Novel Uses of Scoring Functions in Clinical Trial Design Selection
For decades, statisticians have cultivated methods to optimize and de-risk clinical trials for strong regulatory submissions. As the number of possible clinical trial designs have exponentially increased, the need to construct a new set of quantitative tools for accelerated comparison of these designs has become critical for trial selection.
Here I take a closer look at how these three work in tandem, with a special highlight of the use of scoring functions.
A scoring function is a mathematical tool that enables trial sponsors to review millions of designs quickly, and rank them based on a number of performance characteristics. While such functions are a straightforward and intuitive tool, the way in which scoring functions can be used for clinical development have novel new implications.
There are certain forms of de-risking that apply to almost all clinical trials. All trial sponsors need to avoid situations where clinical trials are underpowered. Nearly all sponsors attempt to optimize patient allocation.
Some sponsors though, might have an additional set of worries – highly rigid budget constraints, too few patients to enroll, etc. The majority of statistical consultants will not have the ability to optimize across all of these factors without some guidance from the sponsor.
The scoring function is a customized function, built by statistical consultants with input from sponsors, to create an objective way to measure the performance characteristics of a trial, given sponsor preferences.
Building scoring functions enables sponsors to rank order their clinical trial preferences, when they have millions of options from which to choose. For example, after SolaraTM designs a full and expansive design space, a scoring function can quickly determine which of these millions of designs satisfied a sponsor’s needs for balancing power and speed.
This process also requires research and development teams to go through the process of articulating collective preferences earlier. Oftentimes, different members of the sponsor leadership teams have different goals, or the trade-offs between goals are not clear enough to warrant an elaborate discussion.
A customized scoring function enables sponsor teams to agree on more nuanced tradeoffs, like how much more power justifies delaying interim analyses, or which of myriad secondary endpoints might justify extending a clinical trial.
There are in fact two quantitative pre-requisites to facilitating this process. The first is to build a scoring function, a process that requires deliberation and technical precision. Additionally, the deliberation requires access to knowledge about tradeoffs between performance characteristics. These tradeoffs in turn are illuminated after simulating millions of designs, and seeing patterns in performance characteristics.
A new Cytel position paper outlines a way in which to accomplish this. After designing the full design space, a Pareto optimization method is implemented to determine the tradeoffs of performance characteristics amongst all of the optimized designs. A scoring function is then used to support sponsors in selecting their trial design based on information of these various tradeoffs, and sponsor insights.
About the Author of Blog:
Dr. Esha Senchaudhuri is a research and communications specialist, committed to helping scholars and scientists translate their research findings to public and private sector executives. At Cytel Esha leads content strategy and content production across the company's five business units. She received a doctorate from the London School of Economics in philosophy, and is a former early-career policy fellow of the American Academy of Arts and Sciences. She has taught medical ethics at the Harvard School of Public Health (TH Chan School), and sits on the Steering Committee of the Society for Women in Philosophy's Eastern Division, which is responsible for awarding the Distinguished Woman in Philosophy Award.