Our latest state of artificial intelligence (AI) in life sciences survey takes the pulse of 177 U.S.-based industry executives to help you understand where the use of AI is trending and what peer companies are doing, in R&D and beyond. Here’s the latest:
You might think AI in life sciences is only about finding new drug candidates. But, it’s also about bigger, structural trends. Leaders need to fight decreasing margins, harness insights from consumer health data and support ways to shift care delivery outside of traditional settings.
In our survey, life sciences executives ranked where they think AI will have the most significant benefits to the industry within the next five years. New drug candidates and real-world data top the list for R&D. Interest in AI for commercial operations also shows promise. AI-driven personalization and customer insights ranked number one and two.
AI has become less of a side interest. Half (56%) of life sciences leaders agree that their company has the right management support to bring more AI to the business.
Executives say their top leadership priority is to show how AI supports the business strategy. Priority number two is getting past proof-of-concept stages to embed the use of AI across the enterprise.
Four out of ten leaders (44%) told us their company is good at identifying which uses of AI add business value. Over the next 12 months, 53% plan to explore use cases for clinical trials, but experiments in other areas show traction too.
With all this activity, leaders are seemingly undeterred by estimates that well-trained data analysts and scientists spend nearly half of their time on data preparation and deployment.
Fixing this problem alone could help make data more portable and interoperable while also addressing a set of interrelated concerns about training data, skills, recruitment and change management. Just 34% are confident in their company’s ability to manage data and infrastructure, basic foundations for AI.
To take steps toward scale and value, life sciences executives must look outward to other industries.
Almost two out of three respondents (59%) acknowledge that their organization is in a trial stage—identifying or testing proof-of-concept ideas in some workflows. Only 10% have working models in production.
Given the low deployment rate, we expect few will focus on the critical aspects of operational AI. Looking ahead, less than a third of respondents expect to focus on either AI governance (26%) or corporate ethics for AI (16%) into 2022.
Recognize the necessity of a deployment mindset for getting projects over the proof-of-concept hump.
Have a roadmap for deployment from the beginning. AI deployment needs very different skill sets than model building and training, from tech ops to infrastructure and more. Should your proof-of-concept succeed, this change in thinking will help you orient around deployment and understand the relative time commitment and skills for deployment phases.
Balance model perfection with the risk of losing valuable momentum. Model builders tend to want conditions to be perfect to consider deployment. They want their models to be highly accurate, running on the most up-to-date systems and have data with fewer flaws. But AI models improve with time and companies can lose valuable momentum delaying a deployment phase.
About this survey: ZS surveyed United States-based executives, directors and managers at life sciences organizations in August 2021 about their outlook for using artificial intelligence. Responses represent 177 participants in the survey; two-thirds of respondents are director level or above.