Research Data Management in STEMM

Talk given as part of Data Management elective at the 2017 University of Melbourne Research Bazaar.

Transcript

1. Research Data Management in STEMM Ewan Nurse (@EwanNurse) Depts. of Biomedical Engineering and Medicine University of Melbourne
2. • What does good STEMM data management look like? • Case study – EEG data storage from a clinical trial
4. What does good STEMM data mgmt look like? 1. As small as (reasonably) possible 2. Secure 3. Meets ethics requirements 4. Easy to share
5. 1. As small as reasonably possible
6. 2. Secure • University servers • Reliable, very safe for use • Cloudstor & figshare • Easy storage and file sharing, capped at 100 GB for standard account
7. 2. Secure • Hard-disk drives • Good for portability and offline work • Physically vulnerable • File hosting servers (Dropbox, Google Drive) • Not appropriate for sensitive data • Easy to integrate with multiple computers
8. 3. Meet ethics requirements • All ethics applications need to have a data management plan • Check with local committee • Often worth making ‘de- identified’ version of data
9. 4. Easy to share • Great way to get citations! (make sure covered by ethics) • Think about file types • Github limits to 25 MB/file • Cloudstor lets you send links to stored files (up to 100 GB)
10. Case Study – EEG data storage • Data recorded from first- in-human device trial • Is clinically and commercially sensitive • Majority of data is voltage recordings in .mat files (~50TB)
11. Case Study – EEG data storage • Originally stored on hard drives and copy made to portable hard drives • Now access through online portal via MATLAB plugin • Results saved on university server
12. Summary: • Good data management is essential for high-quality, ethical STEMM research • All data should be: 1. As small as possible 2. Secure & backed up 3. Meeting ethics requirements 4. Easy to share
13. Thanks for listening!