George Achilleos is the CEO of NetraMark Holdings Inc.
Drug development has a strange economic problem.
A pharmaceutical company may be willing to spend tens or even hundreds of millions of dollars advancing a drug into a major clinical trial. However, spending a small fraction of that amount on technology designed to better understand the risk before making that decision can still be a difficult sell.
The challenge is that the cost of prevention is visible today, while the value of preventing a mistake is invisible unless something goes wrong. This is one reason why new technologies designed to reduce clinical development risk can face slower adoption than their economics might suggest.
Drug development is highly regulated, and executives have good reasons to be cautious about changing how major decisions are made. As technology improves, though, the way companies think about that risk will change.
Being first can feel riskier than being wrong conventionally.
Every executive understands the career risk of making an unconventional decision. Imagine two companies considering whether to invest another $100 million into the next stage of a drug program. One follows the traditional process, and the trial fails. The other uses a new analytical technology, changes its strategy based partly on those findings and the trial still fails. Those outcomes may be judged very differently.
That helps explain why new technologies can be difficult to adopt even when they promise to reduce risk. The technology may reduce business risk while initially increasing perceived personal or organizational risk for the executive adopting it.
That changes once competitors begin using the technology successfully. In almost every industry, a point comes when being first feels risky and then, very quickly, being late feels riskier. Pharma is unlikely to be different.
Investors will care less about AI and more about better decisions.
Investors need management teams to make better decisions with shareholder capital, and not just because AI is fashionable.
Consider what happens when a large late-stage clinical trial fails. Investors understandably ask what went wrong, but as analytical technologies become more capable, another question may become increasingly important: What did management do with the data it already had before committing the next major round of capital?
• Did the earlier trial suggest that the drug worked consistently across patients?
• Were there signs that certain types of patients were unlikely to benefit?
• Were there patterns in the data that deserved more investigation before the company committed substantially more money?
One of the most consequential decision points in drug development comes after Phase 2, when a company must decide whether the evidence is strong enough to justify moving into a much larger Phase 3 program. The FDA treats the end of Phase 2 as a critical planning point, intended to occur before sponsors make major commitments of effort and resources to Phase 3.
As analytical capabilities improve, investors may increasingly expect companies to demonstrate that they thoroughly interrogated that data before making one of the most consequential bets in a drug’s development.
The real value is changing the next trial.
Once Phase 2 is complete, a company has valuable clinical data showing how patients responded to the drug and control arm. However, is that data being used to determine whether the trial met its endpoint or whether it can reveal additional information that should influence the design of the next study?
For example:
• Was the treatment effect broadly distributed, or were types of patients driving the result?
• Were there identifiable patients who consistently responded to the control arm and made treatment separation more difficult?
• Are there characteristics that could help define a more appropriate population for the next trial?
It’s important to find those answers before Phase 3 begins and use them to inform prospective trial design, including through explainable AI-driven trial enrichment. This moves analytics beyond explaining what happened and toward generating actionable insights that can shape the next major trial before more capital is committed.
The winning technologies will make existing teams smarter.
Successful technologies will help experienced scientists, statisticians and clinical executives extract more information from the data they already have. That means technology needs to be easily understood. Executives need to know why a system reached a conclusion, scientists need to be able to challenge it and management needs to understand how much confidence to place in the result before changing a major development decision.
The FDA has increasingly focused on whether AI is credible for a particular use and whether there’s enough evidence to support relying on its output, not simply whether a company can say it uses AI.
That’s likely to be the winning model for the industry as well, and AI’s value will be determined by whether it helps people make better decisions.
The economics of prevention are changing.
The hardest thing to sell in business is the solution that ensures a problem never happens. If a company spends money on an additional analysis, the expense is obvious. If that analysis prevents the company from making a bad $100 million decision, there isn’t a failed trial to point to afterward. The value exists partly in an event that never occurred.
If companies can demonstrate that better analysis of existing clinical data changed a major decision, improved the design of the next trial or prevented capital from being deployed against a weak assumption, the technology stops looking like another expense and becomes part of responsible capital allocation.
At that point, the question from boards and investors may change. Instead of asking, “Why spend more analyzing a trial that’s already complete?” they may begin asking, “Why commit the next $100 million before we’ve learned everything we can from the last trial?”
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
