Projects

Eleven positions will be offered in the following projects. One doctoral candidate (DC) will be recruited per project.

Role of Clinical Trial Data Sharing for validation of prognostic and predictive models

Ludwig-Maximilian’s University, Munich, Germany
Prof. Dr. Ulrich Mansmann

More information
Objectives: Use of data from large Phase III or Phase IV trials (available at CSDR, VIVLI, YODA). The DC will explore methodological aspects how to validate prognostic/predictive models over different CTDS platforms and will work on solutions for missing methodology. It is of interest to learn from this activity more general aspects on validation challenges which could be communicated to different fields of clinical medicine and play an important role on best practice implementations.
Expected Results: Protocol for the project, SAP, validation reports for several models over several platforms, two research papers.
External secondments: (1) At BAYER to learn about the process of CTDS in a pharmaceutical company; (2) At ECRIN to learn about the process of CTDS in an academic setting. (3) At YODA to learn about the process of CTDS from a respective sharing platform.
Enrolment in Doctoral degree(s): LMU, Medical Faculty, PhD programme Epidemiology and Public Health.
Requirements: (1) Master in Biostatistics or Data Science; (2) Knowledge of basic concepts in prediction and validation; (3) Basic understanding of the role of prediction models in medicine.

Impact of clinical trial data sharing

University of Rennes, Rennes, France
Prof. Dr. Florian Naudet

More information
Objectives: To assess the impact of clinical trial data sharing (in terms of re-use, published re-uses, impact of those re-uses and costs saved), we will triangulate evidence from different studies including: i) a large-scale observational study of current data-sharing policies in leading biomedical journals, ii) survey of data sharing practices in the biomedical literature and iii) methodological/simulation studies exploring potential pitfalls associated with data re-use.
Expected Results: Evidence about the impact of clinical trial data sharing (in terms of re-use, published re-uses, impact of those re-uses and costs saved): Study protocols, SAPs, reports. Possible improvements of existing methods (publications and policy documents), two research papers.
External secondments: (1) At the Center for journalology to develop skills in journalology essential for her/his research project and to study the Ottawa data champions programme.
Enrolment in Doctoral degree(s): Enrolment in Doctoral degree(s): University of Rennes
Requirements: Basic knowledge in concepts of meta-research and ELSI; Experiences with systematic reviews and evidence synthesis.

Impact of clinical trial data sharing for pivotal trials in oncology

University of Rennes, Rennes, France

Dr. Clara Locher

More information
Objectives: To assess the impact of clinical trial data sharing in terms of reproducibility of study results for pivotal trials shared by the study sponsor. This project will start with a pilot survey of all pivotal trials in oncology submitted to the European Medicine Agency and will reproduce 1) their results on the main outcome(s) and 2) all safety results. This project will develop a strategy for re-analyses and will help to pilot-test a label “reproducible science” for reproducible trials, a new form of incentive for trials sharing their data and being reproduced.
Expected Results: Impact in terms of reproducibility of clinical data sharing in oncology trials – scoping review: Protocol, Report. First proposal of a label “reproducible science” for pivotal trials.
External secondments: (1) At UZH to develop skills in reproducible research practices; (2) At ECRIN to learn more about policies.
Enrolment in Doctoral degree(s): Enrolment in Doctoral degree(s): University of Rennes
Requirements: Basic knowledge in the quality assessment of clinical trials, biostatistics, and concepts of clinical trial regulation.

Innovative approaches to trial data anonymisation and its role for CTDS

Charitè, Berlin, Germany
Prof. Dr. Fabian Prasser

More information
Objectives: To investigate and develop novel methods and best practices for the anonymisation of trial data. Special emphasis will be placed on how anonymisation can be combined with additional safeguards used in CTDS platforms, such as role-based access control. This will provide insights into how to maximise the utility of shared data while ensuring appropriate privacy protection for the data subjects.
Expected Results: Evaluated tools and processes, case study, possible improvements of existing methods (software), two research papers.
External secondments: (1) At BAYER to learn about privacy protection methods implemented in the industrial context; (2) At DIfE to gain insights into real-world data sharing practices for epidemiological studies.
Enrolment in Doctoral degree(s): Charité, Medical Faculty, PhD programme
Requirements: 1) Master in (Medical) Informatics, Statistics or a related field, (2) Knowledge of basic concepts of privacy protection, (3) Fluent in a relevant programming language.

