新准则：AI如何重塑试验设计、执行与信息披露

The New Standard: How AI Is Reshaping Trial Design, The New S tandard: H ow A I Is ReshapingTrial Design, E xecution, and Disclosure Executive summary AI is no longer experimental in clinical trials; it’s already transforming the way work is performed.But the change hasn’t come from moonshot technologies or radical automation. Instead, AI has This strategic playbook explores how artificial intelligence is reshaping clinical workflows — not justin trial design or regulatory compliance, but across the full trial lifecycle. It offers a blueprint for howleading teams are adopting AI in practical, measurable ways, from protocol feasibility and cohort Featuring real-world examples from Citeline’sSmartSolutionssuite andTrialScope Disclose, thisplaybook highlights what the new AI-enabled standard looks like and how clinical teams can start The New S tandard: H ow A I Is ReshapingTrial Design, E xecution, and Disclosure A quiet revolution For years, AI has been positioned as a promise — a future breakthrough that might eventuallychange clinical research. But the revolution is already under way, and it’s far more practical than AI is not replacing clinicians, scientists, or regulatory teams. It’s augmenting them: automatingwhat’s manual, streamlining what’s complex, and surfacing insights where there used to be noise. In today’s environment, trial teams face growing pressure to: According to ClinicalTrials.gov, the median length of completed clinical trials by phase in 2024was as follows: This adds up to nearly nine years for the active clinical trial process. Adding preclinicalactivities and the time it takes to secure approval, sponsors are looking at over a decade The New S tandard: H ow A I Is ReshapingTrial Design, E xecution, and Disclosure •Design more inclusive and realistic protocols As protocols become more complex, the average number of protocol amendments hassoared. According to results of a 2022 study by the Tufts Center for the Study of DrugDevelopment (Tufts CSDD), 76% of protocols in Phases I–IV require at least one amendment, •Improve enrollment performance and trial diversityPatient enrollment remains a thorny problem for clinical trials, with an estimated 37% of sites under-enrolling patients and 11% of sites failing to enroll a single participant.3And oncepatients are enrolled, it can be a challenge to keep them, with patient dropout rates ranging •Deliver regulatory reporting with greater speed and transparencyAccording to a 2024 survey of biopharma executives, complicated regulatory requirements are one of the top challenges increasing clinical trial complexity and one of the factors driving The New S tandard: H ow A I Is ReshapingTrial Design, E xecution, and Disclosure On a daily basis, AI in tools and workflows is already helping meet those demands across protocolplanning, patient feasibility, site selection, and compliance. For example, AI can be used during the clinical trial design process to simulate data that canpredict participant outcomes. One report notes that machine learning (ML) prediction models In a 2019 study, the Cincinnati Children’s Hospital Medical Center found that employing itsAutomated Clinical Trial Eligibility Screener (ACTES), a real-time natural language processing(NLP) and ML-based system, cut patient screening time by 34%. The time saved was applied bythe clinical research coordinators (CRCs) to activities like study-related administrative tasks and AI can be integrated in the site selection process to predict site enrollment rates and forecast siteperformance, as well as plan for any contingencies that could arise during the trial.8And it can From these examples, it’s clear the question is no longer “Will AI change clinical trials?” but “Areyour teams already working differently because of it?” The New S tandard: H ow A I Is ReshapingTrial Design, E xecution, and Disclosure Where AI delivers real operational value Today’s most effective clinical teams aren’t chasing innovation for its own sake. They’re usingAI to solve real bottlenecks that have held trials back for years. Here are areas where AI is already delivering impact: SITE AND INVESTIGATOR SELECTION PROTOCOL FEASIBILITY The challenge: The challenge: When study teams create protocols,assumptions are often made about thepatient population that don’t always fit withactual patient demographics, particularly It can take weeks to build a list of potentialsites for a clinical trial — and that list oftendoesn’t take into account any potential AI’s role: •Match trial needs to investigatorswith relevant experience, based on AI’s role: •Automatically generate inclusion/exclusion (I/E) criteria based on pasttrials and validate against real-world REGULATORY DISCLOSURE The challenge: Disclosure teams often must manuallyextract and structure data fromunstandardized trial documents — a DECISION SUPPORT The challenge: Cross-functional teams can spend hourssearching siloed system