From Algorithms to Outcomes: Stephen’s Journey in Clinical Data Science

You've been with ICON/PRA for 18 years - that's quite a remarkable tenure. Can you walk us through your professional journey from when you first joined the company to your current role as Director of Clinical Data Science? 

I started my career at an entry level as a Clinical Data Coordinator and worked my way through first leading studies in various levels of that role to moving into what is now referred to as the Clinical Data Science Lead position. A highlight included acting as the Project Manager for what became a data management stand-alone study as the study progressed through closeout. It was truly inspirational as this was one of the first studies I started working on when I joined so many years ago and the sponsor supported my career advancement throughout the many years of the registrational oncology study. This same sponsor requested me as the data management representative as they initiated a new biosimilars branch within their organization. I then moved into management and director positions, where I have enjoyed supporting and mentoring new generations of talent and leaders within ICON.

How has the role of data science transformed clinical research over the past two decades, and what emerging technologies are you most excited about? 

Clinical Data Science has been a pivotal mind shift in thinking critically about the data that matters most. We are no longer simply cleaning for data accuracy, but now thinking about what that data means to patient safety and endpoint analysis as the volume and voracity of data and data modalities enable technologies to provide near real time aggregated data insights. The Clinical Data Science team need to not only understand these aspects of clinical trial data but also summarize and present them to their stakeholders in a format easily consumed by their clinical trial functional leads and sponsors. The technology advancements in data connectivity like ICON’s CDR and various analytics tools like Medidata’s Clinical Data Studio are exciting enablers of the evolution of the Clinical Data Science methodology. Artificial Intelligence further enhances the CDS transformation by refining review to once again looking at the data that matters most and automating review where possible while allowing a human-in-the-loop quality gate to affirm the automation outputs. This would not be possible without our legacy experience as Clinical Data Managers.

For aspiring clinical data scientists, what key skills and competencies would you recommend developing to be successful in this rapidly evolving field? 

The great news here, is there is an “easy button,” where the Society for Clinical Data Management has already mapped out a CDM competency framework and supporting Clinical Data Science thought leadership white papers on our industries’ position on this topic. In my opinion, paramount to being successful in CDS is your ability to think critically about the data that matters most for your trial and deploying a risk-based approach to deployment of a clinical trial while welcoming the perspectives of your colleagues and keeping open communication throughout.

How does ICON's approach to clinical data science differentiate itself from other organizations in the pharmaceutical and healthcare research space? 

ICON differentiates itself as an organization by sharing the activities of Clinical Data Science across data management specialist roles (Clinical Systems Lead, Clinical Data Science Lead and Clinical Data Risk Analyst) with an overarching CDS Program Manager as a single point of contact for the sponsor on a portfolio of work to facilitate consistency and continuity of knowledge on expanding and evolving portfolios. This is further enhanced through an overarching aligned CDS director providing leadership support to both the ICON CDS team members and the sponsor’s Data Manager counterpart.

How are artificial intelligence and machine learning currently being integrated into clinical research processes at ICON? 

ICON has many examples of deploying AI and ML solutions within clinical research. Good examples within our Clinical Data Science department include our audit trail review dashboards, auto-coding and AE/CM/MH query automation. We also have cross-organizational technologies like Forward + which help facilitate global resourcing and forecasting. Lastly, we utilize data analytics platforms like Medidata Clinical Data Studio to bring near-real time analytics to our study teams to make timely and informed risk-based decisions through the duration of our study and demonstrate to our regulators both CRO and sponsor informed oversight of our clinical trial through the full trial lifecycle.

Looking ahead, what are the most significant trends or potential breakthroughs you anticipate in clinical data science over the next five to ten years? 

This is a timely question as I am meeting with the SCDM executive board in early 2025 to discuss this exact topic as we look to our vision over the next five years within our industry. I anticipate data convergence will accelerate as data modality becomes unified under a central data repository and query automation through to analysis moves further to human-in-the-loop and beyond. To prepare ourselves, a focus on critical and ethical thinking will be needed to bring in the new normal of an AI-driven trial and society. We are not yet to the stage of thinking about our first interplanetary clinical trial but that will certainly be a new frontier of discovery in the decades to come as well!