Vincent Buchheit, Sebastien Jolivet, Nicolas Frey
Hoffmann La Roche
Objective: For 12 years now, Roche has a group of Clinical Pharmacology Data Scientist (CPDS) embedded within the Clinical Pharmacometric (PMx) group. The added values of the CPDS to the conduction of PMx activities have been highlighted at previous PAGE meetings in 2013 [1] and 2016 [2]. The objective of this abstract is to describe the expansion of the CPDS role to all the different types of Quantitative clinical Pharmacology activities over the last few years and highlight how it further improved efficiency and quality of the quantitative clinical pharmacology analyses.
Methods: The Clinical Pharmacology (CP) group at Roche pRED consists of clinical pharmacologists, pharmacometricians, disease modelers and CPDS. All those roles have a similar objective which to help clinical project teams take the right decisions to transform molecules into medicines for patients. In order to do so, different type of quantitative analyses are conducted ranging from graphical analysis, non-compartmental analysis (NCA), population PKPD analysis to disease modeling. The CPDS’s main accountability is to enable those analyses and ensure full data traceability and reproducibility.
Results: Over the last few years, we have established a group of 7 CPDS, which supports activities at study, project and disease levels. The diversity of the CPDS’s tasks is constantly increasing. The main task remains the creation of data set ready for analysis (using data from different sources, different formats) that are evaluated using whenever applicable previously developed models to identify and fix data inconsistencies (PAGE 2013 [1]). Then it evolves with the data exploration by producing fit-for-purpose graphics to currently perform simulations, conduct NCA, produce outputs for reports, deliver interactive tools for simulation or data exploration, prepare submission-ready data files. In addition [JS{1] to their increasing contribution to the conduction of quantitative clinical pharmacology analyses, the CPDS are also driving the following two recent initiatives to improve efficiency and quality of those analyses:
- Improve Information Technology (IT) environment: in collaboration with IT colleagues, the team delivered a fully validated repository, to ensure complete traceability and reproducibility of all PMx activities. This new platform allows scientists to perform their daily work, in a secured validated and versioning control environment. It combines a global, secure file repository with a robust and versatile modeling management interface for tools like NONMEM®, SAS®, R, PsN.
- Enable data access before database lock: in an attempt to gain critical time especially during submission periods, a new process has been put in place to streamline fillings activities and enable pharmacometricians to start the development of PMx models prior to database lock to gain few weeks to few months.
With usually a Master degree in computational science and engineering, a CPDS brings flexibility and efficiency in the analysis process, increase the quality of the deliverables and also ensure full traceability and reproducibility. Based on our experience, a CPDS free up time of the quantitative clinical pharmacology scientists allowing for more scientific activities to be conducted. The most significant improvement is in the reduction of the time between clinical database lock and dataset ready for analysis which is usually around 70%.
Conclusions: Having data scientists fully embedded in a clinical pharmacology department allow to maximize efficiency and quality of the quantitative clinical pharmacology analyses.
References:
[1] Buchheit, V. and Frey, N., Data quality impacts on modeling results, PAGE 22 (2013) Abstr 2749 [http://www.page-meeting.org/default.asp?abstract=2749]
[2] Buchheit, V. and Frey, N., Added value of the Data Scientist role in a Clinical Pharmacometric group, PAGE 25 (2016) Abstr 5884 [https://www.page-meeting.org/default.asp?abstract=5884]
Reference: PAGE 28 (2019) Abstr 9132 [www.page-meeting.org/?abstract=9132]
Poster: Methodology - Other topics