Automated screening tools for identifying data sharing, data-re-use and common reporting problems in clinical trials

Charitè, Berlin, Germany
Dr. Tracey Weissgerber

More information
Objectives: 1. To extend existing automated manuscript screening tools to detect open data deposited in clinical trial repositories with controlled access (e.g. YODA) and develop a new tool to detect re-use of clinical trial data deposited in a public repository or a repository with controlled access; 2. To develop new automated tools for detecting reporting of CONSORT elements; 3. To use these new tools to assess data sharing, data re-use and reporting quality of clinical trials, and encourage preprint authors to improve clinical trial reporting; 3. To use these tools for Meta-Research studies to assess the prevalence and impact of data sharing and data re-use in clinical trials
Expected Results: 1. Automated screening tools that will detect data sharing, data re-use, and reporting of CONSORT items in clinical trial preprints and papers; 2. Integration of new tools into the ScreenIT pipeline, which will facilitate use of these tools for Meta-Research and interventions to improve data sharing, data re-use and reporting. 3. Meta-Research on the prevalence and impact of data sharing and data re-use in clinical trials (Protocols and SAPs), two research papers.
External secondments: (1) At SciCrunch to learn AI technology for literature reviews.
Enrolment in Doctoral degree(s): Charité, Medical Faculty, PhD programme
Requirements: Experience with the R or Python programming languages (intermediate, beneficial if experience with both); Experience with version control systems (e.g. gitHub, gitLab or comparable repository); Experience with machine learning algorithms (basic skills sufficient, beneficial if practical experience); Attention to detail, and the ability to perform repetitive tasks in a consistently high quality manner; Knowledge of the CONSORT guidelines; Familiarity with reading or performing systematic reviews would be beneficial.

Methodology for FAIRification, Data Enrichment and Data Sharing

University Medical Centre Göttingen, Göttingen, Germany
Prof. Dr. Ulrich Sax

More information
Objectives: To develop and evaluate methodology for improving data reusability and enriching existing data-sets with external knowledge. In this project, we assess best practice approaches for intelligent form building for better interoperability and comparability in different medical domains. Subsequent shared use of the data may benefit from linking the study-specific records with additional data from different contexts from health care or billing. Furthermore, secondary data use will require more context of the data collection and more context towards the interpretation of the data (meta-data). In this project, we examine and measure the success of record linkage and data enrichment approaches in different contexts.
Expected Results: Evaluated toolkit, applied case study, R code/software package, blog/presentation, two research papers.
External secondments: (1) At NFDI4Health (ZB MED) to gain first-hand experience on open data and FAIR data annotation as a prerequisite for the following secondments. (2) At DIfE, to gain practical experience with data capture, data management from primary data, questionnaire data, quality management and analyses. (3) At DIfE, to deepen experience with actual cohorts (e.g. EPIC); to learn about complementary data types in networked clinical research.
Enrolment in Doctoral degree(s): GAU - GAUSS, University of Göttingen.
Requirements: Knowledge in relevant concepts of Medical Informatics; Knowledge in basic concepts of data sharing and data curation.

Added value of meta-analyses of shared individual patient data (IPD) in mental health

University Padova, Padova, Italy
Prof. Dr. Ioana Alina Cristea

More information
Objectives: To establish the impact that IPDMA have had in the field of mental health, specifically with regards to two insofar unanswered questions: (1) Are there harms of psychological treatments, either acute (e.g. side-effects) or at a longer term (e.g. deterioration); (2) Can any prognostic factors be identified to predict response to psychological treatments or otherwise guide treatment selection? To achieve these objectives, the candidate will rely on a suite of Meta-Research methods including i) a large-scale umbrella review of all IPDMA in mental health, ii) a meta-epidemiological approach to examine the role of data sharing by comparing estimates across meta-analyses from trials where investigators made data available and those where they did not and, iii) a qualitative analysis using surveys (about 2000 internet based questionnaires) and about 60 planned interviews with main authors of IPD meta-analyses to identify reasons for lack of participation, beyond what is reported in the paper.
Expected Results: A complete collection of IPDMA in mental health, applied case study quantifying i) the impact of IPDMA in informing clinical practice and research in mental health and ii) obstacles and systematic biases that are hindering the potential of IPDMA in this field, R code, extracted data, blog/presentation. A shiny app will be developed so that findings regarding prognostic factors and harms can be easily consulted, two research papers.
External secondments: (1) At Vrije Universiteit Amsterdam to access the largest collection of mental health IPDMA; (2) At YODA to learn about the process of CTDS from a respective sharing platform.
Enrolment in Doctoral degree(s): UPD.
Requirements: Knowledge of clinical psychology; Knowledge of mental health treatment research; Familiarity with meta-research and meta-epidemiological approaches; Experience with evidence synthesis and meta-analysis; Knowledge of biostatistics and data analysis applied to clinical trials and meta-analysis.

Towards Understanding and accepting data sharing within patients

Hospital Civil de Lyon, Lyon, France
Prof. Dr. Evelyne Decuillier

More information
Objectives: To define the best practice to inform about and engage trial participants in clinical trial data sharing
Expected Results: Survey protocol and survey questionnaire (with their translation in various languages); Qualitative research protocol and interview grids (with their translation in various languages); Analysis report; Recommendations on DS boundaries and on patients’ information. The questionnaire-based survey will be addressed about 2000 trial participants, the interviews will address about 150 persons.
External secondments: (1)At ZB MED to understand the DS technical challenges; (2) At ECRIN to understand how to articulate DS ethical requirements with the setting up of protocols.
Enrolment in Doctoral degree(s): University Lyon Claude Bernard.
Requirements: Basic kowledge of the principles of medical ethics; Experiences in qualitative research; Basic knowledge in clinical trial regulations.

