with AI/ML-based decision making for data-driven technology application. It enables the structured discovery of decision points and their decision context, which incorporates who the decision maker is, the default decision in the absence of data, what alternatives a user might choose, and what supporting data could be used to validate an alternative.

By analyzing business processes and workflows to identify how and where employees make decisions and the context of those choices, leaders can better understand where AI/ML can improve decision-making processes. As a result, DI can help identify where AI/ML or other statistical and rule-based techniques enable users to make better decisions. Over time, DI learns users' preferences as they relate to trusting and consuming AI/ML and adapts accordingly for maximum adoption.

It has been shown that humans are not ideal decision makers, both because we are influenced by internal biases and because we tend to use mental shortcuts, known as heuristics, that enable us to circumvent rational processing of complex information. By contrast, we are highly skilled at making personal connections. The application of DI enables humans and AI/ML to adapt to the strengths of the other and focus respectively on their areas of expertise. This, in turn, drives better business value and ROI from AI/ML investments.

Use Case Discovery: Driving AI/ML with DI

By mapping user workflows and business processes and examining the decision context, as well as the heuristics and biases that influence business-critical decisions, leaders can employ DI to assess how to support better decision making and evaluate tools and resources to help. In turn, they can leverage DI to uncover key functions within the organization where AI/ML can be implemented for optimal adoption and ROI.

Employee insight is essential to the success of the developmental cycle for decision support and intelligence. Leaders should identify key personas as they are discovering decision contexts. These are key inputs for determining the requirements for improved decision support. It's also critical to understand the importance of dimensions such as diversity, fairness, explainability, and actionability. These inputs help define the requirements and the deliverables for the AI/ML projects, as well as aid in determining how to deliver insights to support the user's trust in the AI/ML recommendations.

Still, building AI/ML applications that users trust and want will not automatically guarantee adoption. The user experience (UX) of AI/ML-generated insights lags behind the UX for traditional tech. Thus, the adoption of AI/ML can be greatly enhanced by integrating AI/ML output in ways that users can easily understand, such as in a dashboard. However, the preference for how to consume information may vary by user persona. For this reason, it's crucial to analyze the personas and weigh the relative cost of building the tools that best integrate into each user type's workflow.

To be truly successful, AI/ML must be complemented by precisely defined key performance indicators (KPIs), both from the machine learning perspective and that of the users. These criteria help define and focus the goals of the AI/ML and provide a business justification for continued investment.

A Path to Better AI/ML Implementation

Human nature is such that we will never be able to fully predict what will encourage staff to adopt new tools and processes. However, DI offers a critical bridge to better make these judgment calls based on real-world patterns in behavior. This will contribute to targeting AI/ML on providing decision support where it's most impactful. By taking these data-driven steps to ensure the best possible implementation, AI/ML can do what it was created to do: support better, faster, and more personalized decision-making across industries, including life sciences.

About the Author

Lucas Glass is the vice president of the IQVIA Analytics Center of Excellence (ACOE). The ACOE is a team of over 200 data scientists, engineers, and product managers who research, develop, and operationalize machine learning and data science solutions within the R&D space. Lucas has launched over a dozen machine-learning offerings within R&D, such as site recommender systems, trial-matching solutions, enrollment-rate algorithms, drug target interactions, drug repurposing, and molecular optimization. Lucas’ machine learning research, which is dedicated to R&D, has been accepted at AAAI, WWW, NIPS, ICML, JAMIA, KDD, and many others. You can contact the author via LinkedIn or Twitter.

TDWI Membership

Accelerate Your Projects,
and Your Career

TDWI Members have access to exclusive research reports, publications, communities and training.

Individual, Student, and Team memberships available.