How human-enabled AI is creating a new map for navigating site selection
The clinical research industry’s recent abundance of data is transforming decision-making strategies at key stages in the development process. With the right approach, integrating artificial intelligence (AI) can effectively manage diverse datasets and generate insights to unlock the potential of data-driven decisions. Our implementation of AI in the site selection process has had a transformative effect, from reshaping understanding of the clinical research ecosystem, redefining ‘core clinical fit’, and informing the approach we use to rank and select sites.
In this blog, we explore how AI facilitates new ways of navigating data balanced by human-enabled intelligence to extract meaningful insights for informed site selection decisions.
Exploring the clinical research ecosystem
Employing AI in site selection has been transformative. We built our One Search site selection system to ingest vast data from multiple sources, generating nuanced insights into site suitability. Unlike traditional methods, One Search rapidly ranks sites based on a wide range of data, uncovering critical connections within the millions of datapoints that characterise the clinical research ecosystem. These interconnected relationships between investigators, sites, institutions, and studies reveal that the best-fit investigators have multiple high-quality connections. This AI-enabled exploration resulted in redefining clinical fit as a function of enhanced ecosystem connectedness, significantly improving patient enrolment and study start-up timelines.
Mapping connections in the ecosystem
AI’s processing power maps the vast, overlapping relationships within the clinical research ecosystem, providing a deeper understanding of site performance drivers. Studies drive higher-level connections between sponsors and institutions, with connections deepening as sponsors run multiple studies, and institutions participate in various projects. Data shows that the best investigators have multiple high-quality connections within this web, representing baseline clinical fit. By tailoring One Search’s parameters, our feasibility experts prioritise the most valuable connections within therapeutic areas and indications to better select suitable sites.
Tiered analysis and site feasibility
Once mapped, we combine human-enabled intelligence and machine learning to rank and analyse the connections within the network. Connectedness is heavily weighted, alongside other factors like previous performance, enrolment speed, quality, diversity data, and site capabilities. Our specialists review and prioritise protocol nuances for AI to incorporate into the ranking. This layered analysis provides a comprehensive picture of a site’s potential, fast-tracking top-tier sites for selection. One Search sorts sites into four tiers. Tier 1 and 2 sites have sufficient data to demonstrate that that are most likely to enrol the patients required for the study. Tier 3 and 4 sites lack sufficient data or have data that reflects they are less likely to enrol the required patients.
Human-enabled intelligence for complex analysis
AI must be balanced with human expertise to achieve a holistic understanding of site potential. Tier 3 sites, despite having lower data scores, may still have significant potential that simply hasn’t yet been demonstrated. Some less experienced sites might also face challenges as they navigate the learning curve at the beginning of their clinical research journey – but with the right guidance, they can succeed.
Our expert analysts review the site profiles comprehensively to determine what a site may need to realise its potential and if it could feasibly perform to the required standards with additional support. This balanced approach allows us to allocate trial sites effectively, ensuring diverse and capable participation.
Connecting to diverse populations in clinical trials
The ecosystem approach also enables novel strategies for diversity in trials, which is increasingly important for sponsors and regulators. Proactive inclusion of representative, diverse communities is an important part of planning for successful studies, especially those with registrational intent in the US. The FDA’s 2022 guidance supports the further diversification of clinical trial populations to be more representative of the research’s target populations.
One Search helps us to assess population data and demographic breakdowns to understand site-investigator-community relationships. One Search also ingests data from our partners, and claims, census, and third-party data to identify sites and investigators more likely to enrol underrepresented populations. This approach enhances clinical trial access and diversity leading to more meaningful therapies for broader patient demographics.
AI’s predictive potential for ideal site selection
One Search combines AI’s computational power with the nuanced expertise of human analysts, to map, organise, categorise and analyse clinical research connections quickly and with greater accuracy. AI’s assessments, layered with additional data from external and proprietary sources, helps us to generate more robust site profiles for deeper assessment. This holistic approach enables decision-makers to identify high-performing sites with greater precision, for optimal patient enrolment and study start-up timelines.
By leveraging AI algorithms augmented by human insights, One Search can uncover data trends and effectively forecast which sites may be more likely to excel. Part of the analysis uses data clustering techniques, and when new or unfamiliar sites are clustered with sites that have demonstrated high performance, we can extrapolate the metrics across the cluster and forecast performance. As datasets grow and we continue to refine AI applications, this predictive capability will mature, extending across tiers, specific therapeutic areas, and indications to provide earlier clarity on ideal site selection.
Unlocking innovation and efficiency
Embracing human-enabled AI allows us to combine AI’s technical capabilities with our medical and operational expertise, driving progress, ensuring quality data and bringing important therapies to patients faster. By integrating AI-driven insights into site selection processes, One Search helps sponsors navigate the clinical research ecosystem with agility and precision, paving the way for continued advancements in patient care.
Connect with us to learn how ICON’s technologies and agile approaches can optimise your clinical program, starting with data-driven site selection.