Cognizant’s Health Sciences Innovation Day took a deep dive into how to scale and operationalise AI in the industry. Two expert sessions delivered some crucial takeaways.
Our second Health Sciences Innovation Day brought together industry experts across life sciences, data, AI and consumer health to discuss how to take AI and other new technologies from innovation to practical implementation.
Across one fireside chat and one panel session, they shared insights and experiences from the front line of innovation in a highly regulated industry. Here are three big takeaways from those sessions.
Get the data foundation right—but it doesn’t have to be perfect from day 1
One point that came across very clearly was the need to have the right data foundation in place. In the keynote, Nitish Mittal of Everest Group had spoken of the disillusionment of one large pharma company which had discovered that the data it wanted to use for AI initiatives was not accessible.
In a fireside chat, Saqib Mir, Director of Engineering for Generative AI and Data in Development Tech at GSK, spoke about GSK’s Data Fabric project, aimed at making data findable, accessible interoperable and reusable. Started four years ago before Gen AI’s potential became apparent, it has given GSK a reliable data pipeline for its AI models and Gen AI applications. The message: do the hard data engineering work first, and it will ease things down the line.
That said, waiting to have the ‘perfect’ data foundation might put companies at a disadvantage. The importance of experimenting now with the data you have was also emphasised by Saqib Mir, as well as by Cognizant’s Global Head of Generative AI David Fearne. The key thing, according to GSK’s Senior Director of Manufacturing Technology Dominic Thornton Flowers, is being able to decide whether (and when) to abandon such an experiment or scale it across the business.
Get the right people on board—including compliance and procurement
As important as the data foundation is having the right people on board. Kabir Patel, Global Head of IT at AstraZeneca, noted that the skills required to conduct small-scale pilots or proofs of concept are not the same as the skills required to scale, operationalize and manage a business-as-usual implementation of a new technology like AI. Identifying and acquiring the skills required to take the pilot to the next stage is essential.
One area of expertise that’s often missing at the experimentation and innovation stage is compliance. A strong message from the day is that involving compliance from the get-go can ensure that innovation heads in the right direction to achieve future regulatory approval.
While there’s a general feeling in the industry that compliance can put the brakes on innovation, David Fearne said it can actually act as a catalyst, since guardrails and constraints can be triggers for creative thinking. Several speakers agreed that cultivating advocates within both compliance and procurement can create a smoother path to scaling AI, given that both functions are traditionally left out of innovation conversations and as a result can become blockers down the line.
Domain expertise isn’t the only type of skill required to successfully scale AI. Salesforce’s Health Care and Life Sciences Industry Advisor Nikki Pallaras encouraged life sciences businesses to “look for the people with magic in them”—the people with the passion to drive things forward.
Those people need not always be the stellar performers. Nikki cited Google’s experiment to find the perfect team, which involved pitting an ‘A’ team of star performers against a ‘B’ team of people of mixed ability. The ‘B’ team won out because its members were more supportive of each other, thought more like a team, and gave each other space to be heard.
Get the right mindset embedded across the business
Adopting the right mindset is a third imperative for taking AI from innovation to implementation. Several speakers made the point that an atmosphere of psychological safety is necessary to allow for experimentation and failure. One theme that came up is the need to get business-side stakeholders and internal sponsors comfortable with the idea that results may not be perfect from Day 1.
GSK’s Dominic Thornton Flowers, for example, said that stakeholders need to be realistic about the work required to scale a prototype to multiple global sites. Results achieved in the early stages may not be sufficiently high-quality to scale without further investment.
Echoing this point, David Fearne compared embarking on a Gen AI project with hiring a new employee. Generally, you wouldn’t expect them to do everything perfectly from the moment they walk in the door—it takes time for them to learn what’s expected of them and how they can excel. Gen AI is no different; it requires feedback and iteration to get it delivering the results expected.
In the case of the virtual AI clinician that Cognizant developed for a national healthcare agency, for example, multiple rounds of feedback from clinical specialists were required until it was able to diagnose symptoms with 98% efficiency.
Cross-industry collaboration is key
Lastly, speakers agreed on the importance of having a collaboration mindset—within the organisation, with partners, and even with competitors.
Where uses of AI are non-differentiating, such as using Gen AI to accelerate report writing or to “talk to your data” for analytics, it makes sense for companies to share best practices so the whole industry can deliver new therapies faster and improve patient outcomes—the common goal that binds the industry together.
It’s an approach wholeheartedly taken by Salesforce, which works with the Life Sciences industry and tech partners like Cognizant to experiment with new applications of AI in the sector. You can see many of the results at Salesforce Plus.
With both industry and tech collaborating to experiment, learn and scale, the potential for discovery and innovation seems limitless.
