STRATOS accelerated guidance for real world data analysis

Background (from the Biometric bulletin 2023, Issue 4)
Notwithstanding some recent improvements, major concerns remain about the methodological quality and standards of empirical research. Especially in the analysis of observational data, problems arise from poor application and reporting of statistical methodology. In the recent COVID crisis, van Calster et al (2021) discussed research waste resulting from the current organization of scientific research, which tends to prioritize the ‘need to publish’ over publication quality. Their proposals for change refer to initiatives aiming to improve the methodology and reproducibility of research. Among these, our STRengthening Analytical Thinking for Observational Studies (STRATOS, www.stratos-initiative.org) initiative is the only group with a primary focus on developing and disseminating evidence-supported guidance concerning statistical analyses. Since its inception in 2013, the STRATOS initiative has grown to more than 100 members from 20 countries. Its activities are organized into 9 topic groups (TG) and 12 cross-cutting panels. Activities of all TGs and some panels were presented in articles in the Biometric Bulletin (BB) over the last 5+ years. In BB March 2022 we published a brief update on achievements of the STRATOS

At the Lorentz Center (September 16-20 2024)
To advance the cross topic STRATOS developments and help set the course of action for the next 3 years, we organize an all-STRATOS scientific workshop at the Lorentz Center (16-20 September 2024). In this workshop experienced experts meet and work together alongside promising young researchers which build a bridge to future generations.   Some 55 scientists will gather live and start each day from an expert-led plenary session for which they are joined by an online community. To then work in parallel groups sometimes homogeneously composed, and for a large part purposefully mingling the topic groups, to bring diverse insights and skills to the table. In late afternoon, we come all together again to summarize, debate and conclude. This creative exchange is set to yield workable methods and solutions that solve real and complex problems in an effective way. Our 4 key cross-topic objectives for the meeting center around:

1. Combining Machine Learning with Statistical Methods
2. Estimands in time-to-event analysis
3. Simulation studies: how to generate synthetic datasets?
4. Developing an open-source case study dataset

The biometric bulleting articles in addition refer to future foci on `Open science’, `Initial Data Analysis’ and `Guidance for researchers with limited statistical knowledge and experience’. For more information on exciting activities in topic groups and panels, please see the STRATOS website: https://stratos-initiative.org/