Considerations for Integrating Artificial Intelligence

ACT: What should industry stakeholders be keeping top of mind when implementing artificial intelligence (AI) into their workflows?

Lacroix: When we started thinking, or when I was brought into the initiatives at eClinical to start thinking about artificial intelligence and where we want to apply it; immediately you start thinking, I want to do it everywhere, right? I think it's critically important to focus on where are your inefficiencies and redundancies, and where can you automate repetitive tasks that aren't adding value. The thinking is if there are things that can be done, and again, I'm thinking about data management, if there are things that can be done with rules that are successful, there's no need to necessarily innovate and automate everything—focus on the areas that truly are redundant, time and effort, and resource intensive and aren't adding value.

I would say, for organizations that are thinking about implementing, think about starting small. Think about what outcomes you're trying to achieve and set a roadmap of what are your objectives and what are you expecting that innovation to accomplish? The other is ensuring that you have high quality data to be able to introduce artificial intelligence into your organization. I think there's the need to have a clinical data management platform with data that you can use to train models and to build out your AI workflows is really important and I think that oftentimes, as organizations are engaging on this journey, they aren't necessarily thinking also about the investment in artificial intelligence. It's not a magic bullet. It takes time to build out these models. It takes time to refine these models and interestingly, our industry survey at the end of last year was respondents expected within 12 months to be able to see significant benefits from introducing AI and ML, yet the majority of them hadn't even started on their AI/ML journey.

I think being realistic about, again, what you're trying to achieve, and putting some realistic timelines around what is acceptable, having the investment from the organization into those initiatives, and then consider the people in the process as well. As I said, AI is a technology. It has infinite potential to impact the life sciences so significantly, but yet, it's not going to solve all the world's problems alone. You need to think about the individuals that are going to be using the tool in the processes around the artificial intelligence and machine learning, so that individuals are truly optimizing the technology and you're achieving the business outcomes that you want to achieve. I think people often forget about the processes and the people when they're talking about implementing AI into their workflows, because it does and it will significantly change how people are functioning in their roles and in their day-to-day.