Generative AI Is Reshaping Protocol Design
Generative AI tools are increasingly being explored in protocol design, where they can help analyze complex information, identify potential operational issues, and support faster evaluation of study-design options. Protocol design does not identify better drug candidates; rather, it determines how a selected intervention will be evaluated in a clinical study. Sponsors can use predictive models and scenario analysis to explore enrollment assumptions, assess potential dropout risks, and pressure-test protocols before sites open.
These capabilities build on established prospective feasibility, risk-assessment, and protocol-planning processes rather than replacing a previously reactive approach. The stakes could be significant. Protocol amendments can introduce additional costs, operational disruption, and delays, particularly when changes affect recruitment, study procedures, or multiple countries and sites. A Tufts Center for the Study of Drug Development analysis of 836 protocols found that 57% had at least one substantial amendment, while 45% of those amendments were considered avoidable. The study also reported median direct implementation costs of approximately $141,000 for Phase II amendments and $535,000 for Phase III amendments.
Generative AI is one of several analytical approaches that may support protocol planning. Predictive models, simulation, digital twins, and automated feasibility assessments are distinct technologies with different capabilities and applications that can work together to improve healthcare operations. By applying large language models and other analytical approaches to appropriate clinical datasets, organizations may be able to identify operational bottlenecks earlier and adjust study designs accordingly. Digital twins, simulation, and automated feasibility assessments could provide additional insight as programs move from Phase II toward Phase III, although their value depends on the quality of the underlying data and the specific use case.
The broader opportunity is less about replacing established clinical development processes and more about augmenting decision-making with data-driven analysis. Used appropriately, these technologies could help sponsors balance scientific requirements with the practical realities of running complex global studies.
Decentralization Has Moved Beyond Experimentation
Decentralized and hybrid approaches have become increasingly established in clinical research. What gained momentum during the pandemic has evolved into a broader set of practices that can bring selected trial activities closer to participants. Home health services, local health care providers, telehealth, and digital technologies can reduce the need for some site visits and help address geographic barriers to participation.
These approaches may be particularly relevant in rare diseases and specialized oncology studies, where eligible patients may be geographically dispersed or face substantial travel burdens. The FDA notes that decentralized elements may improve participant convenience, reduce caregiver burden, expand access to more diverse populations, and facilitate research in rare and less-mobile patient populations.
The convergence of real-world data with traditional clinical data could also provide a broader view of treatment use and patient outcomes. Electronic health records, registries, claims data, patient-generated data, and device-generated data can be potential sources of real-world data. When appropriately collected, integrated, and analyzed, these sources may complement clinical trial data and support evidence generation across the product lifecycle. Their use also requires appropriate attention to privacy, consent, access controls, security, data provenance, and permitted data use.
The organizations making progress in this area are likely to view data complexity as something to manage deliberately rather than simply as an obstacle. Effective governance, interoperability, validation, and data-quality controls remain essential to realizing the potential of these approaches.
Cloud-Native Platforms Can Help Reduce Data Silos
The shift toward cloud-based data platforms could change how clinical information moves across global research networks. Rather than relying on disconnected systems, modern architectures can support more timely data exchange among sponsors, sites, vendors, and other stakeholders. However, cloud adoption alone does not eliminate data silos. The extent of any improvement depends on interoperability, data standards, system integration, governance, access controls, and data quality.
This can create opportunities for faster safety review, more timely interim analyses, and shorter feedback loops. Improved data exchange may also support the operational preparation and execution of interim analyses, but formal interim analyses remain controlled study activities governed by the protocol, statistical analysis plan, and applicable regulatory requirements.
EHR interoperability may also reduce manual data entry and reconciliation. Capturing information closer to the point of care can reduce opportunities for transcription errors and provide a more current view of study activity. When systems are appropriately validated and integrated, this approach may improve operational visibility and support more consistent documentation.
For sites, reducing repetitive administrative work could create more time for patient-facing activities. Sponsors may gain earlier visibility into study performance, while regulators may receive more structured and auditable information. These outcomes are potential benefits rather than automatic results of adopting cloud-based technology.
Hybrid Models Balance Care with Patient Convenience
Early decentralized approaches highlighted both the potential and the practical limitations of moving trial activities away from traditional sites. Current hybrid models can combine digital consent, remote monitoring, local health care services, and direct-to-patient approaches while retaining in-person visits where clinical procedures require them.
This model may be particularly useful in oncology and rare disease research, where participants may face substantial travel and treatment burdens yet still require hospital-based assessments or procedures. Satellite sites, mobile nursing services, and local providers can provide additional options for selected activities while maintaining access to specialized care. The feasibility of these approaches may depend on the study protocol, jurisdiction, investigational product, provider qualifications, applicable regulatory requirements, and the nature of the activity being performed.
