The Critical Need for a Strategic AI Blueprint before Implementing AI Solutions

Artificial Intelligence is not a panacea. It will not fix every problem the pharma C suite throws at it. Even if it could, that opens the business up to a raft of pitfalls, from liability concerns to poor adoption, and poor impact and return. When AI is doing such remarkable things for pharma success, the temptation to jump on the AI bandwagon is as enticing as a once-in-a-lifetime road trip to the world’s most iconic sights. But without an AI roadmap, you will get lost before getting to the car.

The enticing bandwagon of AI vs the challenges

There’s a common theme when I enter a pharma company and see how they’ve approached AI. There’s often immense frustration. They see Artificial Intelligence as the solution, but they have addressed their solutions in an ad hoc way unaligned to their strategic objectives and without really examining it in terms of strategic alignment, potential impact and metrics.

 The problem of poor AI impact comes down to a lack of strategy and pre-strategy. We know AI is impressive, and we see the results all around us. So, boards jump on the bandwagon, ready to capitalise on those results. But many buy tech for the sake of the tech rather than consciously and strategically.

 There is a plethora of evidence as to why not having a strategic AI blueprint before you begin is problematic. First and foremost, if the project does not move the needle in some way (typically either efficiency or growth), the C suite complains about the poor impact and lack of returns. Other problems include inadequate data sets, integration difficulties, insufficient experience to get the projects through internal legal and compliance (especially those with personalised data), not addressing the change management effectively, reputational damage if poorly implemented, governance and liability concerns, and so on. An AI roadmap is critical to avoid these pitfalls.

Before strategy

When choosing AI for pharma success, it’s often the case that leadership sees the hype and wants a slice of the pie, and understandably so because the pharma companies that succeed with AI are becoming industry leaders. However, by jumping feet first blinded by hype, they overlook challenges and only look at limited products or capabilities.

 Instead, with a strategic AI blueprint, we flip our thought process. Before we consider a specific AI solution or product, even before we work out our AI strategy, we must uncover our challenges and identify our objectives. Chances are, with the plethora of AI solutions and capabilities, AI can come up trumps. However, it’s not actually the starting point.

 It takes solid C-suite leadership to stand back and avoid AI temptation, but I can promise it’s worth it and necessary. You cannot take potluck AI; you must deploy it with expertise and care.

Your starting point is understanding the corporate vision and where the gaps currently lie in achieving that vision, and only then can you start identifying where AI can offer the most significant value to healthcare professionals, patients, payers, internal stakeholders and company growth. Looking at your challenges and objectives helps to identify areas and opportunities ripe for AI transformation. Start with a simple question: Where do we want to be and what is missing to help us get there? Or even more simply  – What problem do we need to solve?

The opportunity of data

Having a strategic aim, and definable gaps to achieve serves as the beginning of your strategic AI blueprint, but it’s not the blueprint itself. Any Artificial Intelligence solution is only as successful as the data and algorithms behind it. Overlooking the need for high-quality data is another common pitfall of the AI bandwagon. Poor quality, limited and generic data will give poor, limited and generic results, no matter how capable the tech is.

As such, you need to face the data issue head-on to prevent future bafflement about the AI solution failing to deliver its promises. So, once you know the areas to focus on, data must be considered. In most circumstances, it’s not about having one big data source. Instead, it’s about triangulating multiple big data sources for a specific situation. Data sources vary for different pharma businesses looking to solve different problems in different countries.

It’s time for strategy

Once you have done a thorough examination of your corporate vision, strategic pillars,  gaps and unmet needs, you can focus on the AI strategy and draw the AI blueprint. So, what should you include?

• What do we want to achieve?
It won’t work for leadership to consider AI as ‘The tech or the data science teams problem to solve’.  Instead, they need to get involved enough to understand how the technology potentially works for the type of challenges that they are facing. The C Suite members don’t have to be experts, but they need to understand an AI solution’s capabilities in this context so they can guide the tech team’s implementation. Tech and data science are not business strategies. AI has to be led by business strategy. Otherwise, you end up trying to fit a square peg in a round hole.

• Who do you intend to serve?
Failure to identify and overlook the human element at the point of AI implementation limits its success. One needs to be very clear on who the AI is there to serve.

• A thorough understanding of the end users or customers (internal or external)
Just as one creates a patient journey for understanding the patient and how best to serve them, one also needs to have a thorough understanding of the stakeholders in the areas of identified unmet needs. These can be internal pharma clients or your external customers, or both.

• What are the parameters of this project?
I reiterate that it’s easy to wrongly think that a single AI solution can fix anything and everything. As such, it’s tempting to start any AI blueprint with a scope that’s far too wide. Instead, be clear about the parameters of each AI strategic plan within the broader overall business and department objectives. This will also help to determine AI priorities. You won’t solve everyone’s problems in one hit or in one way.

• What data do we need, and where can we find it?
The data issue may seem daunting, it’s about knowing where to look. You need to know what data sources will make the project a success and where to find them. At Eularis, being in this space for 20 years now, our reach and experience give us access to excellent data sources with hundreds of millions of big data sets – including EMR data sets for over 30 countries. We have even successfully generated synthetic EMR data and synthetic biological data that replicate real human data for rare diseases which helps clients avoid falling foul of many of the existing and future privacy regulations. So, if you cannot find data – you know who to ask. 

• What resources do you have, and what resources do you need?
Any AI solution will require a collection of new resources, specifically technical and skill-centred, so assess what resources you have and what you need to bring on board.

• What risks exist, and how can you mitigate them?
In the strategy stage, it’s vital to identify, assess and mitigate risk. In pharma, the risks of AI implementation go further than in any other industry. Every risk must be considered, from cybersecurity to compliance. And the regulations around this space are increasing constantly. For example, the Digital Services Act that will be enforced in the EU in 2024 has some wide-reaching implications for many of the pharma omnichannel Next Best Action systems that will need to be changed in advance. Keeping abreast of these changes is important when planning your blueprint.

• How will you analyse success and value?
Determine how you will measure and analyse whether the AI solution adds value and impact in the short and long term, as this helps you establish if the AI solution will live up to its promises.

• How will you integrate, refine and scale the AI solution?
Time and again, AI projects start too big and have poor buy-in. You must consider how you will use your analytics to refine and scale your AI solution. Significantly, ensure AI is fully utilised and embedded by considering how to approach communications and training.

Vitally, consider your AI blueprint as an ongoing process. It needs to adapt and flex as the AI journey unfolds. With the fledgling development of your AI blueprint, it’s finally time to consider whether you buy or build your AI solution for the recommended elements. I do not believe in reinventing the wheel so if a high-quality solution designed for pharma is on the market for parts of your roadmap, then they should b considered.

The AI solution: buy vs build

With your AI blueprint outlining core strategy and priming the pharma business for success, it’s time to consider whether you buy in the AI solution or build it with the help of an external vendor. Learn about the pros and cons of each in my article, Build versus Buy for Pharmaceutical AI: Should I Build or Should I Buy?
There are sound reasons why, in my experience, build or buy works best in different scenarios. In both, your AI blueprint will prove invaluable to ensure that functionality meets expectations. Indeed, your AI blueprint (and the steps taken to develop it) ensures that the C Suite and the tech developers or implementers speak a common language and have a common vision and goal for the company. This is the bedrock of AI development and implementation success