Translated by Reuters using machine learning and generative AI, please refer to the following disclaimer: https://bit.ly/rtrsauto

The Impact of AI on Research Organizations

The stocks of contract research organizations (CROs) have plummeted due to fears that advancements in artificial intelligence may allow drug manufacturers to internalize clinical trials. However, industry experts argue that this drop overestimates the extent to which technology can replace essential capabilities of the sector.

IQVIA, Medpace, and Charles River Laboratories have seen significant declines since Anthropic launched advanced AI agents in February, fueling expectations that drug manufacturers could rely less on CROs.

A recent wave of partnerships between pharmaceutical companies and AI firms has further reinforced these concerns.

“Could AI eat into CROs? Yes, I think that’s a possibility,” stated Thomas Laur, CEO of data analytics company DNAnexus.

However, the nature of services provided by CROs, from patient recruitment to global trial execution, will be challenging to automate or replace, according to several analysts, industry experts, and policymakers interviewed by Reuters.

CROs maintain global networks of trial sites and hold exclusive data that pharmaceutical companies, especially small biotech firms, cannot easily replicate, as noted by Jailendra Singh, an analyst at Truist Securities.

Wall Street shares this view. TD Cowen analysts estimate that even a fully AI-based clinical trial setup would only allow drug manufacturers to achieve 10-15% savings.

The pharmaceutical industry’s reliance on large-scale execution is at the core of the argument.

Just finding a small group of eligible patients for an early trial across diverse demographic and geographic areas requires extensive data standards and site networks built by CROs over decades. “Pharmaceutical companies don’t have the same level of data and expertise,” added Mr. Singh.

The Human Element

CRO leaders emphasize that while AI can streamline certain parts of the process, it cannot replace the human backbone of trials.

“AI alone cannot contact the physician, enroll the patient, ensure they show up on time for appointments, record all data,” said Brigham Hyde, CEO of Atropos Health.

Although AI can automate high-volume tasks like patient pre-screening, critical decisions still require human oversight, according to Ami Bhatt, head of the FDA’s digital health advisory committee.

Site execution, informed consent, and safety monitoring firmly remain in human hands, with ultimate responsibility resting on individuals, she added.

Others highlight more fundamental constraints of the technology. AI cannot replace essential drug safety laboratory tests, and its use in direct patient care remains limited by regulatory scrutiny and liability risks, noted William Pierce, former HHS assistant secretary for public affairs.

An Opportunity to Seize

Analysts believe that rather than replacing CROs, AI could enhance their value by accelerating trials and improving efficiency.

TD Cowen analysts estimate that an entirely AI-managed late-stage trial could be completed in 47 months, down from the initial 58 months, representing an 11-month reduction.

This time compression could become a powerful competitive advantage for CROs heavily investing in AI.

For example, launching a drug one year earlier with estimated annual revenues of $1.5 billion could generate approximately $44 million in additional revenue, according to TD Cowen.

“We expect the emergence of new types of contracts, including gain-sharing deals on AI efficiency,” said the brokerage firm.

However, the recent stock decline suggests lingering investor anxiety.

Jim Lee, from drug manufacturer Incyte, noted that investors are likely concerned about services that CROs could potentially reduce, impacting their revenues. Currently, analysts maintain that there is no evidence that pharmaceutical companies are cutting expenses with CROs due to AI and describe the sell-off as “more panic than a real threat”.

For Mr. Singh, the conclusion is simple: “We don’t see AI as a headwind for the industry; it’s more of a tailwind.”