The approach could also provide greater resilience during logistical disruptions. By distributing certain trial activities across sites, homes, and local health care settings, hybrid designs can reduce the impact of individual disruptions without removing the need for careful contingency planning.
The key to success is thoughtful protocol design. Not every visit is appropriate for home delivery, and not every assessment can be collected remotely without affecting data quality. Successful hybrid models should identify which activities require in-person interaction and determine where remote or local alternatives can provide an appropriate balance of participant convenience, clinical oversight, and data integrity.
Recruitment Strategies Designed With Diversity in Mind
Patient recruitment is increasingly moving from broad outreach toward more targeted approaches that can identify potentially eligible participants across different communities and care settings. Electronic health records and other data sources may help organizations identify candidates while expanding outreach beyond traditional research networks, subject to applicable privacy, consent, and data-use requirements.
This approach is relevant to efforts to improve the representativeness of clinical research. The FDA’s current guidance recommends strategies to increase enrollment of representative populations, including consideration of demographic and non-demographic characteristics such as age, sex, race, ethnicity, geographic location, comorbidities, disabilities, and disease prevalence.
Identification alone is unlikely to be sufficient. Building trust can require sustained community engagement, clear communication about trial objectives and participant protections, and meaningful input from patients and advocacy organizations during study design. Incorporating these perspectives earlier may help sponsors identify practical barriers that could otherwise affect participation.
More representative enrollment can also strengthen the relevance of clinical evidence to the populations expected to use a medical product. Differences in response, including potential differences in pharmacokinetics, efficacy, or safety, may occur across population groups, and representative enrollment can provide opportunities to evaluate potential gaps.
Patient-centered protocol design may also help address barriers to participation. Rather than assuming that a single study model will work for everyone, sponsors can consider transportation, visit frequency, caregiver responsibilities, technology access, and other practical factors when designing participation requirements.
The Clinical Research Organization Role Has Expanded
As trial designs become more complex, clinical research organizations may increasingly serve as strategic partners rather than purely transactional service providers. Sponsor-CRO relationships can involve greater collaboration around study planning, operational execution, data management, technology, and risk management.
This collaborative approach can also support more innovative trial designs. Adaptive trials allow prospectively planned modifications based on accumulating data, while master and platform protocols can allow multiple treatments or questions to be evaluated within shared infrastructure.
These approaches require careful planning and statistical oversight. Their potential efficiencies do not remove the need to protect trial integrity, maintain reliable data, and meet applicable regulatory requirements.
As trial complexity increases, CRO expertise in local regulations, site management, patient recruitment, logistics, and study execution may become increasingly valuable. Sponsors can contribute scientific and strategic direction, while specialized partners can provide operational capabilities that complement those internal resources.
Scaling Success While Addressing Equity Gaps
The clinical research landscape is moving toward more integrated approaches that combine digital technologies, patient-centered study design, and increasingly connected data sources. These approaches may create opportunities to improve operational visibility, participant access, and the efficiency of evidence generation.
At the same time, technological progress can introduce new challenges. Scaling digital approaches into underserved therapeutic areas may require investment in infrastructure, connectivity, training, and technology support. Without those considerations, digital approaches could create barriers for participants who have limited access to devices, reliable internet connections, or digital services.
The next phase of clinical research might therefore depend not only on deploying new technologies but also on determining where those technologies add meaningful value. Effective implementation may require a balance between automation and human oversight, centralized data infrastructure and local expertise, and digital convenience and access to specialized care.
For sponsors and research partners, there’s an opportunity to build clinical research models that are more connected, adaptable, and responsive to participant needs while maintaining scientific rigor and data quality. The path will vary by therapeutic area and study design, but the direction is toward more deliberate use of technology, data, and patient input throughout the research lifecycle.
Conclusion
Clinical research is entering a period in which technology, data, and patient-centered design are becoming increasingly interconnected. Generative AI, decentralized and hybrid trial models, cloud-native platforms, real-world data, and more targeted recruitment strategies can create new opportunities to improve how studies are designed, executed, and adapted. At the same time, these technologies are not solutions in isolation. Their value depends on the quality of the data, the strength of the underlying processes, and the ability of research organizations to integrate innovation without compromising scientific rigor, participant safety, or data integrity.
It’s an opportunity to apply the right tools to the right clinical-development challenges. Generative AI and other analytical approaches can augment established planning and decision-making processes, while decentralized models, connected data platforms, and targeted recruitment strategies can support more participant-centered study designs. Their implementation requires appropriate governance, regulatory oversight, interoperability, data protection, and human judgment.
For sponsors, CROs, sites, and other stakeholders, the focus can therefore remain on building clinical research systems that are connected and adaptable without losing sight of scientific rigor, participant needs, or data quality. The most appropriate balance between digital and in-person activities, automation and human oversight, and centralized infrastructure and local expertise will vary by therapeutic area and study design.
