Menu Back to Poster-Presentations-Details

P112: Data-driven Clinical Trial Execution from Start-Up to Closure

Poster Presenter

Sanghita Bhattacharya

Associate Director Data Science , Feasibility and Analytics, Clinical Operations
Johnson & Johnson Innovative Medicine
United States

Objectives

Data-driven methodologies like AI/ML methodologies as well as usage of real-world datasets are crucial during the trial feasibility enabled trial acceleration strategies.

Method

AI/ML algorithms as well as real-world datasets were used to predict site performance as well as diversity potential. Data-driven tiering algorithm; as well as regression analysis to predict site selection gap enabled realistic targets through a suite of reporting dashboards across the program.

Results

With a suite of data-driven end to end suite of analytical tools the “journey of a site” was captured with the ability to intervene and mitigate trial risks proactively throughout their lifecycle during the clinical trial from start-up to enrollment. Insights generated at snapshots on an ongoing basis led to the identification of any bottlenecks during the feasibility phase. Comparison of real-world data sets as well as site surveys enabled realistic diversity targets at the indication level, as well as opportunities for site engagement strategies for top recruiting diverse sites. Patient finding methodologies with geo-proximity algorithms led to the identification of regional "patient hotspots" at zip codes leading to opportunities for sponsor naïve sites and expansion of site footprint. Market trends, as well as competitive insights, led to the creation of real-time competitive risk analysis for overlapping, often over-burdened high-priority sites to ensure enrollment targets were met as well as tailored site engagement strategies. Predictive models were also utilized during the site feasibility process to reduce the gap to reach site targets with alternative backup plans used globally. Additionally, the usage of customized CRM tools enabled real-time site-sponsor interactions to understand and solve key roadblocks to site recruitment plans. These methodologies enabled centralized and real-time information and actionable insights that were shared across cross-functional teams leading to prompt decision-making and corrective actions.

Conclusion

With the emergence of new covid vaccines now more than ever the clinical trial space is in the public eye and the need to recruit faster and better is stronger. Complex study protocols with staff constraints at sites, and multiple sponsors vying for similar sites across the globe is a norm. Site feasibility has become an extremely competitive landscape with declining recruitment rates across multiple TAs, especially in large megatrials. From research naïve sites in untapped geographies to identifying under-represented populations, these newer methods will enable us to look beyond past relationships and historical performances; with the ability to intervene and mitigate trial risks proactively throughout their lifecycle. Validation of these predictive algorithms will be a necessity as real-time site performance emerges, to build trust among key stakeholders. The need of the future would be to proactively acquire novel datasets, with a fit-for-purpose end-to-end ecosystem across multiple stakeholders who would work collaboratively from site feasibility to site performance with built-in trial engagement tools that could enable real-time feedback and site support.