Developing investigational compounds is a team sport. Within that team our expectation for statistical experts is to be strategic partners to clinicians, program managers and other members who shape, design and implement development strategies. The motto of our group, Quantitative Sciences, is to contribute to “make the correct decision, at the earliest time point, in the most efficient manner possible”, across our portfolio of R&D opportunities. Apart from technical expertise, this requires our statisticians to clearly articulate, successfully influence, and ultimately achieve the desired impact in shaping the thinking on HOW the team will make the decisions. Being effective requires deep understanding of adjacent functions, including preclinical, translational medicine, pharmacology, clinical, regulatory, and commercial. First and foremost our statisticians need to become experts in drug development, equipped with fundamental knowledge in statistics and experimental design and analysis methods.

We strive to move away from a merely transactional approach – phase 1/2/3 - to a culture of “critical questions driven drug development”. To make the correct decision, we need to first identify and align on the key decision problems. Then develop the appropriate research questions. Only once this is done, start work on the experimental design and analysis methods. The greatest impact is achieved by looking at the overall program, rather than solely at an individual clinical trial in isolation, and to some degree even condition our thinking by portfolio level information (e.g. how do we manage uncertainty, given the priority setting of a program relative to other efforts across the portfolio).

Our statistical experts enable teams to connect the dots: - By integrating externally available information with internally created knowledge – model based meta-analysis; - By building seamless linkage through longitudinal modeling from early biomarker observations all the way to clinical endpoints in confirmatory trials; - By fully integrating translational modelling into the development effort; - By proposing experimental designs, where decisions are based on early biomarker readouts, but subjects within the trial continue to be observed to also create data within the same subject on both early and late clinical readouts; - By ordering the critical research questions and answer them sequentially, making full use of innovative designs, including adaptive designs and opportunities for model informed drug discovery and development; - By routinely conducting scenario analyses of options and simulations to evaluate the operating characteristics of the design and its execution.

Our ambitions: - To apply statistical innovations broadly and consistently, across the portfolio; - To convince all our Therapeutic Areas to better manage the risks and uncertainties of development by applying estimation methods rather than hypothesis testing in “learn” trials; - To improve the information value of Proof-of-Concept and dose-response finding designs by consistently applying model based designs; - Wherever appropriate to implement interim analyses to enable earlier decision making. - Where appropriate, to also apply innovative designs in confirmatory settings, and engage regulatory experts to achieve buy-in.

This requires a continuous handshake between design and execution. We are building an execution framework that aims to enable “real time learning and decision making”, by integrating the operational aspects of in-stream data capture, cleaning and analysis, with a goal to execute interim analyses and adaptive designs in a scalable and efficient manner.

Ultimately our statistical experts invite all other members of the team to engage in a scientific dialogue that is driven by quantitative principles, allowing for the “correct decision to be made at the earliest time point in the most efficient manner.”