Figure 1. The stated objective of Robocup is for a team of robots to beat the winner of the human Word Cup by 2050.
The revolution of Artificial Intelligence (AI) is on. According to estimates from McKinsey and PwC, its impact on our global economy could range from $4.4 trillion (for generative AI only) to $15 trillion (for all use cases). By comparison, US healthcare spending reached $4.3 trillion in 2021. And one question immediately comes to mind: will AI help fix our broken healthcare systems? Closer to home, will AI help solve market access challenges, and eventually replace us, ‘the experts’?
It didn’t take long for entrepreneurs and vendors to get in there. The offering of end-to-end automatized AI-powered tools for market access is booming. From pricing predictive analytics to patient subgroup identification, and all the way to dossier writing, your worst market access nightmares are one subscription away from the ultimate solution.
Well, not so fast...
1. Market access is data-hungry: AI can help synthesize high volumes of data as often required by HTA bodies.
2. Market access requires a lot of writing: Generative AI can bring significant productivity gains to value dossier development and HTA submission.
3. Market access needs the right pricing strategy: AI-powered predictive pricing models are coming of age.
4. Market access involves contracting with payers: Contract generation has been identified by analysts as a highly feasible application of Al .
It may very well be that, in the future, some or all of market access activities will be performed by bots. But in my opinion, that future is distant.
Hopefully when I’m retired. Here is why:
Al sucks at strategy
According to McKinsey, when looking at the potential impact of AI on each functional area, strategy and pricing are predicted to be the least impacted. By opposition, customer operations, software engineering, marketing, and sales will see the greatest impact. I believe that market access, at its core, is about strategy development. Hence, the goal of fully automatizing it sounds elusive.
There is not much of a learning sample
The biopharmaceutical market is incredibly fragmented, segmented, and targeted. And now it is becoming personalized. There aren’t many patterns to learn from for those poor little algorithms. These gluttons need to eat through a lot of data to be able to regurgitate something predictive. That’s why their playground is banking (lots of transactions) or drug discovery (lots of molecules). For each new product launch, the combination of indication, target population, mechanism of action, and line of therapy is unique and often differs per market. It makes it nearly impossible to teach an algorithm if and how a new product can achieve market access success. This is why the demos from AI vendors fail.
The data is just one part of the equation
Data rarely suffice to explain market access decisions. If it did, coverage and pricing decisions would be consistent within (and in many cases across) geographies and jurisdictions. After all, the data is the same everywhere. Payers’ decision are often case-dependent with policymakers, politicians, advocates, and other pressure groups playing a key role. These intangibles are more difficult to model for a machine, no matter how sophisticated it is.
Zero tolerance for approximations
Market access is the last gatekeeper between a medicine and patients. HTA bodies, payers, and pharma market access professionals have a huge responsibility and a duty of accuracy. At the moment, AI can’t draw a hand and makes up scientific references when it can’t find one. This is not good enough for market access. AI cannot replace market access professionals until there is appropriate safeguarding for patients.
Undeniably, AI is set to have a huge impact on market access. Entire segments of the market access value chain will be revolutionized through productivity gains and automation. But be reassured – there is a future for market access professionals, and we will hold our ground! The very serious French sport outlet L’Equipe predicted that by 2050 a team of robots could beat a team of professional players on a football (soccer) pitch.
My prediction – by then, our all-virtual competitor InbeeoGPT will beat us at developing successful market access and pricing strategies.