Methodology for cross-design synthesis

University Medical Center Utrecht, Utrecht, The Netherlands
Assistant Professor Dr. Valentijn de Jong

More information
Objectives: We will explore how decision making and health technology assessment (HTA) is affected by (changes in) the amount, type (e.g. non-randomised) and format (e.g. individual participant data (IPD) instead of aggregate data) of available evidence. To this purpose, we will develop and evaluate statistical methodology to combine data from randomised and non-randomised studies. To help ascertaining the validity and credibility of meta-analysis results, we will develop instruments to ascertain the transparency, openness and reproducibility of evidence generated from individual data sources.
Expected Results: Statistical methods, applied case study, R code/software package, blog/presentation, two research papers.
External secondments: (1) At METRICS to develop indicators of transparency, openness and reproducibility for clinical trials; (2) At NICE to apply cross-design synthesis in a case study to study the real-world effectiveness of interventions for COVID-19. Data will be combined from published trials (aggregate data), observational studies (IPD) and patient registries (IPD). The PhD student will also be externally supervised by Dr. Thomas Debray and Dr. Dalia Dawoud; (3) At BAYER to investigate the reproducibility of trial results across multiple (randomised and non-randomised) data sources.
Enrolment in Doctoral degree(s): The PhD student will be enrolled at the Graduate School of Life Sciences of Utrecht University
Requirements: Master in Statistics, Epidemiology, Data Science or related field. Sound knowledge in regression models. Preferred: Good knowledge of meta-analytic concepts, experience with coding (e.g. R, Python, Julia).

Using shared historic data to augment prospective clinical trials

Medical University Vienna, Vienna, Austria
Prof. Dr. Martin Posch

More information
Objectives: (1) To assess and develop statistical methodology for the design and analysis of randomised clinical trials utilising external data sources, e.g. as external controls or for purposes of extrapolation. (2) To review current practice and reporting of the selection process of external data to augment data of randomised clinical trials. To achieve objective (1) the DC will use analytical methods and simulation studies to develop and assess Bayesian and frequentist methods to incorporate external data in clinical trials. Objective 2 will be addressed by a systematic literature review. The potential biases introduced by the knowledge of control group data at the time of trial planning will be assessed in a simulation study as well as in a case study based on a large clinical trial data base.
Expected Results: Trial design and analysis methods utilising external data with a description of limitations and quantification of potential biases. Recommendations on bias mitigation strategies. Review of current practice and report on potential biases induced by the selection process of external data. Statistical software, case studies, two research papers.
External secondments: (1) At BAYER to work on a case study and corresponding analysis methods; (2) At UMCU and SDAaS to work on IPD meta-analysis approaches and their potential use to incorporate external data, as, e.g., historic controls.
Enrolment in Doctoral degree(s): Medical University Vienna
Requirements: Sound biostatistical knowledge; Experiences in systematic reviews and meta-analyses; Knowledge of clinical trial regulations.

Evaluating outcome reporting bias in clinical trials

University of Zurich, Zurich, Switzerland
Prof. Dr. Leonhard HeldFunded by Swiss Confederation

More information
Objectives: Selective outcome reporting occurs when only a subset of the originally recorded outcomes of a clinical trial is reported. Outcome reporting bias (ORB) refers to the situation where the selection is influenced by the study results. For example, benefit outcomes that are non-significant and harm outcomes that are unfavourable to the treatment considered may be less likely to be reported. ORB is a serious threat to the validity of meta-analyses of both benefit and harm outcomes. Several statistical methods are available for ORB adjustments of combined treatment effect estimates based on reported summary statistics. The availability of individual patient-level data will provide complete information on primary, secondary and harm thus enabling to assess the accuracy of meta-analytic adjustment methods in the presence of ORB through simulation. Furthermore, the extent of ORB in the medical literature will be assessed through a systematic comparison of the outcomes reported in available publications against the set of originally recorded outcomes.
Expected Results: Neutral comparison simulation study of methods to adjust for ORB: Study protocol, SAP, report. Possible improvements of existing methods (Software, Publications). Systematic comparison of recorded and published outcomes.
External secondments: (1) At University of Rennes to develop of a protocol for simulation study based on systematic literature review on ORB in clinical studies. (2) At METRICS for a systematic comparison of the outcomes reported in available publications against the set of originally recorded outcomes. Quantitative assessment of the amount of ORB in clinical studies.
Enrolment in Doctoral degree(s): University of Zurich, Life Science Zurich Graduate School, PhD programmeme in Epidemiology and Biostatistics
Requirements: Obtained a Masters degree in Biostatistics or Data Science.