AI transforms life sciences as we know it, yet human validation remains vital – for now

With Sam Laermans, Head of Digital Advisory, NNIT

Nearly half of the respondents in NNIT’s Expectation Barometer 2023 survey place heightened integration of AI at the top of their priority list. However, prior to fully embracing AI, life sciences companies should prioritize the people and business processes that will become AI’s closest allies.

In the traditionally cautious and conservative life sciences industry, the adoption of groundbreaking technologies often trails behind other sectors.

Yet, paradoxically, it is in the field of life sciences where the transformative potential of emerging technologies like artificial intelligence (AI) is most significant – all the way from clinical trials to drug development and post-market approvals.

However, using AI in life sciences raises profound questions and possibilities:

What happens when AI does not just assist in the laboratory, but also contributes to the medicine we consume? And how can you as a pharma company effectively use AI to optimize business processes across business units in a secure and compliant manner?

We have asked Sam Laermans, Head of Digital Advisory at NNIT, to share his thoughts on the strengths and pitfalls of welcoming AI into the realm of life sciences.

Prep before AI: fix the fundamentals

Data, both in terms of volume and format, is crucial in enhancing business processes through AI. And you will most likely discover significant diversity in data quality and reliability:

"The data might have been generated in the 1990s, perhaps in a spreadsheet, or scanned using a traditional copy machine. It could also be generated automatically in a modern cloud-based system. There’s a huge variety when it comes to the way data is produced and processed, explains Sam Laermans."

One issue is the lack of homogeneity within the data. Another is data access:

"Consider a worker on the shop floor in a large manufacturing hall who’s calibrating his machine, or a staff member responsible for reviewing regulatory documents prior to FDA submission. Their interaction with data differs greatly," Laermans notes and continues:

"The person in the hall might use a computer that still runs on an old operating system, and his biggest concern right now might be a poor Wi-Fi connection, while his colleague in the back office uses a sophisticated cloud-based system and no longer relies on paper records."

Therefore, Laermans advises, fix the operationing system and identify the data needs of different departments before making your AI investment.

AI unchained: the non-data driven approach

As previously mentioned, determining your needs is crucial before embarking on AI. However, what happens if your data is entirely unmanaged? Or if you don’t have a coherent strategy in place?
NNIT’s Expectation Barometer 2023 survey indicates that only half of all participating life sciences companies have a corporate data strategy.

According to Sam Laermans, lack of a data strategy should not be a barrier to getting started with AI:

"There are so many experts and pundits out there saying that ‘if you don’t have control over your data, you’re not ready to start your AI journey’, which is probably right to a certain degree. However, in my opinion you can also use AI to improve your data," he says and gives an example:

"Let’s say your company has a thousand documents in various languages, but you don’t know what’s in them and how many of them have actually been signed off. In this case, AI can be used to solve the problem. So, my advice is: don’t let poor data or the absence of a data strategy be a showstopper. It is perfectly fine to build an AI strategy and begin to move forward without having your data fully under control."

Sam Laermans suggests perceiving AI as the capabilities required to solve your tasks, similar to when you employ staff, while the data strategy is more about ensuring that the right people have access to the right data at the right time.

With great potential comes great responsibility

One thing is getting the data foundation in order. Another is the people interpreting and validating the data. Because no matter how great a potential AI comes with, we must not underestimate that the human brain is still vital in a pharma company:

"I’m still waiting for a company to claim their drug was entirely developed by AI from start to finish, and the auditor raised no concerns," states Sam Laermans, further explaining:

"Pharma 4.0 is very much about the human dimension in a digital world, because at the end of the day, people making critical decisions are vital for the life sciences industry. While AI is progressing rapidly, no medications have yet been approved by the FDA that were solely developed through AI."

But even though the human brain is still indispensable, paradoxically it also has the ability to intervene and disrupt AI’s ability to deliver accurate and representative results:

"An AI system, built and trained by humans, is influenced by the biases inherent in its human-generated data. This underscores the need for human critical scrutiny and judgment, such as ensuring a medicine is tested on a representative group of people," says Laermans.

The future of life sciences

AI has truly proven its potential, and the technology is going to have a profound impact on all of us:

"Just look at the adoption rate of ChatGPT in 2023, you have seen everyone using it, from my mom using it for spaghetti sauce recipes to my colleagues for debugging issues in the Microsoft Power Platform. It just goes to show how quickly the way we interact with computers is changing," says Laermans.

But where does that leave the life sciences industry?

"I’m a firm believer that in the upcoming years we will see drastic changes to the way companies in the life sciences companies work. The healthcare industry produces more data than Netflix, Disney, Facebook and Instagram combined. Therefore, I see the life sciences sector as a major candidate for embracing AI," says Sam Laermans and concludes:

"In the future, I believe many pharma companies will increasingly view themselves as tech companies that also happen to produce medicine."

Sam Laermans’ 3 tips for digital leaders in life sciences

1. Identify the AI potential

Before making your AI investments, you should 1) clarify where AI can step in and boost efficiency or restructure work processes for the better and 2) address the challenges that stand in the way of getting started with AI. For instance, it's pointless to provide the employees on the shop floor with Chat GPT if they are still operating on Windows 98 or having trouble establishing a stable Wi-Fi connection.

2. Delegate data responsibility

I believe a data strategy set up solely by the IT department is destined to fail. It might be successful from a cybersecurity standpoint, but when you ask the organization if their lives have become easier with the new data strategy, the likely response will be, 'what strategy?'. Therefore, let the business units be assigned as data owners and the IT teams as data enablers. This approach ensures everyone contributes, with each team taking part in data ownership.

3. Use AI to fix your data issues

If you’re lacking a polished data strategy, or even if you’re not fully in control of your data, it doesn't imply that AI is beyond your grasp. Rather, AI can be instrumental in aiding with the cleanup process and in enhancing and organizing your data.