Structured missingness workshop

We are inviting researchers to participate in a virtual workshop on the topic of structured missingness, funded by the Turing-Roche strategic partnership. The three-part workshop will take place between 15:00 – 17:00 GMT on Wednesday 10 November, Wednesday 17 November and Wednesday 1 December 2021.

The workshop aims to convene a community of researchers interested in developing new tools and methods to learn from heterogenous data with structured patterns of missing readings. Our intention is to bring a diverse community of researchers together – including established and early career researchers – to seek synergies and develop collaborative project ideas.

At the end of the three-part workshop, we will be issuing a call for research proposals to support substantive projects of work in the area of structured missingness. Successful proposals will be fully funded by the Turing-Roche strategic partnership.  

All active attendees of the workshop will be eligible to apply for this funding. The call will also be open to researchers who were not able to attend the workshop however priority will be given to project proposals emerging from this event, particularly collaborative proposals that combine different areas of data science and/or different academic institutions.

About structured missingness

Missing data are a common problem that arise in many fields and can significantly complicate analysis. While there are established ways to handle data that is missing at random, it is often the case that missing readings are structured in some way. Dealing with such structured missingness is substantially more challenging yet occurs commonly in healthcare (and other) contexts. For example, one cohort of patients may have been given a certain test, another may not, leading to blocks of missing readings. Simply excluding such missing (or potentially inaccurate) data can significantly bias subsequent analysis. Consequently, there is a need to develop rigorous ways to understand structured missingness and develop tools to handle it appropriately. 

As an example, the Roche Clinico-Genomic Database, combines real-world patient level clinical data with in-depth genomic profiling for over 30,000 patients. While this data set of enormously rich, it also possesses significant structured missingness: the sets of genes measured varies between patients depending upon which tests they received; similarly, the clinical measurements vary between patients depending upon their particular indication. 

The objective of this series of three meetings is to scope out approaches to structural missing data, particularly using the Clinico-Genomic Database as motivation, and formulate substantive research questions that merit further exploration. At the end of the event, we will be issuing a call for research proposals to support projects that arise from these discussions. All active attendees of the workshop will be eligible to apply for this funding.