I work in the UK National Health Service [ NHS ] and am considering a pilot project to curate our medical linac commissioning and quality control data . Existing commercial and linac vendor projects exist, as does a similar open system qatrack+ : see http://qatrackplus.com/news.html
The scale of the data is not huge/ i.e. not 'big data'; until one envisages national or even international pools.
The rules for processing/parsing/ analysing the data to essentially look for common performance records and deviations are relatively straightforward; but some of the data would need to be carefully stripped of identifiers for data protection purposes.
I was considering CERN ROOT but have put that to one side due to lack of c++ skill and not a huge amount of 'positive hits' in my searches for db approaches in ROOT. I have had a little experience in R, Python and the SciDB 16.9 VM seems of both relevance and simplicity, in terms of interface and (inherent) processing , and the scaling seems to be 'there'. (As in , one may be able to creaate a model with a relatively few contributors or just local data, and later, securely roll it out (somehow)).
anyhow, if it rings any bells do let me know,